Environmental

Report
2024
 Introducti AI for Our Our Appendi
on sustainability products operations x

12

Table of contents

Introduction 2 Appendix 60

Executive letter 3 About Google 61

Our sustainability strategy 5 Sustainability governance 61

2023 highlights 6 Risk management 61

Targets and progress 7 Stakeholder engagement 62

summary
from our 2023 fiscal year (January 1 to
December 31, 2023), and mentions some Searching for sustainability 8
notable achievements from 2024 to date.

AI for sustainability 9

Our products 14

Mitigation 16

Adaptation and resilience 23

What’s Our operations 27

inside
Net-zero carbon 29
Additional resources Water stewardship 42
• Explore the 2024 Environmental
Report with AI Circular economy 49
• Sustainability.google
Nature and biodiversity 56
About this report • Sustainability reports
• Sustainability blog
Google’s 2024 Environmental Report provides an overview
• Google’s mission, values, and
of our environmental sustainability strategy, our targets,
commitments
and our annual progress toward them. Throughout this
• Alphabet SASB and TCFD Index
report, we use the term “sustainability” to refer to
environmental sustainability. This report features data, • About Google
performance highlights, and progress against our targets
 Introducti AI for Our Our Appendi
on sustainability products operations x

and partnership

Multi-sector products 67

Ecosystems for 68
collaboration
Environmental data 70

Certifications 80

Recognitions 80

Endnotes 81
Introduct AI for Our Our Appen

Introduction
Executive letter

Our sustainability strategy

2023 highlights

Targets and progress

summary Searching for

sustainability
 Introduct AI for Our Our Appen

A letter from our Chief Sustainability Officer and

our Senior Vice President of Learning and
Sustainability

Since our earliest days, we’ve

Our annual Environmental Report We strive to build the world’s most and a vehicle’s engine to suggest the How we're driving sustainability across
been on an ambitious journey offers a deep dive into our efforts to energy- efficient computing most efficient route. It's estimated to our operations
to help build a more harness infrastructure, supported by responsible have helped enable more than 2.9
sustainable future. technology—particularly AI—to drive water use practices and a commitment million metric tons of GHG emissions In 2017, Google became the first major
positive environmental change and to minimizing waste. A Google- owned reductions since the feature launched company to match 100% of our annual
An important part of that is sharing what operate our business sustainably. This and -operated data center is, on in late 2021 to the end of 2023— that’s electricity consumption on a global
we’ve learned along the way and being year, we’re also offering a new average, approximately 1.8 times as equivalent to taking approximately basis with renewable energy, which
transparent about our progress and our experimental AI chatbot, powered by energy efficient as a typical enterprise 650,000 fuel-based cars off the road for we’ve achieved every year since. 9
challenges. NotebookLM, to help summarize key data center. 3 In 2023, the average annual a year. 6
This is especially true given the findings, clarify complex topics, and power usage effectiveness for our data Building on our first two decades of
• Improving prediction: We built a
urgency of the moment—a time when explore the details about our centers was 1.10 compared with the progress, in 2020 we launched our third
breakthrough global hydrological AI
technological advancement is environmental work. industry average of 1.58, 4 meaning that decade of climate action—our most
model and combined it with publicly
converging with the need for energy our data centers used about 5.8 times ambitious yet.
available data sources to predict
transition. Our approach to enabling AI for less overhead energy for every unit of IT floods up to seven days in advance in
sustainability equipment energy. over 80 countries. This covers
Last year we introduced a water risk territories where more than 460
We know that scaling AI and using it to
million

accelerate climate action is just as crucial framework to further identify people live, 7 helping these communities
climate- prepare for and respond to riverine floods.
as addressing the environmental 1,000conscious cooling solutions that consider carbon-free energy (CFE) availability, watershed
impact associated with it. times.health and future water needs. We see our growing infrastructure as an opportunity to drive
All the innovations and investments needed to power a low-carbon economy.
To help minimize our environmental these
footprint, we’ve built world-leading practic
AI holds immense promise to drive climate action. In fact, AI has the potential to help mitigate 5–
efficient es 10% of global greenhouse gas (GHG) emissions by 2030. 5 We’re advancing climate action
infrastructure for the AI era, including used through AI in three key areas:
Trillium, our sixth-generation Tensor at
• Organizing information: Fuel-efficient routing uses AI to analyze traffic, terrain,
Processing Unit (TPU), which is over 67% Google
more energy-efficient than TPU v5e. 1 today.
We’ve also identified tested practices
that our research shows can, when used
together, reduce the energy required
to train an AI model by up to 100 times
and reduce associated emissions by up to
 Introduct AI for Our Our Appen

u reduction in emissions at
• Better optimization: Green Light p intersections. 8
is an AI- based tool that helps
Through our products, we aim to help
city traffic engineers optimize t individuals, cities, and other partners
the timing of traffic lights to o collectively reduce 1 gigaton of carbon
reduce stop-and-go traffic and
equivalent emissions annually by 2030,
fuel consumption. This 1 and we’ll continue to develop
technology has the potential 0 technologies that help communities
for up % adapt to the effects of climate change.
to 30% reduction in stops and

3
 Introduct AI for Our Our Appen

We have a bold goal to reach net-

zero emissions across all of our
• We implemented a Google
Renewable Energy Addendum that
Our ongoing work to build a of clean energy projects and the
resulting GHG reductions from them.
Kate E. Brandt
sustainable future
operations and value chain by 2030,
asks our largest hardware Kate E. Brandt
supported by a goal to run on 24/7 To continue to drive progress
manufacturing suppliers, based on In spite of the progress we're making, Chief Sustainability Officer, Google
CFE on every grid where we operate. toward a low-carbon economy, we
spend, to commit to achieving a we face significant challenges that
In addition, we’re working to most recently
100% renewable energy match by we’re actively working through. In 2023,
advance water stewardship, build a
2029. 12 our total GHG emissions increased 13% introduced a clean transition rate that Benedict Gomes
circular economy, and restore and brings customers and utilities together to
• Our water stewardship projects year-over-year, primarily driven by Benedict Gomes
enhance nature and biodiversity. This drive new clean energy projects in the
replenished an estimated 1 billion increased data center energy
year’s report shows how we continue to SVP, Learning & Sustainability, Google
consumption and supply chain United States, and we unveiled an
make progress across all of these areas: gallons of water, 13 which represents
emissions. investment to enable 1 gigawatt of new
18% of our 2023 freshwater
• Ten of our grid regions 10 achieved at solar capacity in Taiwan.
consumption and tripled our
least 90% CFE, and even with our total While we advanced clean energy on
replenishment progress of 6% in 2022.
many of the grids where we operate, A sustainable future requires systems-
electricity load increasing across our
• For new Google products launched and there are still some hard-to- level change, strong government
data centers, we maintained a global
manufactured in 2023, our packaging decarbonize regions like policies, and new technologies. We’re
average of 64% CFE. We also
was at least 99% plastic-free. 14 Plus, Asia Pacific where CFE isn’t readily committed to collaboration and playing
celebrated a first-of-a-kind enhanced
packaging for our Pixel 8 and Pixel 8 available. In addition, we often see our part, every step of the way.
geothermal project now delivering CFE
Pro uses 100%plastic- free materials. 15 longer lead times between initial
to the grid.
investments
We signed contracts to purchase approximately 4 gigawatts of clean energy generation capacity and such
11 in locations construction
as , , and —more than in
any prior year.

Our campus in Mountain View, California.

 Introduct AI for Our Our Appen

Our sustainability strategy

We believe Google has a unique opportunity that extends beyond managing the environmental impact of our own operations and value chain. Our products, which are used by
billions of people every day, can enable decisions that drive positive action for the environment. The figure below illustrates these two pillars of our sustainability strategy,
supported by our dedication to accessible information and technological innovation. To learn more about our sustainability governance, stakeholder engagement, environmental
data, and more, see the Appendix.

Our products Our operations

wering people with information to mitigate
We’re
andworking
adapt to
toclimate
drive sustainability
change—focusing
acrosson
oursectors
operations
like energy
in four key
and ways:
transportation
accelerating the transition to a net-zero carbon future, advancing water stewardship, building
e’s unique capabilities can have a meaningful impact.

Net-zero carbon Water stewardship

Mitigation Adaptation and resilience
Energy Extreme event forecasting and early warning systems
Transportation

Circular economy Nature and biodiversity

Learn more in the section. Learn more in the section.

Information and innovation

We’re helping to lead the transition to a more sustainable future by making information accessible and by driving innovation forward.
 Introduct AI for Our Our Appen

2023 highlights
For additional highlights, see the approach page of each report section. And for a more complete overview of our performance over time, see the Appendix for our Environmental data tables.

Our products Net-zero carbon

Reached 1 billion users Displayed flood forecasts for over Achieved 80 countriesat least
covering
90% 460 million people 100%
Maintained Contracted 4 GW
For the past two years, Google has provided information
Flood Hub displayed to forecasts
over 1 for over 80 countries on five continents—
carbon-free energycovering
in more than 460 million
billion users to help them make more sustainable choices
people globally. 17 annually through our products. 16 renewable energy of clean energy
Learn more on 15 Learn more on 25 10 grid regions match for 7 years We signed contracts to
Ten of our grid regions 19 purchase approximately 4
We achieved seven
achieved at least 90% CFE and GW of clean energy
consecutive years of 100%
we maintained a global generation capacity 21— more
renewable energy matching
average of 64% CFE across our than in any prior year. Learn
on a global and annual
data centers and offices—even more on page 35
basis. 20
as our total electricity
Learn more on page 33
consumption increased. Learn
more on page 35

Enabled 2.9 million metric tons of GHG emissions reductions from fuel- efficient routing
Fuel-efficient routing is estimated to have helped enable more than 2.9
million metric tons of GHG emissions reductions since the feature launched in late 2021—equivalent to taking approximately 650,000 fuel-based cars off the road for a year. 18
Learn more on 20 Water stewardship Circular economy Nature and biodiversity
Doubled our water Achieved 100% plastic- Created or restored
replenishment free packaging for Pixel 67 acres of habitat
portfolio 8 and 8 Pro As of the end of 2023, we created
We nearly doubled our water Packaging for Pixel 8 and or restored approximately 67 acres
replenishment portfolio, Pixel 8 Pro uses 100% of habitat and planted roughly
increasing from 38 water plastic-free materials. 22 4,500 native trees on Google’s
stewardship projects in 2022 to campuses and the surrounding
Learn more on page 55
74 projects in 2023. Learn more urban landscape, primarily in the
on page 46 San Francisco Bay Area. Learn
more on page 58
 Introduct AI for Our Our Appen

Targets and progress summary

This table summarizes our targets and progress as of December 31, 2023. For additional details and year-over-year trends, see the target boxes in relevant report sections and the Targets data table
in the Appendix.

Targets 2023 progress

Run on 24/7 carbon-free energy on We maintained a global average of approximately 64% carbon-free energy across our
every grid where we operate by data centers and offices from 2022 to 2023—despite growth in electricity demand
Net-zero carbon 2030 over this period. Learn more on page 35
We aim to achieve net-zero emissions across
all of our operations and value chain by Our total GHG emissions were 14.3 million tCO 2e, representing a 13% year-over-
2030 year increase and a 48% increase compared to our 2019 target base year—
Reduce 50% of our combined Scope 1, 2
(market-based), and 3 absolute emissions primarily due to increases in data center energy consumption and supply chain
emissions.
(compared to our 2019 base year) by 2030 23
Learn more on page 31

Water stewardship
Replenish 120% of the freshwater volume Our water stewardship projects replenished approximately 18% of our freshwater
We aim to replenish more water than we consume and help
we consume, on average, across our offices consumption from our data centers and offices—tripling our replenishment progress
improve water quality and ecosystem health in the communities
and data centers by 2030 of 6% in 2022.
where we operate
Learn more on page 46

Achieve Zero Waste to Landfill 24 for Following our updated waste accounting methodology, 29% (8 out of 28) of our
our global data center operations Google-owned and -operated data center campuses met our Zero Waste to Landfill
Circular economy goal. 25
Learn more on page 53
We aim to maximize the reuse of finite
resources across our operations, products, and
supply chains Divert all food waste from
landfill by 2025 We diverted 82% of food waste from landfill—a slight decrease from 85% in
2022—due, in part, to limited composting infrastructure in certain regions.
Learn more on page 52

Use recycled or renewable material in at least 34% of the plastic Google used in products manufactured in 2023 was recycled
50% of plastic used across our consumer content. 26 This decrease from 41% in 2022 27 was due to changes in our product
hardware product portfolio by 2025 mix—some product types use less plastic than others, which can reduce
opportunities to use recycled content. Learn more on page 54

Make product packaging 100% For new Google products launched and manufactured in 2023, our packaging
plastic-free by 2025
Introduct AI for Our Our Appen

was at least 99% plastic-free, 28 an increase from at least 96% plastic-free

packaging in 2022, primarily due to packaging innovations. 29
Learn more on page 55
 Introduct AI for Our Our Appen

Searching for sustainability

Google Search provides a unique lens into the world’s perspectives on sustainability and the issues people care about most. Google Trends 30 show that the world is
searching for sustainability information and insights like never before. For example in 2023, search interest in “jobs related to sustainability” reached an all-time high. 31 In fact,
we estimate that 36% of Google Search users in the United States have searched for a sustainability-related topic over the course of a year. 32 Below are trends that demonstrate
the impact climate change is having around the world, and what people are searching for to better understand it.

Extreme weather events Clean energy Recycling

Search interest in “coastal flooding” in the United States
“Energy development” and “solar power” were among the top In the United States, “electronic waste recycling” was searched
reached an
five trending topics searched with “climate change more than ever in 2023, 38 and top materials searched with
all-time high in January 2024, 33 and the top natural disasters
mitigation” in 2023. 35 “Solar” and “wind power” are the two “recycling” worldwide in 2023 included plastic, paper, metal, glass,
searched for worldwide in 2023 include earthquakes, floods,
most searched renewable energy types in the world, 36 and and cardboard. 39
tsunamis, storms, and tropical cyclones. 34
global searches for “rooftop solar power” specifically reached
an all-time high in 2024. 37
Search interest in coastal flooding since 2004,
Search interest in electronic waste recycling since
in the United States
100 2004, in the United States
Coastal flooding Search interest in solar power versus wind power in 100

80
2023, worldwide
80
Electronic waste recycling
Wind Solar
60 60
power
power
40
40
20

20
0
2004 2010 2016 2022
2004 0
2 2
2

Top searched natural disasters in 2023, worldwide Top materials

searched with recycling
Earthquake Flood Stor Tropical cyclone* Tsunami 10%
m in 2023, worldwide
100
12%
Search interest in rooftop solar power
Cardboard
since 2004, worldwide 80
Glass 42%
Rooftop solar 60 Metal 13%
Paper
40
Plastic

20
*includes hurricane searches

2004 2010 2016

23%
0
2022
Introduct AI for Our Our Appen

AI for sustainability
Our approach

Boldly accelerating climate action with AI

Responsibly managing the resource consumption of

AI Working together
 Introduct AI for Our Our Appen

Our approach to enabling AI for sustainability

FIGURE 1
AI can have a transformative With a bold and responsible approach,
Harnessing AI-powered climate solutions with environmental responsibility
effect on climate progress. we’re taking steps to make this
We’re technology helpful for everyone.
harnessing our years of
While scaling these AI applications and
experience as an AI-first Boldly accelerating climate action with AI Responsibly managing the resource consumption of
finding new ways to use AI to accelerate
company to enable others— climate action is crucial, we’re also
people, businesses, working to
communities, and By providing solutions for information, prediction, and optimization
responsibly manage the environmental We’re committed
acrosstosectors,
developing
AI can
AI play
responsibly
a by working to address its environmental footprint through m
transformative role in climate progress. efficient infrastructure, and emissions reductions.
governments— to use AI for impact associated with it.
both mitigation and
Our strategy on enabling AI for
adaptation.
sustainability is underpinned by the
following approach (see Figure 1):
As highlighted in our AI Opportunity
Agenda, we’re currently at an inflection
point where choices made today will
ensure that AI is used to benefit as
many people as possible.

Information Prediction Model optimization Efficient Infrastructure

Optimization Emissions reductions

 Introduct AI for Our Our Appen

The world has in its hands the potential to apply artificial intelligence (AI) and machine learning (ML) to drive forward the net-zero transition and give us a chance to stay within 1.5℃. AI and ML can contribute massively to the pace of
pollution.

Nicholas Stern
Chair, Grantham Research Institute on Climate Change and the Environment, London School of Economics
Working together to enable AI for climate progress

Working together across industry, governments, experts, international organizations,

To read the full article from Nicholas Stern, see and individuals is essential to unlock the potential of AI for tackling climate change.
.
 Introduct AI for Our Our Appen

Boldly accelerating climate action with AI

AI has a critical enabling • Near-real time data and monitoring:

role to play in accelerating By combining data from
Optimization FIGURE 2
Cool Roofs tool
MethaneSAT,
mitigation,
Environmental Defense Fund’s new One of the challenges of addressing
supporting adaptation, and
methane-tracking satellite, with climate change is the fragmented way that
building foundational capabilities Google’s
for the
AI and infrastructure mapping information is gathered and used. AI offers a
transition to a low-carbon capabilities, 21).
transformative solution by not only gathering and interpreting vast amounts of data,
future. we’ll create a more comprehensive
view of methane emissions from oil but also optimizing complex systems with unprecedented efficiency.

According to an estimate from Boston and gas— enabling a better

• Home energy use: For over a decade, our Nest Learning Thermostats have used
Consulting Group, AI has the potential to understanding of how to mitigate
machine learning to help people save energy and money at home. In 2023, we estimate
help mitigate 5–10% of global GHG them (see page 22).
that Nest thermostats helped customers save more than 20 billion kWh of energy 43 (see
emissions by 2030 40—the equivalent of
page 19).
roughly the total annual emissions of
Prediction • Grid decarbonization: Tapestry—a part of X, Alphabet’s moonshot factory—is creating a
the European Union. 41 And the positive
impact of AI could grow as it contributes Breakthroughs in forecasting by using single virtualized view of the electricity system through AI-powered tools that can predict
to breakthroughs that open new AI enable improvements in and simulate what might happen on the grid from milliseconds to decades into the future
pathways for climate action. preparedness efforts and facilitate (see page 19).
interventions in climate mitigation.
In this context, the three distinct These solutions demonstrate that the applications of AI to meaningfully bolster mitigation,
capabilities that AI enables are • Extreme heat: When people search for adaptation, and climate resilience aren’t just speculative, but rather are having real impact
information, prediction, and information on extreme heat, they’ll today.
optimization. see details on when a heat wave is
predicted to start and end, tips on

Information staying cool, and related health

concerns to be aware of—all
Information generated by AI through prominently displayed in Search
the rapid processing and analyzing of results.
vast amounts of data can empower Since launching, we’ve provided The deployment of AI in crafting and executing energy strategies presents an unparalleled opportun
individuals and organizations in their information about extreme heat on
Search in more than 100 countries (see Damilola Ogunbiyi
own sustainability efforts.
CEO and Special Representative of the UN Secretary-General for Sustainable Energy for All, and Co-Chair of UN-E
page 25). To read the full article from Damilola Ogunbiyi, see
• Cool roofs: We’re exploring how AI • Contrails: We’re working with the
.
algorithms and aerial imagery can airline industry to develop contrail
help implement reflective roofs— forecast maps using AI-based
called “cool roofs”—which save predictions, allowing pilots and
energy and reduce indoor and airlines to optimize routes that avoid
outdoor temperatures. As of the end creating climate-warming contrails. In
of 2023, this pilot was available in 15 one trial consisting of 70 test flights,
cities (see Figure 2 and page 26). use of the predictive technology
reduced contrails by 54% 42 (see page
 Introduct AI for Our Our Appen

Our Cool Roofs tool

uses AI and aerial
imagery to generate
high resolution roof
solar reflectivity
measurements
 Introduct AI for Our Our Appen

Responsibly managing the resource consumption of AI

While AI offers new solutions heavily in improving the efficiency of our

AI models and infrastructure. While these brought necessary increased attention to its energy consumption and resource demands.
for climate action, it also has
efforts have helped mitigate some of AI’s
its own environmental In 2023, our total data center electricity consumption grew 17%, despite maintaining a 100% global renewable energy match. As Google’s
environmental footprint, the rapid
impact. We have an advancement of AI has infrastructure continues to power the digital transition, providing numerous economic benefits across the globe, we expect this trend to
understanding of its current continue in the future. But we see our growing infrastructure as
demands, but its future an opportunity to drive the innovations and investments needed to power a low-carbon economy.
trajectory
Overall, our total GHG emissions increased by 13%—highlighting the challenge of reducing emissions while compute intensity increases and
remains uncertain. Here’s what we grow our technical infrastructure investment to support this AI transition. For more details, see the Net-zero carbon section.
we know and what we’re still
learning. Predicting the future environmental impact of AI is complex and evolving, and our historical trends likely don’t fully capture AI’s future
trajectory. As we deeply integrate AI across our product portfolio, the distinction between AI and other workloads will not be meaningful.
Contextualizing Google’s impact

Based on the most recent

estimates as of 2022, global data
center electricity consumption is
240–340 TWh, or around 1–1.3% of
global electricity demand. 44
Using these estimates as a proxy for
2023, Google’s data center
consumption of more than 24 TWh in According to the IEA, estimated global data center electricity consumption represents around
2023 translates to approximately 7– 1–1.3% of global electricity demand. 47
10% of global data center electricity
consumption and approximately
0.1% of global electricity demand. 45

While data centers consume electricity

and contribute to emissions, cloud and
hyperscale data centers collectively
represent only an estimated 0.1–0.2% of
global GHG emissions, based on the
most recent global estimates
as of 2022. 46 Google’s emissions
contribute a smaller fraction—in 2023,
our total GHG emissions were
approximately 14.3 million tCO2e.

AI’s resource demands

AI has been deeply integrated into our

products for years, and we’ve invested
 Introduct AI for Our Our Appen

S n (and hence, the environmental impact) of

Through increased collaboration, innovation, and implementation, together with the development of
o c AI. Golestan (Sally) Radwan
, e Chief Digital Officer, United Nations Environment Programme
Uncertainty ahead
To read the full blog from Golestan (Sally) Radwan, see .
w t
e h AI is at an inflection point and many
’ e factors will influence its ultimate impact—
r y including the extent of AI adoption, our
e ability to mitigate
i its footprint, and the pace of continued
f n innovation and efficiency. Additionally,
o c system-level changes are needed to
c l address challenges such as grid
u u
decarbonization, evolving regulations,
s d
hard-to-decarbonize industries, and the
i e
availability of carbon-free energy. While
n
we remain optimistic about AI’s potential
g t
to drive positive change, we’re also
h
clear-eyed about its potential
o e
environmental
n
o
impact and the collaborative effort Google’s energy and efficiency highlights
required to navigate this evolving
d v
landscape.
a e
t r We’re committed to responsibly
a a managing the environmental impact of AI 100% 64%
l by deploying three major strategies: global renewable energy match 48 carbon-free
c l energy
model optimization, efficient
e infrastructure, and emissions reductions.
n r
t e
e s
r o
- u 1.10 1.8x
w r fleet-wide average annual PUEas
49 energy efficient as a typical enterprise data center 50

i c
d e
e
c
m o
e n
t s
r u
i m
c p
s t
i
s o
i n
 Introduct AI for Our Our Appen

Working together

Model optimization Efficient infrastructure in the world, and we continue working

Emissions reductions Our AI for climate policy
to optimize their use of electricity,
Google has long been at the We strive to build the world’s most water, and materials. In 2023, the We aim to reach net-zero emissions agenda
forefront of AI and machine learning, energy- efficient computing average annual power usage across all of our operations and value
effectiveness 57 (PUE) for our global fleet Harnessing the potential of AI for
evolving years of deep learning infrastructure, supported by responsible chain by 2030. In 2023, we achieved
of data centers was 1.10 (see Figure climate action requires collective action.
research into techniques that make water use practices and a commitment 64% carbon-free energy on average
3) , compared with the industry average Policymakers, in particular, can help by:
training faster and more efficient— to minimizing waste. across all of our data centers, and we
enabling models that are higher of 1.58 58—meaning that Google data purchased over 25 TWh of renewable
centers used about 5.8 times less Encouraging data sharing, ensuring
quality, faster, and less compute- Chip hardware efficiency: We’re electricity—including from PPAs,
overhead energy for every unit of IT affordable technology access, and building
intensive to serve. continually improving the power on-site renewable energy generation, and
equipment energy. For more details, see awareness: Accessible and standardized
efficiency of AI hardware. For example, grid renewable energy.
the Net-zero carbon section. data (exemplified by tools like Data
Development and training: We’ve our TPU v4
Commons and MethaneSAT), robust AI
identified tested practices that our was 2.7 times more energy efficient We deployed innovations like advanced
Responsible water use: The expansion compute availability, and widespread AI
research shows can, when used than our TPU v3, 54 and we’ll soon geothermal, carbon-intelligent computing,
together, reduce the energy required to of AI products and services is leading expertise (fostered through programs like
offer and demand response capabilities, and
train an AI model by up to 100 times to an increase in data center the Government AI Campus, with funding
Nvidia’s next-generation Blackwell partnered with others to leverage our
and reduce associated emissions by workloads and the associated water from Google.org)
GPU to Cloud customers, which Nvidia demand, like our demand aggregation
up to 1,000 times, 51 which are all used at footprint required to cool them and procurement model for advanced are crucial for developing effective
estimates will train large models using
Google today. We’ve sped up AI model efficiently. In 2023, our data centers clean electricity technologies with climate- related AI solutions.
75% less power
training through techniques like than older GPUs to complete the same consumed 6.1 billion gallons of water— Microsoft and Nucor and our
quantization, boosting large-language 17% Defining market-specific priorities,
task. 55 Additionally, our new Google participation in Frontier.
model training efficiency by 39% on more water than the previous year, delivering on public sector use cases,
Axion Processors are up to 60% more
Cloud TPU v5e. 52 And our Go Green mirroring similar growth in electricity use. We work with our supply chain, and encouraging private sector action:
energy efficient than comparable
Software guide helps developers reduce To put this into perspective, in 2023 our partnering to improve environmental Policymakers can reduce regulatory
current-generation x86-based instances.
their digital footprints. 56
These advancements, including data centers used data collection and reporting and to hurdles, invest in enabling
AI-powered optimizations like the same amount of water needed to develop decarbonization roadmaps— infrastructure, encourage innovation,
Deployment and usage: Google’s AlphaZero, show how we’re constantly irrigate roughly 41 golf courses annually, focusing on reducing emissions in high- and align incentives. In turn, the
Gemini 1.5 Pro delivers dramatic improving hardware efficiency. on average, in the southwestern United impact areas. private sector can focus on high-impact
improvements and achieves States. 59 We’re committed to areas, pilot solutions, and partner to
comparable quality to Gemini 1.0 Ultra Data center energy efficiency: Our data responsible water use—using our We’re also working to reduce the scale them— exemplified by Google
while using less compute. 53 centers remain some of the most water risk framework to identify embodied carbon impact of growing Research’s partnership
climate- conscious cooling solutions with Breakthrough Energy and American
efficient machine learning
that consider carbon-free energy
FIGURE 3 availability, watershed health, and
future water needs. For more
1.12 details, see the Water stewardship
Energy efficiency at Google and 29% (8 out of 28) of our data centers
1.10 section.
data centers met our Zero Waste to Landfill goal. 60
1.08
We also maintain servers for as long as
1.20 Waste: In 2023, we diverted 78% of
Legend Trailing 12-month PUE possible by refurbishing, reusing, or
1.18 operational waste from disposal
reselling components, and we work to
1.16 across our global fleet of Google-
Trailing 12-month

ensure device longevity. For more details,

1.14 owned and -operated data centers,
see the Circular economy section.
 Introduct AI for Our Our Appen

deman machine placement, promoting the reuse and associated with our products, including AI, Policymakers can promote
d at Airlines to develop technology that
upcycling of technical infrastructure and we remain dedicated to their transparency into the impact of AI,
our predicts contrail formation zones
data hardware, and collaborating with organizations prioritization as we continue to innovate. For encourage voluntary commitments from AI
and informs flight paths accordingly.
centers like Building Transparency and the iMasons providers to mitigate this impact, and work
more detail, see the Net-zero carbon
by Climate Accord. to better enable AI providers to source
optimiz section. Promoting environmentally
carbon-free energy for their operations
ing responsible AI deployment:
These efforts help to reduce the emissions and supply chain.
0 2012 2014 2016 2018 2020 2022
2010
Introduct AI for Our Our Appen

Our products
Our approach

Mitigation

Adaptation and resilience

 Introduct AI for Our Our Appen

Our approach to product sustainability

We’re developing tools and technologies that help people mitigate and adapt to climate change through
our products

A sustainable future will be As an information company, we bring For the past two years, Google has We’re tackling climate change by In addition to our own products, our
built upon countless accurate, authoritative information to provided information to over 1 billion focusing on some of the most critical platforms like Google Cloud enable
decisions made by people to help them make key decisions users to help them make more challenges: partners to build solutions that help
governments, organizations, in their lives, including information about sustainable choices annually through our reducing emissions in sectors like others mitigate and adapt to climate
sustainability. We don’t tell people what products. 61 We achieved this by offering energy and transportation, and better change. For more details on Google
businesses, and individuals,
choice to make, but we know that when sustainability features like predicting extreme weather to lessen its Cloud, see the Multi-sector products
which they are given high quality information, fuel-efficient routing in Google Maps effects. By applying section in the Appendix.
will need to be grounded in they often make more sustainable and more transportation options in AI and other cutting edge technologies
helpful and authoritative choices. Search, such as train routes. to these problems, we can help
information. individuals and organizations achieve
their goals and maximize impact.

We’ve developed an AI-based technology to map wildfires in near real-time in Google Search and Maps.
Introduct AI for Our Our Appen

Mitigation
We aim to help people make more sustainable choices to mitigate climate change

Our approach
1 gigaton aspiration

Energy
Home heating and cooling

Solar

Electricity grid decarbonization

Clean energy partners

Transportation

Driving

Transit and urban mobility

Flying

Additional sectors
 Introduct AI for Our Our Appen

Our approach to climate mitigation

In 2023, global searches their health. We also strive to enable areas like geospatial mapping (Google
for “climate change businesses to track emissions, Maps, Google Earth), smart home
mitigation” reached an optimize operations, and drive climate technology (Nest thermostats), and grid 2023 highlights Global challenges
all-time high. 62 action throughout their value chains. decarbonization (24/7 CFE), Google is
uniquely positioned to have the most Green Light: Recommendations from —an AI tool for optimizing
Say-do gap: There’s
traffic
a significant
signals—were
opportunity
implemented
to help
inc

With solutions across Google products, Across sectors that have been impact in these sectors through our
Transportation options in Search: We now featureInformation
more , showcasing
quality: Deliberately
long-distance
false
train
or routes
misleading
in 38inc
we aim to help individuals make identified by experts as having the products and research.
informed choices about their largest mitigation Solar API: Our helps accelerate solar development,
Needand
for now covers
localized more than
solutions: The472
complex,
million localized
buildings a
environmental impact. Providing people potential, 63 we’ve chosen to focus on We’re also developing tools that support developing product sustainability features with wides
with information on sustainable energy and transportation—which a broad range of mitigation solutions
alternatives often provides co-benefits— collectively account for roughly half of across multiple sectors. For more details,
like saving them time or money, or global emissions 64 (see Figure 4 and see the Multi-sector solutions section in
improving Figure 5). With expertise in the Appendix.

FIGURE 4
Our sector-based approach to mitigation

Sectors of primary focus

Energy Transportation
Empowering individuals, businesses, and grid operators to navigate the clean energy transition—optimizing renewable energy sources and decarbonizing the grid
Supporting more sustainable travel choices across various modes of transportation, including driving, transit, and flying
 Introduct AI for Our Our Appen

FIGURE 5
Estimated global emissions (by sector)

Sectors of focus for Google

A view of Earth shows the vast network of electric grids worldwide.

SPOTLIGHT

Our 1 gigaton aspiration

We’ve set an aspiration to help individuals, cities, and other partners collectively reduce 1
gigaton of their carbon equivalent emissions annually by 2030.

To put the scale of this • Nest thermostats: In 2023, we estimate that

ambition into perspective, 1 GT Nest thermostats helped customers save
Our 1 GT aspiration of emissions more than 20 billion kWh of energy, 69 which
reductions is comparable to the we estimate enabled approximately 7 million
entire annual emissions of Japan. 67 metric tons of GHG emissions reductions. 70

So far, we’ve estimated the As other products continue to be developed

Sector GT Secto GT enabled GHG emission reductions
r and mature—including some of those
from described throughout this chapter—we’re
Forest, land use,
Energy 20 13 two products: working to understand their real-world impact
and agriculture
in enabling GHG emissions reductions. We
• Fuel-efficient routing: In 2023 alone, expect to share additional enabled GHG
8.7 Industrial 14 we estimate that fuel-efficient emissions reduction estimates in the future as
Transportation routing enabled more than 1.7 we progress toward our 1 GT aspiration.
A gigaton of million metric tons of GHG
Buildings 3.3 1
emissions emissions reductions—
Source for global emissions estimates: IPCC equivalent to taking
approximately 380,000 fuel-
 Introduct AI for Our Our Appen

While we’ve made some progress, many of the solutions to help others achieve a gigaton of carbon emission reductions
are either still in development or don’t yet exist, and estimating enabled emissions reductions is inherently difficult and
imprecise (see the
Our gigaton aspiration section of our 2023 Environmental Report to learn more about our estimation approach).

This ambition pushes us to innovate and be audacious in our approach and to collaborate with others to drive systemic
solutions.
 Introduct AI for Our Our Appen

Energy

To support the clean energy They can help save energy in a number
transition, we’re working to of ways, including setting up energy-
Solar Electricity grid Clean energy partners
enable more informed choices efficient schedules or automatically
Solar plays a significant role in the clean
decarbonization We have a number of Google Cloud
and reduce the climate impact making small changes to temperature partners that are building solutions to
energy transition. 74 To help, we’ve One of the biggest barriers to bringing
settings when energy use is surging and decarbonize the energy sector,
of energy use. developed several products and tools to more clean energy online is the
prices are high. These actions lead to a including mCloud, ElectricityMaps and
make solar more accessible, bringing complexity of connecting it to the grid
significant collective impact. In 2023, we Flexidao. Among
Home heating and estimate that Nest thermostats helped
solar information and technology to while maintaining overall reliability. Since
these, LevelTen Energy not only provides
homeowners, solar companies, and city renewable sources like wind and solar
cooling customers save more than 20 billion kWh
of energy 72—which we estimate enabled
planners around the world. are intermittent, governments and
access to the world’s largest clean
energy marketplace and the tools
Energy use in residential buildings approximately 7 million metric tons of utilities need to use sophisticated
In 2015, we launched Project Sunroof needed for efficient transactions, but is
contributes 50% to global building GHG emissions reductions. 73 models to do so safely. This can be a
to help people determine their homes’ also the first partner to benefit from
carbon emissions. 71 But making sense of slow process, with barriers like aging
solar potential. Demand for solar data Google’s partnership with HSBC to
the options can be overwhelming, from We’re also making it easier for people infrastructure and the inability to plan for
continues to grow and in 2023, with finance fast-growing climate technology
navigating financial incentives and total looking to upgrade their home heating or severe weather events often causing
help from our partners, we launched firms.
costs, to understanding efficiency ratings. cooling system. Search queries like delays.
the Solar API on the Google Maps
“boilers,” “heat pumps,” or “furnaces” in We’ve worked to integrate their
Platform. This enables us to That’s why we started Tapestry, a part
Nest thermostats use algorithms the United States, Germany, and France solutions on our Google Cloud
provide data on solar potential, high of X, Alphabet’s moonshot factory.
powered by AI and machine learning to now show information about sustainable Marketplace, supporting companies like
resolution rooftop imagery, and financial Tapestry aims to unlock access to
optimize residential heating and options—including their capabilities, Iron Mountain who share our ambition
insights to our customers and partners— clean, reliable, and
cooling systems and reduce energy energy efficiency, and financial to reach 24/7 CFE and accelerate
making obtaining solar energy more affordable electricity worldwide by
consumption. incentives—all in one place (see Figure progress toward grid decarbonization.
efficient, transparent, providing greater insight into our
6).
and effective. increasingly dynamic and complex
Our predictive analytics are offering
electricity system. Tapestry
We’ve used AI to provide detailed rooftop innovative ways to promote clean
is creating a single virtualized view of the
electricity system through AI-powered energy. For
tools
FIGURE 6 Sharing information about home cooling solar potential data for processes.

and heating options in Search more than 472 million

In 2023, we introduced a new
buildings across 40
generative design feature in
countries. This includes
Google Earth that makes it
newly expanded
easier for planners to
coverage to over 95%
determine the best building
of
designs and solar options for
all buildings in the
urban areas. It’s especially
United States. 75 Solar
helpful when optimizing
companies, like
designs for places with ample
Monalee, are using our
space, like car
solar data to transition
parks. This will help cities plan
more people to
for development with the right
renewable energy at a
information to meet their
fraction of the time
sustainability goals.
and cost of traditional
 Introduct AI for Our Our Appen

that can predict and simulate what might happen on the grid example, Engie partnered with Atlas AI to use its
from milliseconds to decades into the future. Google Cloud-powered analytics platform to
successfully identify potential off-grid solar
Tapestry worked with Google DeepMind to improve the grid customers in Kenya, boosting their sales by 48%
planning process by applying and enhancing GraphCast, an in one region.
AI model designed for fast and accurate global weather
forecasting. Tapestry and GraphCast’s collaborative
model
outperformed the state-of-the-art model, the European Centre
for Medium-Range Weather Forecasts’ HRES, by up to 15%. 76
These highly accurate wind forecast insights have already aided
wind prediction in Chile and can give grid operators worldwide
higher confidence in relying on variable renewable energy to
power their network.
 Introduct AI for Our Our Appen

Transportation

How to get around is a high- 2023, we started rolling out fuel- efficient Canada, Egypt, and nearly 40 European locations in Google Maps—by the end of
We’re also making the charging
impact choice people make routing to India and Indonesia, adding to countries. And, in India and Indonesia, 2023, Google Maps included more than
experience easier for EV drivers in
every day. We’ve designed our our existing coverage in the United we’re including fuel-efficient routes for
Google Maps and
400,000
products to help people make States, two- wheelers to help even more EV charging locations globally. Additionally,
people travel more sustainably. in the Waze app, drivers can find compatible
more informed travel
charging stations along their route. Thanks to
decisions.
Electric vehicles local map editors from our Waze Community,
EV data is reviewed and updated in near-real
Driving Electric vehicles (EVs) are a key time.
solution for decarbonizing road
70% of direct transport emissions When EV drivers who have cars with Google
transport and an
come from road vehicles, 77 so we’re built-in search for “EV charging” on Google
increasingly popular option with
helping to make driving more Maps, they’ll see charging stations nearby with
consumers. We’ve improved the
sustainable by providing information charger availability and last use. Google Maps
experience for searching for EVs with
on the most fuel- or energy- efficient users on Android and iOS are routed to the
new comparison features and tools to
route, available and compatible EV
understand cost and range (see Figure 7).
chargers, sustainable commuting
options, and more. FIGURE 7
• EV comparison: In the United States, EV features in Search
when people search for terms like
Fuel-efficient routing “best electric cars,” they can
quickly compare
By building AI models on the emissions the prices, battery range, and power
profile of different vehicle types, fuel- output of individual models. They can
efficient routing in Google Maps also find federal government
analyzes traffic, terrain, and the incentives for qualifying vehicles in the
vehicle’s engine (gas/petrol, diesel, United States and parts
hybrid, or electric) to find the most of Europe.
efficient route. This may mean fewer
• Fuel Cost Calculator: Our updated Fuel
stops for gas engines, routes favoring
Cost Calculator, currently available in
highway speeds for diesel vehicles, and
21 countries, now includes results for
maximizing downhill stretches for
both electric and fuel-based cars,
electric cars to boost regenerative
helping drivers understand the cost of
braking—all while providing the same
charging compared with filling up at
or similar ETA.
the pump.
As of the end of 2023, fuel-efficient • Battery Range Explorer: With Battery
routing is estimated to have helped Range Explorer, searching for an EV
enable more than 2.9 million metric tons shows how far one can drive on a
of GHG emissions reductions since the single charge, specific to that model.
feature launched in late 2021—equivalent These battery range insights launched
to taking approximately 650,000 fuel- in the United States in 2023 and are
based cars off the road for a year. 78 In planned to roll out in Europe in 2024.
 Introduct AI for Our Our Appen

most convenient charging location based excessive fuel consumption and engineers can use to optimize Recommendations in these cities
on factors like traffic, charge levels, and harmful air pollution. That’s why we the timing of traffic light can save fuel and lower emissions
charger speed, and can see data on real- created Green Light, an AI tool that changes to reduce stop- and- for up to 32 million car rides
time charging availability, charger measures driving trends at go traffic. monthly. 79 Early numbers from
compatibility with their vehicle, and intersections and develops Green Light indicate the potential
charger speed. recommendations that city traffic In 2023, recommendations from to reduce stops by up to 30% and
Green Light were implemented in reduce emissions at intersections
Urban traffic 12 cities across four continents, by up to 10%. 80
including Bengaluru, Kolkata, Rio
Many cities suffer from congested de Janeiro, and Budapest.
urban road traffic which results in
Waze. This starts with having EV EV comparison Fuel Cost Calculator Battery Range Explorer
charging
 Introduct AI for Our Our Appen

Transit and Flying Since only a small number of flights avoidance could be in the range of $5–
need to be adjusted to avoid the $25/ton CO2e, making it a very cost- Authoritative information
urban mobility majority of contrail warming, the effective climate solution. 84 Now, we’re
We’re working on a number of solutions When people come to our products and platforms lo
We’re working to empower people with incremental fuel impact could be quite broadening our efforts via a new
that help mitigate the climate impact of
small when considering an airline’s partnership with EUROCONTROL to test On Search, for example, we’re making it easier for pe
information on a wider range of transit flying. When individuals search in
entire fleet—potentially as low as 0.3%. contrail avoidance technology in dense
and urban mobility options. We now Google Flights, they see carbon 83
Our analysis suggests that the cost of European airspace. On YouTube, we raise authoritative voices on climate
feature more transportation options in emissions estimates for nearly every
reducing contrail formation through in search results and videos, particularly for newswor
Search, showcasing long-distance train flight—right next to price and duration
navigational
routes in 38 countries and long-distance (see Figure 8). If people want to view And we have a for Google advertisers, publishers, an
FIGURE 8
bus routes in 15 countries. For only flights that have lower emissions
instance, high-speed rail suggestions compared to the average for similar trips, aviation community so it can safely reroute planes to reduce the chances
may appear when searching for flights they
of contrail formation. In one trial consisting of 70 test flights, use of the predictive technology reduced contrails by 54%. 82
when the travel time is similar. can simply tap the “Less emissions only”
filter.

We’re also making public transit and

We created the Travel Impact Model (TIM),
commuting in urban areas easier with
a public model for predicting the per-
Google Maps:
passenger CO2e emissions produced by a
flight. This powers the emissions
• Public transit: Users can access bus,
estimates you see
train, subway, and ferry directions by
on Google Flights as well as other
tapping on the Google Maps transit
leading travel sites through our work in
icon. Google Maps provides, on
the Travalyst coalition. We’ve seen
average, more than
positive industry adoption of our model
3 billion kilometers (1.86 billion
in aviation and, in 2023, we formalized
miles) of public transit results per
our efforts by establishing an
day.
independent advisory committee to
• Bike routes and bike shares: In 2023, we oversee future changes to the TIM.
added over 30,000 kilometers (18,000
miles) of bike lanes to Google Maps. Contrails—the thin, white lines that
Individuals can also find nearby bike can appear behind aircraft—trap heat
and scooter shares in over 400 cities within the earth’s atmosphere and are
around the world. significant contributors to climate
• Walking directions: Pedestrians can change, accounting for about 35% of
easily get walking routes, turn-by-turn aviation’s contribution to global
directions, and use Live View to get warming emissions. 81 To mitigate
augmented reality assistance on the contrails, Google Research partnered
go. Google Maps offers Immersive View with
and Street View to preview a route Breakthrough Energy and American
before you go. Airlines to develop an AI-based tool to
predict where contrails will form. We
share insights from this tool with the
 Introduct AI for Our Our Appen

Google Flights emissions

estimates
 Introduct AI for Our Our Appen

Additional sectors

emissions reductions and receive perception tools to bring visibility to our

Forests, land use, payments based on positive results. Farmland ocean ecosystems, so we can better
protect them and help solve some of
humanity’s biggest challenges—from food
and agriculture For more details on Google Earth
Increasing agricultural productivity and
understand and production to climate change. In 2023,
Engine and Google Earth, see the Tidal was included in Time Magazine’s
Forests sustainability is imperative to feed the
Multi-sector products section in Best Inventions of 2023 for its undersea AI
planet’s growing population. 87
the Appendix. capabilities. Tidal is currently working with
Every year the world loses around 10
Google Cloud provides customers and aquaculture partners across the globe,
million hectares (24.7 million acres) of
partners with a number of insights including Norway-based Mowi, the world’s
forest, 85
related to agricultural productivity. Our largest Atlantic salmon producer. From
an area roughly the size of Hungary. 86
Cloud partner Airbus owns and launching commercially in early 2023 to
operates optical and radar satellites, February 2024, Mowi has installed Tidal
Google’s geospatial tools are
and uses their data to, for example, systems in nearly 300 pens across Norway—
empowering innovative solutions to
help farmers minimize excessive water making Tidal’s insights available for improving
protect forests and avoid
and fertilizer use while maintaining or the sustainability and efficiency of its
deforestation. For example, we’re a
increasing crop yields. KfW—a German aquaculture operations.
founding partner of the Forest Data
Partnership, which aims to stop development bank—uses satellite data
through our
deforestation caused by commodity
Cloud partner UP42 to monitor
Industrial
production. Our
Cloud partners, like NGIS (and their irrigation infrastructure along the Niger The industrial sector is critical in the fight
solution TraceMark) and Satelligence, River, which is key to food security against climate change and it presents unique
help companies on their journey toward across the region. challenges in mitigating GHG emissions.
compliant and sustainable supply chains At Google, we’re using cutting-edge
by assessing deforestation and other Ocean ecosystems technologies to support organizations in
environmental risks using satellite addressing emissions like methane leaks from
technology and data processing on Understanding and protecting the oil and gas sector.
Google Cloud. And our Cloud partner ocean ecosystems is essential
Kumi Analytics can help estimate the for promoting sustainable
carbon captured from reforestation for practices and healthy marine
carbon credit issuance, using a scalable, environments.
transparent, and Google Cloud-powered
assessment tool. Founded in 2015 as a collaboration
between Oceana, SkyTruth, and Google,
The UN Food & Agricultural Organization Global Fishing Watch is an open-source
has been working for nearly a decade to platform that uses satellite data and
create measurement, reporting, and machine learning to monitor the planet’s
verification (MRV) tools that countries fisheries. In 2023, Global Fishing Watch
can use to map, monitor, and report was awarded $60 million by the TED
their forest change. Google Earth Engine Audacious Project. To learn more, see
Global Fishing Watch’s 2023 TED talk.
and Google Earth are being used for
MRV, enabling countries
Tidal—founded as a project at X—uses
to better measure associated
underwater camera systems and machine
 Introduct AI for Our Our Appen

Methane leaks Additionally, our Cloud partner

BlueSky Resources uses Google
Methane is a potent greenhouse gas, Earth Engine to integrate a range
over 80 times more powerful than of sensor data and satellite
carbon dioxide. Tackling methane imagery, empowering
emissions effectively will require organizations to pinpoint
collaboration between industry methane emission sources and
stakeholders and the public sector to develop targeted mitigation
track emissions and implement strategies.
mitigation efforts.

Recycling
To help, the Environmental Defense
Fund (EDF) will use a new satellite—
In addition to recycling features
MethaneSAT— to map, measure, and
within our products, we’re
track methane
working on technology pilots that
with unprecedented precision, offering
advance recycling process
a comprehensive view of oil and gas
optimization for the industry. In
methane emissions globally. EDF’s
2023, X deployed its
algorithms, powered by Google Cloud,
first advanced sortation
will calculate the amount of methane
technology pilot in a materials
emitted in specific places and track
recovery facility in Oregon,
those emissions over time (see Figure
demonstrating the ability to
9).
achieve high resolution
materials identification and
As part of a partnership with EDF, we’re
intelligent sortation at
using this information to build a global
industrial speeds and scale.
map of oil and gas infrastructure to
help identify the components most
Meanwhile, Google’s AI
responsible for methane emissions.
technology, CircularNet, powers
These insights will be available later this
Recykal’s waste management
year on MethaneSAT’s website, and
platform—Asia’s largest circular
accessible through Google Earth
economy marketplace—
Engine.
enabling the identification of
recyclable materials with high
accuracy. FIGURE 9 MethaneAIR on Google Earth Engine
EDF’s aerial methane emissions data, or MethaneAIR—available in Earth
Engine—shows both high-emitting point sources as yellow dots, and diffuse area sources as a purple and yellow heat map. Metha
with more frequency.
RESOURCES
Introduct AI for Our Our Appen

Adaptation and resilience

We aim to develop tools and technologies that help communities adapt to the effects of climate change

Our approach

Extreme event forecasting

and early warning systems
Weather and climate
forecasting Wildfire
detection
Flood
forecasting
Extreme heat
Air quality conditions
 Introduct AI for Our Our Appen

Our approach to climate adaptation and resilience

We’ve discussed the ways FIGURE 10

expanding extreme event forecasting and early warning systems.
we’re helping people make 2023 highlights Global challenges
more sustainable choices to Crisis Alerts for wildfires: Our provided timely wildfire information
Data gaps to over 30
and uncertainties: million users
Accessing across
reliable local18
cl
mitigate climate change, but Flood Hub: displayed forecasts for over 80 countries on five
Long-term continents,
planning: Climateincluding some
adaptation often requires
its effects are already here. of the territories with the highest percentages ofLocal
population exposed
collaboration: to flood risk—covering
Successful more tha
adaptation requires co
Extreme heat information: Metrics and measurement: The inherent complexity of
2023 was the hottest year on Since launching, we’ve provided information about on Search in more than 100 countries.
record, the past decade was Weather and climate forecasting Cool Roofs pilot: Our pilot was available in 15 cities.
Our innovations in forecast modeling and cloud simulation technology are providing more accurate
Tree Canopy and timelyTree
data: We expanded weather anddata
Canopy climate information
in EIE to more than 2,000 cities globally on
the hottest ever recorded, 88 four continents.
and communities worldwide—
especially the most vulnerable
— are experiencing the
devastating effects of floods,
wildfires, food insecurity, and Wildfire detection
more. Our alerts, simulation models, and wildfire risk forecasting systems are helping provide accurate wildfire information to affected communities and fire authorities

During extreme weather events,

people often turn to our platforms for
information. Global search interest in
“heat wave” and “heat exhaustion”
both spiked during the Flood forecasting
month of July 2023, 89 during the warmest Our AI-powered flood forecasting platform can predict when and where riverine flooding will occur up to seven days in advance
July ever recorded. 90 For events like these,
timely information is critical to help
minimize damage and loss.

To help communities adapt to the

effects of climate change and stay safe,
Extreme heat
we’re building technologies to model, Through extreme heat alerts on Search and tree canopy and reflective roof data in the Environmental Insights Explorer (EIE), our insights are helping keep people cool and safe
predict, and respond to extreme
weather events, wildfires, floods,
extreme heat, and air quality conditions
(see Figure 10).

AI plays a critical role in powering these

platforms—enabling more accurate Air quality conditions
predictions, real-time data analysis, and Our tools provide hyperlocal air quality data, enabling informed health decisions and strategies for cleaner air
personalized insights to aid in
adaptation efforts. We’re also leveraging
AI to create more accurate models in
data-scarce regions, improving and
 Introduct AI for Our Our Appen

A ng and early warning systems

r
e
a
s

o
f

f
o
c
u
s

f
o
r

e
x
t
r
e
m
e

e
v
e
n
t

f
o
r
e
c
a
s
t
i
 Introduct AI for Our Our Appen

Extreme event forecasting and early warning systems

planning and post- fire recovery.

Weather and climate clouds, the most prevalent cloud type. patterns with unprecedented detail, engine for the Earth and everything on it—has
Using TPUs and high-performance offering a glimpse into the future built an AI-first wildfire risk forecasting system
forecasting computing originally designed for climate.
To help prevent catastrophic fires in
to understand the future probability of wildfire
the future, X’s Project Bellwether—as
Our innovative weather and climate machine learning applications, we’re in the United States, Canada, and Australia. X
part of its prediction
forecasting tools are providing able to simulate cloud Wildfire detection partnered with SwissRe Reinsurance Solutions
and others to bring these critical insights to
communities with crucial information to
plan for and respond to a changing Climate change is increasing the the insurance market, where better tools are
climate. frequency and severity of wildfires, needed to keep homeowners insured.
which are having a devastating impact
on communities and This work is important for fostering a more
Weather forecasting
ecosystems around the world. 92 This resilient society, as it enables better
growing threat is reflected in the all-time understanding of wildfire risks and informs
Google’s MetNet-3 is a state-of-the-art
high global searches for “air quality actions that can protect communities and
neural weather model that outperforms
index” and “wildfire” in June 2023, 93 homeowners from the devastating impacts of
leading operational forecasting systems.
coinciding with the peak of the Canada these events.
Its precise precipitation forecasts for 12
hours into the future have been wildfires that year. 94
integrated into various Google products,
To minimize the impact of wildfires on
Flood forecasting
enhancing the information and tools
communities, Google has developed an Floods are among the world’s deadliest
available to people. Meanwhile, Google
AI- based technology to map wildfires natural disasters, and climate change is
DeepMind’s GraphCast model delivers
in near real time in Google Search and causing more frequent and severe floods. 95
10-day weather predictions at
Maps. The tool Our Flood Forecasting Initiative, launched
unprecedented accuracy in under one
uses data from geostationary satellites in 2018, uses advanced AI and geospatial
minute, and offers earlier warnings of
such as NOAA’s GOES constellation to analysis to predict when and where
extreme weather events. And, in 2023,
show the size of a wildfire, with data riverine flooding will occur up to seven
Google Research launched
refreshed roughly every 15 to 20 minutes. days in advance.
WeatherBench 2, a platform that
In 2023, our Crisis Alerts provided timely
enables researchers to compare global
wildfire information to over 30 million
weather models and drive innovation in
users across 180 wildfire events around
this field.
FIGURE 11
the world. Google’s Flood Hub platform
Cloud simulations We’ve also been working closely with
U.S. agencies, such as the U.S. Forest
Google Research, in collaboration with a
Service (USFS), on wildfire research by
Climate Modeling Alliance (CliMA) lead, is
using advanced simulation and AI
achieving breakthroughs in cloud techniques to improve their fire-spread
modeling, significantly improving model. Additionally, USFS used Google
climate model accuracy. Today’s climate Earth Engine and Google Cloud
predictions vary widely, largely because technology to build a Landscape
it’s so difficult to simulate clouds Change Monitoring System to map and
accurately—which is why we’re monitor land cover change across the
focusing on simulating stratocumulus United States, supporting forest
 Introduct AI for Our Our Appen

The system, called Flood Hub, displays

flood forecasts to help governments, Extreme heat
aid organizations, and at-risk
In 2023, we launched extreme
communities take timely action (see
heat alerts, so when people
Figure 11). These breakthroughs are a
search for information on
result of innovative modeling
extreme heat, they see details
approaches that use machine learning
they need
to create scalable models in real-
about when a heat wave is
world settings.
predicted to start and end, tips for
staying cool, and related health
In 2023, Flood Hub displayed
concerns to be aware of from the
forecasts for over 80 countries on
Global Heat Health Information
five continents, including some of the
Network (see Figure 12). Since
territories with the
launching, we’ve provided
highest percentages of population
information about extreme heat on
exposed to flood risk—covering more
Search in more than 100
than 460 million people globally. 96 In late
countries.
2023, we announced the expansion of
riverine flood forecasts on Flood Hub to
the United States and Canada, covering FIGURE 12
Extreme heat
more than 800 locations by rivers where
more than 12 million people live. 97
alerts

In 2023, we generated nearly 700 Crisis

Alerts that shared critical flood
information, including mappings of
estimated flooded areas, which were
viewed by more than 10 million users.
 Introduct AI for Our Our Appen

Tree canopy 80% of the U.S. population. The success vulnerability. Alongside Google.org’s efforts, WRI’s project will help decision-makers understand where to implement cool surface
of this project has led to the development infrastructure—such as trees and reflective surfaces—to reduce the impact of extreme heat.
Cities are looking for ways to prevent of a similar tool for the United Kingdom.
“heat islands”—urban areas that
experience higher temperatures due to Cool roofs
structures like roads and buildings that
absorb and re-emit heat. Our Tree Canopy We’re exploring how our technology,
tool, part of our EIE platform, uses AI and such as AI algorithms and aerial
aerial imagery to detect and imagery, can help
map tree canopy coverage in cities to implement reflective roofs—called “cool
help them plan future tree planting roofs.”
projects more effectively (see Figure
13). These energy-saving roofs reduce
indoor and outdoor temperatures,
For example, the city of Austin, Texas making them especially impactful in
used insights from this tool to develop communities without reliable air
Austin’s Community Tree Priority Map, conditioning. By mapping urban solar
helping to focus tree planting efforts in reflectivity, we can help planners and
the eastern part of governments identify areas where cool
the city where tree coverage was lower roofs would be most impactful. As of
and ambient temperatures were higher. the end of 2023, this pilot was available
in 15 cities.
In 2023, we expanded Tree Canopy
data to more than 2,000 cities Addressing extreme heat requires
globally on four continents. With the collaboration, so Google.org is
aim of making shade in cities more supporting nonprofit projects that
equitably distributed, we identify which communities are most
partnered with American Forests in the affected by extreme heat and where
United States to make our tree canopy there are opportunities to help. In
data available on its Tree Equity Score 2022, we announced a $30 million
tool. American Forests now provides tree Google.org Impact Challenge—an
canopy data for nearly open call for nonprofits to submit big
ideas for climate action.

FIGURE 13 Tree Canopy data

In 2023, we announced the first recipient:
for Lisbon, World Resources Institute (WRI). WRI
Portugal received
$5 million to support its project to use
sensors, satellite imagery, and AI to
close data gaps and model air
temperature, humidity, surface
reflectivity, tree cover, and heat
 Introduct AI for Our Our Appen

n erlocal air quality data empowers

Air quality conditions e individuals, companies, researchers, and
W policymakers to make informed health
i p choices and develop strategies for cleaner
t o air.
h l
l In 2023, we introduced a new suite of
c u Environment APIs from Google Maps
l t Platform, which included:
i i
m o • Air quality: The Air Quality API shows
EIE’s air quality data for Bengaluru, India.
a n robust air quality data, pollution
t , heatmaps, and pollutant details for
e nearly 100 countries around the world.
IKE Smart Cities is using this data to SPOTLIGHT
a
c n alert residents of poor air quality and
h d provide city partners with a Hyperlocal air quality data in Bengaluru, India
a dashboard to monitor conditions EIE provides select cities with street-by-street air quality data, helping the
n a alongside other relevant data. This
g l helps cities implement protective
e l measures for citizens and visitors.
CSTEP
By equipping Google Street View vehicles with air collected air quality
pollution sensors anddata by installing
partnering sensors
with local org
e • Pollen: Our new Pollen API shows analyzing hyperlocal
real-time, hyperlocal modeled air quality intelligence, data.
Google aims to make
i r current pollen information for the invisible air pollution easier to address.
n g most common allergens in over 65
t y countries. The API provides localized
e pollen count data, heatmap
n s visualizations, detailed plant allergen
s e information, and actionable tips for
We then integrated CSTEP’s calibrated data into EIE,
i a allergy-sufferers to limit exposure. partners.
f s
y o RESOURCES
i n
s In 2023, we launched street-by-street in EIE—our first city to launch in India. This launch was t
n
(CSTEP) and .
g ,

9
w 8
i
l G
d o
f o
i g
r l
e e
s ’
, s

o h
z y
o p
Introduct AI for Our Our Appen

Our operations
Our approach

Net-zero carbon

Water stewardship

Circular economy

Nature and biodiversity

 Introduct AI for Our Our Appen

Our approach to operating sustainably

We’re working to advance our ambitious sustainability goals, showing the way forward through our own
operations

In 2023, Google had offices customers and users rely on— key ways: accelerating the transition to a
and data centers on six including Google Cloud, Google net-zero future, advancing water
continents, in over 200 Search, Google Workspace, and stewardship, building a circular economy,
cities, across nearly 60 YouTube. and restoring and enhancing nature and
biodiversity.
countries.
That’s why our work on sustainability
started with understanding and working We’re also working worldwide with
We use energy and natural
to address the environmental footprint suppliers through our Supplier
resources to build and operate our
of our operations. Responsibility Program. We collaborate
data centers and offices around the
with stakeholders across our supply
world, and to power the many
At Google, we’re working to drive chain to uphold our high standards for
products and services that our
sustainability across our operations in respecting workers and the
four environment.

FIGURE 14
Google’s global operations

6 200+ ~60
continents cities countries

24 44 ~40 Wind turbines spin near our Eemshaven, Netherlands data center.
center locations data
99 center electricity grid regions (including third party)
Introduct AI for Our Our Appen

Net-zero carbon
We’re working to accelerate the transition to a net-zero future at Google and beyond

Our approach

Our net-zero goal

Reducing carbon emissions

Scope 1 emissions
Electrification
Refrigerant
mitigation
Renewable fuels
Scope 2 emissions
Energy
management
Carbon-free
energy
Scope 3 emissions
Supplier
engagement
Embodied carbon
Sustainable travel

Managing residual emissions

 Introduct AI for Our Our Appen

Our approach to net-zero carbon

We’re working to accelerate help accelerate global

the transition to a net-zero decarbonization. To ensure our efforts
future, and we’ve taken maximize impact, we’ll continue 2023 highlights Global challenges
significant steps over the past evaluating our plan on a regular basis
to ensure it’s rigorous, grounded in CFE: We maintained 64% carbon- free energy, on average,
Need acrosschanges:
for systemic every grid
Transitioning
where we operate—even
to a net-zero future
as our will
electricity
requireuse
stronger
increased.
government polici
two decades to minimize our
science, and realistic in light of
GHG emissions. evolving challenges and standards.
Clean energy procurement: We signed contracts Evolving
to purchase approximately
standards 4 GW ofFaster
and regulations: clean energy
reform and
generation
greatercapacity
harmonization
100— more
of various
than in climate-relate
any prior yea

100% renewable energy: Hard-to-decarbonize industries and regions: Global operations and value chains that span a diverse ran
In 2021, we set an ambitious goal to
We’re engaging in advocacy efforts, We achieved seven consecutive years of 100% renewable energy matching on a global and annual basis. 101
reach net- zero emissions across all of Collaborating with suppliers: The maturity of supplier climate programs and their commitments to tak
exploring data center innovations,
our operations and value chain by 2030. Energy innovation: We developed, piloted, and advanced innovative new approaches to energy management and tracking, including and (T-EACs).
accelerating global grid
We’re working toward this goal in two Availability of carbon-free energy: There are often long lead times between investments and resulting
decarbonization, and advocating for
key ways: first, we’re focused on Carbon removal procurement: We completed our first carbon removal credit offtake deals through Frontier in 2023—including deals with , , and .
GHG Protocol reform to help drive
reducing emissions across our operations Cost and availability of carbon removals: Due to the early stage of the carbon removals market, the vo
system- level change. For more details,
and value chain (including advancing
see the Policymakers and Partners
24/7 CFE), and after reducing our
sections in the Appendix, as well as
emissions, we’re addressing our residual
Energy in the Our products section for
emissions with carbon removals (see
how we’re working
Figure 15).
to advance clean energy development
through Google’s core capabilities in
Ultimately this isn’t just about Google
data and software.
—our net-zero goal is a key tool by
which we can

FIGURE 15 We prioritize emissions reductions first before

compensating for residual emissions

Reduce emissions across our

Reducing carbon Reedmuicsisnigo ncsa rbon emissions
operations and value chain tions
Reduceueemissions across our opera
chain through
and val through variousvariou
levers s levers

After reAdf
tuceirn rge odurc ienmg iossuiro enms, iwsseio’lln s, work tow aed’dll rweossrk o tuor aredsdidreusasl oemuri ssions throughr ecsairdbuoanl eremmisosvioanl s othlurtoioungsh
Managing residual carbon removal solutions
emMiasnsaiogninsg residual
emissions
 Introduct AI for Our Our Appen

Our net-zero
goal TARGET

Reduce 50% of our combined Scope 1, 2 (market-based), and 3 absolute GHG emissions
In 2021, we set a goal to reach to rise before dropping toward our
absolute emissions reduction target. by 2030, 102 and invest in nature-based and technology-based carbon removal solutions
net- zero emissions across all
of our operations and value to neutralize our remaining emissions
Our net-zero goal is aligned with the
chain by 2030. IPCC’s definition of “net zero Year set: 2021; Base year: 2019 103; Target year: 2030
emissions,” which is “when
To meet this goal, we aim to reduce 50%
anthropogenic emissions of
of our combined Scope 1, Scope 2
greenhouse gasses to the
(market-based),
atmosphere are
and Scope 3 absolute emissions
(compared balanced by anthropogenic removals 2023 PROGRESS TREND
to our 2019 base year) by 2030, and we over a specified period.” The world’s Emissions reductions: In 2023, our total
plan to invest in nature-based and understanding of “net zero” remains in Emissions reductions: Total GHG emissions were 14.3 GHG emissions increased 13% year-
technology- based carbon removal a dynamic state and is subject to
million tCO2e, representing a 48% increase compared over- year, partially driven by a 37%
solutions to neutralize our remaining refinement as global consensus year-over-year increase in our Scope 2
emissions. We’ve formally committed to develops. We’ll proactively monitor to 2019 (market-based)
the Science Based Targets initiative (SBTi) the evolution of global standards to emissions. Our total GHG emissions
to validate our absolute emissions ensure our definition maintains general Residual emissions: Signed offtake deals for increased at a slower rate compared to
reduction target. alignment while maximizing our positive approximately 62,500 tCO2e of removal credits the previous two years. For trend details
impact on the planet. for each scope, see the Scope 1
Reaching net-zero emissions by 2030 is emissions, Scope 2 emissions, and
an extremely ambitious goal and we In 2023, our total GHG emissions were Legend Scope 3 emissions sections.
know it won’t be easy. Our approach will 14.3 million tCO2e, representing a 13% Total emissions
Residual emissions: 2023 marks the first
continue to evolve and will require us to year-over- year increase and a 48%
Business as usual (footprint in absence of PPAs) year of implementation of our carbon
navigate significant uncertainty– increase compared to our 2019 target
removals strategy, and while we have a
including the uncertainty around the base year. This result was primarily due
20 long way to
future environmental impact of AI, which to increases in data center energy
go to meet our 2030 target, we’ve begun
is complex and difficult to predict. In consumption and supply chain
establishing impactful partnerships and
addition, solutions for some key global emissions. As we further integrate AI
14.3 15 have started contracting for carbon
challenges don’t currently exist, and will into our products, reducing emissions
removal credits. For more detail, see the
depend heavily on the broader clean may be
12.6

Million
Managing residual emissions section.
energy transition. challenging due to increasing energy
10.8 10
demands from the greater intensity of AI
As our business and industry continue
9.7
compute, and the emissions associated 8.6 DETAILS
to evolve, we expect our total GHG with the expected increases in our Emissions reductions: All Scope 1, 2 (market-
5
emissions technical infrastructure investment. based), and 3 absolute emissions across our
operations and value chain, including our
data center and office operations, supply
chain, and consumer hardware devices.
Residual emissions: Carbon removal
credits to neutralize our residual
 Introduct AI for Our Our Appen

emissions by 2030.

0
2019 2020 2021 2022 2023
Rødby solar farm in Denmark (55 MW for Google)
 Introduct AI for Our Our Appen

Reducing carbon emissions

In 2023, our total GHG Electrification Refrigerant mitigation

emissions were approximately Carbon footprint Scope 1 emissions
14.3 million metric tons of recalculation In 2023, our Scope 1 emissions were Electrification—followed by clean We’re working to more accurately
carbon dioxide equivalent In 2023, we recalculated certain
approximately 79,400 tCO2e, energy procurement—is an essential measure refrigerant leak rates and
(tCO2e), which represents our representing approximately 1% of our first step toward decarbonizing our develop new technologies to reduce
previously reported GHG emissions
total carbon footprint. Compared to office and shuttle fleet operations. them. At our data centers, we’re
Scope 1, Scope metrics in accordance with our internal
2022, we reduced our Scope 1 developing and deploying cooling
2 (market-based), and Scope 3 recalculation policy for improved
We’re working to electrify our office
emissions by 13% due to building solutions that include natural, low-
emissions (see Figure 16). accuracy.
electrification and decreases in portfolio, including retrofitting existing GWP refrigerants. And in 2023, we
emissions building systems, incorporating completed 24 Technical Feasibility
We’ve analyzed our operations and See the Recalculation of previous
from transportation and data center electrification standards into new Studies for refrigerant phase-out for
value chain to pinpoint specific levers environmental metrics section in the
generator use. Our main sources of Scope developments (such as designing offices in the Asia-Pacific region.
that will drive carbon reductions across Appendix for more information on our
1 emissions include natural gas use, all-electric kitchens), and working with
Scope 1, Scope 2, and Scope 3 carbon footprint recalculation, and the
refrigerant leakage property owners on lease terms that We also continued working with
emissions. While our reduction efforts Environmental data tables for more
in our data centers and offices, fuel use support building electrification. industry partners to identify
are crucial, they alone won’t get us all details on our GHG emissions.
from back-up generators, and innovations in technology and
the way to our net zero goal. In 2023, we completed 14 full or refrigerants. In 2023, we submitted a
transportation (i.e., company vehicles
and aircraft). Given this, some of our partial office building electrification follow-on patent to the one we filed in
key Scope 1 emissions reduction levers projects around the world—including in 2022, incorporating the use of large
include electrification, refrigerant India, Brazil, and the San Francisco language models for “intelligent
mitigation, and renewable fuels. Bay Area—and we pledged to upgrade diagnostics.” Based on our pilots,
the iconic Thompson Center in intelligent diagnostics can provide
FIGURE 16 Our 2023 carbon footprint Chicago to an all-electric office operators with guidance to locate
building. And in 2023, we signed our leaks, helping to both curb refrigerant
Scope 1: Scope tCO2e % first leases in the Google office emissions and reduce energy
79.4K tCO2e portfolio with consumption from faulty equipment.
Scope 1 79,400 1%
electrification provisions—located in the
Scope 2:
Scope 2 (market-based) 3,423,400 24% Asia- Pacific region—with more
3.4M tCO2e Renewable fuels
currently under negotiation.
Scope 3: (1) Purchased goods and services 4,038,000 28%
We’re exploring ways to reduce the
Scope 3: (2) Capital goods 1,605,000 11% Our Moffett Park Thermal Plant in carbon footprint of our data center
Scope 3: (3) Fuel- and energy- Sunnyvale, California, demonstrates backup power. One solution is
1,186,000 8%
related activities (not included in electrification at scale—as an all-electric renewable diesel, which offers lower
Scope 1 or 2)
central utility plant that will efficiently life cycle carbon emissions compared to
Total emissions: Scope 3: (4) Upstream transportation 584,000 4% deliver heating and cooling services to fossil fuels. In 2023, we piloted
14.3M tCO2e some nearby Google offices.
Scope 3: (5) Waste generated in 10,000 <1% renewable diesel in select U.S. and
operations European data centers, and plan to
Scope 3: (6) Business travel 283,000 2% And for our shuttle fleet operations, in scale this program globally as
2023 we incorporated additional renewable diesel availability increases.
Scope 3: (7) Employee
113,000 <1% electric shuttle buses into our fleet in
commuting (including And we use renewable diesel
teleworking) the San Francisco Bay Area, increasing for some of the non-electric vehicles in
Scope 3: Other categories 2,993,000 21% the electric vehicles in our fleet to 14%. our corporate fleet.
Scope 3:
Scope 3 (total) 10,812,00 75%
10.8M tCO2e
0
 Introduct AI for Our Our Appen

Scope 2 emissions agreements (PPAs), we achieved a 63% projects becoming operational, and the compute they process. On average, a In 2023, we piloted a new demand
reduction in the emissions from our current mismatch between our Google-owned and -operated data response capability. By shifting some
In 2023, our Scope 2 (market-based) electricity use in 2023 (see Figure 18). And approach to CFE and the GHG center is approximately 1.8 times as non-urgent compute tasks to other
emissions were approximately 3.4 million Protocol’s Scope 2 guidance. In fact, energy efficient as a typical enterprise times and locations, we work with
from 2011 to 2023, our carbon-free
tCO2e, representing 24% of our total despite achieving a 100% data center, 107 and compared to five local utilities to reduce our data
energy purchasing has resulted in
carbon footprint. Our main source of global renewable energy match, our years ago, our data centers deliver centers’ power consumption during
cumulative emissions savings of more
Scope 2 emissions is purchased electricity Scope 2 emissions have increased (see nearly four times as much computing high-stress periods to help grids
than 36 million tCO2e 105—equivalent to
for our data centers more details on this mismatch on the power with the same amount operate more reliably and meet the
taking more than 8 million fuel-based
and offices. Given that we have more following page). of electrical power. 108 needs of
cars off the road for a year, or the carbon
direct control over our data centers local communities.
sequestered by more than 42 million
and offices than many other parts of Energy management In 2023, the average annual power usage
our value chain, Scope 2 emissions acres of U.S. forests in a year. 106
effectiveness (PUE) for our global fleet of To improve energy efficiency in our
are a key focus of our Our data centers remain some of the data centers was 1.10, compared with the offices, we use data analytics to guide
However, compared to 2022, our Scope
decarbonization efforts. Some of our most efficient in the world, and we industry average of 1.58 109—meaning that strategies like equipment upgrades,
2 (market-based) emissions—which
key Scope 2 emissions reduction continue working to optimize their use Google data centers used about 5.8 times optimized building settings, and
originate primarily from our data center
levers include energy management of electricity, water, and materials. less overhead energy for every unit of IT efficient lighting retrofits.
electricity consumption—increased by
and carbon-free energy procurement. equipment energy. Since 2012, our For example, in 2023, we rolled out
37%, despite considerable efforts and
To optimize energy consumption at our average annual fleet-wide nearly 20 energy conservation
As of the end of 2023, we achieved progress on carbon- free energy. This
data centers, we strive to build the PUE has stayed at or below 1.12. For initiatives across our offices in the Asia-
seven consecutive years of 100% was due to data center electricity
world’s most energy-efficient compute more information on our practices Pacific region, which we
renewable energy matching on a global consumption outpacing our ability to
infrastructure, outfitting each data managing AI’s energy consumption and estimate helped reduce energy
and annual basis, even as our electricity bring more CFE projects online,
center with high- performance servers environmental impact, see the AI for consumption by nearly 290 MWh, as
consumption increased (see Figure 17). specifically in the United States and
designed to use as little energy as sustainability section. compared to the prior year.
104
Through our power purchase Asia-Pacific region, CFE contracts
possible compared to the amount of
terminating prior to those

FIGURE 17 FIGURE 18
Renewable energy purchasing compared with total electricity Annual impact of CFE purchases on Scope 2 emissions
Legend Legend 10

Total electricity 25.3 26

Scope 2 location-based emissions (without CFE
24
consumption Renewable 100%
22 purchases) Scope 2 market-based emissions (with
energy purchases 20 9.3
CFE purchases)
18 8
16

Terawatt-
63%
14 8 reduc
tion
12 6 6

Million
Google first 10
achieved 5
100% 8 5 4
renewable
energy 6
match
Google
announced
4 3 2
100% 2

0 0
 Introduct AI for Our Our Appen

2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023
 Introduct AI for Our Our Appen

Golden Hills wind farm in California (43 MW for

Google)

SPOTLIGHT

Why our Scope 2 emissions have increased despite achieving a global 100% renewable energy match and maintaining 64% carbon-free energy
We believe that our greatest impact lies not only in advancing CFE for our own operations, but also in accelerating decarbonization of electrical grids around the world.

We started addressing our Scope 2 Europe), while in other regions we transition. Our pursuit of 24/7 CFE remains one of our primary approaches to reducing our Scope 2 emissions. For
emissions in 2010 with our first power purchase less (such as in the Asia-Pacific more details about our methodologies, see the Methodology section in the Appendix.
purchase agreement from a new wind farm, region) due to significant regional
and in 2017, we became the first major challenges in sourcing clean energy. The Further, we believe standards should move toward this approach, and to that end, we’ve
company to match 100% of the annual discrepancy between our energy shared feedback on how the GHG Protocol could update its Scope 2 Guidance to
electricity consumption of our global consumption and clean energy purchases improve the accuracy
operations with renewable energy in these regions results in Scope 2 of carbon inventories and still allow companies to take actions that demonstrably
purchases. However, despite achieving a emissions. support
100% global renewable energy match in decarbonization. The even more granular local and hourly accounting of 24/7 CFE
2023 for the seventh consecutive year, our • Bundled energy and EACs: Our approach to
provides a more accurate representation of corporate carbon footprints, and research
reported Scope 2 emissions increased. clean energy procurement to date
has found that targeting 24/7 CFE drives greater impact on grid-level decarbonization
prioritizes purchasing physical electricity
than other procurement goals, such as a 100% annual renewable energy match. 111
If we’re matching 100% of our electricity along with associated “bundled” EACs. 110
consumption with renewable energy, why This helps ensure the additionality of our
Learnings
does the accounting for our Scope 2 clean energy projects (i.e., that they’re
• Our various goals (e.g., 100% renewable energy matching and 24/7 CFE) are divergent
(market-based) emissions show an increase new to the grid). While the GHG Protocol
from the Scope 2 accounting rules, making progress more complex. To simplify our
instead of being zero? It’s because our allows companies to apply “unbundled”
execution, we’ll need to better align across these approaches.
approach to date for EACs to their Scope 2 footprint, including
from existing clean energy projects, our • We’ll continue to evaluate all high-quality options to manage Scope 2 emissions,
100% renewable energy matching differs
“bundled” approach prioritizes new clean including and beyond corporate PPAs, in the future.
from the way Scope 2 emissions are
calculated in accordance with the energy projects, which helps drive the
Greenhouse Gas Protocol (GHG Protocol) clean energy transition but leaves us with
(see Figure 19): some regional gaps.

Key differences Looking ahead to 24/7 CFE

• Global vs. market boundaries: We’ve taken While our approach to date has resulted in
a global approach to our 100% renewable a discrepancy between our Scope 2
energy matching goal. In contrast, the emissions trend and our 100% renewable
GHG Protocol establishes regional market energy match, we believe our shift in focus
boundaries for clean energy sourcing to to 24/7 CFE procurement—matching our
reduce Scope 2 emissions. In some consumption on an hourly basis from
regions, we purchase more clean energy within the same grid—will have
than our electricity consumption (such as a more significant impact on grid
in decarbonization and the clean energy
 Introduct AI for Our Our Appen

FIGURE 19
Our 100%
renewable
energy
matching
and
implicatio
ns for
Scope 2
emissions

Google’s 100% renewable Scope 2 emissions

Key
differences energy match implications

Global Clean energy purchases Clean energy purchases

vs. market are matched globally, must be matched
boundaries which is broader than following GHG Protocol
GHG Protocol regional regional market
market boundaries. boundaries, creating a
gap where
we operated but
didn’t purchase
enough clean
energy in 2023.

Bundled energy Prioritization of Given our

and EACs purchasing bundled prioritization of
energy and EACs, with new-to-the-grid CFE
some market-specific projects, we didn’t
exceptions (see the purchase unbundled EACs
Carbon- free energy to reduce our Scope 2
procurement section for emissions in 2023.
more details).
 Introduct AI for Our Our Appen

Carbon-free energy working with utilities or developers to • Asia Pacific: In Australia, we signed a contract to add 25 MW of clean energy generation capacity, with the solar farm expected to be
buy and deliver carbon-free energy; operational in 2025.
Our primary approach to reducing our structuring
Scope 2 emissions is through the energy supply contracts with energy
procurement of carbon-free energy. 112 In providers through the CFE Manager
2020, we set a goal to run on 24/7 model; and making targeted investments
carbon-free energy—every hour of every in renewable energy to enable
day on every grid where we operate—by additional projects on the grids where
2030. We’re working to achieve this we operate.
through three main initiatives:
purchasing carbon-free energy, From 2010 to 2023, we signed more than
accelerating new and improved 115 agreements to purchase over 14 GW
technologies, and transforming the of clean energy generation capacity 115—
energy system through policy, the equivalent of more than 36 million
partnerships, and advocacy. 113 solar panels. 116 Through these
agreements, we estimate we’ll spend
Carbon-free energy procurement more than $16 billion to purchase clean
energy through 2040. 117
Achieving 24/7 CFE is a far more
complex and technically challenging In 2023, we signed contracts to
pursuit than annually matching our purchase approximately 4 GW of clean
energy use with renewable energy energy generation capacity 118—more
purchases, but we see this effort as than in any prior year—including:
crucial to a bigger picture: scaling new,
global solutions for clean energy. • North America: In Texas, Google
signed a contract for 150 MW of
We buy electricity directly from new clean energy generation capacity.
clean energy projects through various We also signed a contract for new
methods depending on the market, wind, solar, and battery storage in
including: contracting directly via long- Arizona.
term PPAs;
• Europe: In Poland, Google signed
CFE beyond our operations our first contract for 42 MW of
clean
Beyond purchasing CFE for our own operations, we alsoenergy generation
make targeted capacity in CFE development. From 2010 to 2023, we entered into agreements to invest more than $3.1 billion in renewable energy projects
investments
with an expected combined generation capacity of approximately
from the Przyrów4.5 GW.
wind farm.
114 And in
Ireland, we signed our first contract
for 50 MW of clean energy
generation capacity from the Tullabeg
solar farm. In 2023, Google also signed
contracts for new clean energy
generation capacity in Belgium and the
Netherlands.
 Introduct AI for Our Our Appen

In early 2024, we announced a new set of o eration capacity to European grids.

PPAs
— d Our 24/7 carbon-free energy goal
TARGET
i a
n t In 2023, we maintained a global
c e average of approximately 64%
Run on 24/7 carbon-free energy on every grid where we operate by
Year set: 2020; Target year: 2030
l — carbon-free energy across our data
u t centers and offices.
d h
i a Data centers
n t Despite our total electricity load across
g all data centers increasing by roughly 3.5 2023 PROGRESS
w TWh (17%), 64% global average carbon- free energy across our data centers
o i in 2023, we maintained a global average TREND
25
u l of approximately 64% CFE across all of We maintained our CFE percentage for Google’s global portfolio of data centers and offices at 6
r l our data center sites, inclusive of those in the availability of CFE on many of the grids where we operate, progress was also impacted in
operated by third parties. This is due to DETAILS
l both an increase in Contracted CFE 119 The load-weighted average of carbon- free energy percentages 122 across Google’s global portf
b 20 purchased to meet our electricity needs, ever
data centers, and it represents the clean energy
a r (up by roughly 1.2 TWh, or 9%, from every grid where we operate.
r i 2022) as well as improvements in overall

Terawatt-
g n Grid CFE. 120
15
e g 64% CFE
s We’ve worked hard to continue
t 7 advancing CFE in parallel with load 64% CFE
0 growth across our data center
10
o 0 portfolio. In 2023, 10 of our 44 grid
f regions 121 achieved at least 90% CFE
f M (see Figure 20).
s W 5
h Four grid regions—Great Britain, Brazil,
o o France, and Switzerland—reached or
r f surpassed 90% CFE for the first time,
0
e while one grid region— IESO (Canada)—
2022 2023
c fell to slightly below 90%
w l CFE due to accelerated load growth in
i e that region. Google’s CFE percentage in
Legend
n a the ERCOT grid region, which powers our Total electricity consumption across our data centers and
d n Texas data center, nearly doubled from Global average CFE percentage
41% in 2022 to 79% in 2023.
p e
r n
o e
j r
e g
c y
t
s g
e
t n
 Introduct AI for Our Our Appen

FIGURE 20
Global data center carbon-free energy map
Google CFE percentage in every grid region in which we have data center operations, including third-party-operated facilities

In 2023, Google maintained 64% CFE

globally on an hourly basis across our
offices and data centers, and 10 of
Netherlands 80% our 44 grid regions achieved at least
Denmark
90% CFE.
Belgium 82% 92%
98% Finland
Canada
U.S. (Hydro-Québec) % hourly match with
U.S. Canada Great Britain 92% 90% carbon-free energy
(BPA) (MISO) (IESO) Germany
U.S. 100% Ireland 43%
84 31% Poland
U.S. (NVE) U.S. 95% 87% 90%-100%
% 94 Japan
(PACE) (SPP) France 30%
52% U.S. (PJM) % 52% Italy (KEPCO)
26% 80%-89%
29% 87% 92% South Korea 35%
76%
Japan 70%-79%
U.S. (CAISO) 63% U.S. (DUKE) India
55% Switzerland 16% (TEPCO)
Spain 29%
(Northern) 60%-69%

48% 29% U.S. (Santee Cooper)

65% 50%-59%
5% 0% 18%
79% Taiwan
U.S. 42% U.S. (SOCO)
U.S. 40%-49%
(SRP) Israel Qatar
U.S. (TVA) 28%
0%
(ERCOT) 14% 30%-39%

Saudi Arabia Hong Kong 20%-29%

India 4%
(Western) 13% Indonesia 10%-19%
Singapore
0%-9%

90%

Brazil Australia
91% 16% 33%
Australia (NSW)
40%
Chile South Africa (Victoria)
 Introduct AI for Our Our Appen

More broadly, we’re seeing great Grid CFE being 92%, our Consumed CFE for commercial real estate faces corporate agreement to develop a next-
Accelerating new and
progress on CFE in some parts of the Grid CFE was only 21%. For more different challenges, including variability generation geothermal power project in
improved technologies
world: in central North America 123 and details, see the Environmental data in building characteristics, work patterns, Nevada, which became operational in 2023.
Latin America, our average regional tables in the Appendix. and geographic locations with a broad We’re using Google’s engineering
Google CFE 124 across our data centers range of electricity grid carbon expertise and purchasing power to
are both above 90%. However, other While this represents our approach to intensities. Still, we continue to work accelerate the commercialization of a
regions remain challenging for CFE, date, this approach may evolve as we toward advancing CFE across our offices: wide range of CFE technologies,
such as the Asia-Pacific region and continue to enhance our methodology In late 2022, we signed a contract to add working to demonstrate and scale
eastern and data availability improves. For clean energy generation capacity to the their climate impact:
North America, 125 where our average example, we don’t currently grid that powers Google’s offices in New
regional Google CFE across our data • Optimizing generation from existing
use residual grid mix 128 when calculating York City—
centers are 12% and 47%, respectively. CFE technologies: In addition to
grid contributions to Google CFE because our first office-only deal outside of the
using
hourly residual grid mix data does not yet San Francisco Bay Area.
As we continue to enhance our CFE innovative dragonscale solar rooftops,
exist. We’re supporting the development
reporting, we’re newly disclosing our we’ve installed facade solar panel
of time-based tracking methods, which Innovative purchasing models
Contracted CFE 126 and Consumed Grid systems (which are mounted vertically
are essential for making this calculation
CFE 127 percentages on building walls) at our Moffett Park
possible in the future. We’ve pioneered innovative
by grid region, helping to demonstrate Thermal Plant and at our Humboldt
contracting models and industry
the impact of our global CFE office campus. These systems are
Offices partnerships, accelerating the
procurement. For example, in Finland designed to generate power during
For our offices—which represent a adoption of corporate clean energy
in 2023, we achieved 98% Google times of the day when traditional
small portion of our global CFE practices and improving access to
CFE, and over three-quarters of this rooftop solar might be less productive.
percentage—we achieved a global more buyers. In 2023, we advanced
percentage was matched with our And, at wind farms in
average of approximately 56% CFE in the following:
Contracted CFE within the Finnish grid the United States, we’re deploying
2023, up from 54% in 2022. 129
region. This is meaningful because, • New, scalable procurement approaches: machine learning solutions to optimize
despite Finland’s We piloted a new approach to clean wind power forecasting and project
energy requests for proposals (RFPs) economics.
CFE procurement headwinds that can reduce the time to negotiate
• Managing energy demand across our
and execute
In 2023, our efforts to increase CFE procurement faced a number of headwinds, including interconnection a PPA,
delays enabling
(such easy,
as in the PJM grid region), higher development costs and project demand, supply chain issues, U.S. regulatory tariffs for so
global operations: We developed a
transparent, reliable, and efficient
Notably in 2023, the termination of some CFE projects by their owners prior to those projects becoming operational also impacted our data center“carbon-intelligent
CFE progress. Wecomputing
estimate that without these project terminations, our 2023 data center CFE p
contracting in today’s market.
platform” that optimizes the timing
• Contracting principles: We partnered and location of computing tasks
with the Eurelectric 24/7 CFE Hub on based on local grid carbon intensity.
a set of contracting principles that This allows us to shift some of
clean energy buyers and sellers can our computing tasks to different times
use to move toward greater hourly and relocate tasks geographically
energy matching. across data centers so that we can do
• CFE transition tariffs: We worked with more computing in regions where and
the Regulatory Assistance Project to when the grid is cleaner. We’re also
define the optimal design of 24/7 CFE working on demand response at our
transition tariffs and contracts, which data centers.
can make 24/7 CFE more accessible • Accelerating next-generation energy
for electricity users of sources: In 2021, we signed the first
all sizes.
 Introduct AI for Our Our Appen

• Improving clean energy data,

tracking, and trading: Google is
driving the development of
advanced methods for
tracking and validating clean
energy, including time-based
energy attribute certificates (T-
EACs). We’re also working to
improve the quality,
consistency, and granularity of
grid electricity data through
collaborations with
organizations including
EnergyTag, Electricity Maps,
Flexidao,
and Linux Foundation Energy.
And we’ve partnered with
LevelTen Energy and others to
launch the Granular Certificate
Trading Alliance, which aims to
incentivize new clean energy
availability, reduce energy costs,
enable quick transactions, and
help more buyers and sellers
achieve their clean energy
goals.

Google and clean-energy startup Fervo signed the world’s first corpora
 Introduct AI for Our Our Appen

Scope 3 emissions are purchased and not over the life of the Reporting environmental data To support this, we engage our Focusing on emissions hotspots in our value
assets, as is the case for financial suppliers to improve their chain
In 2023, our total Scope 3 emissions were accounting. Primary activity data is essential for environmental data collection and
approximately 10.8 million tCO2e, more accurately calculating our supply accounting, including for their Scope 1, We’re also working directly with
representing 75% of our total carbon Addressing our Scope 3 supply chain chain carbon footprint and developing 2, and 3 emissions. For example, we suppliers of hotspot commodities—or
footprint. Our Scope emissions is particularly challenging due life cycle assessments. We expect all of provide training on reporting to CDP as commodities with disproportionately
3 emissions are indirect emissions from to the geographic diversity of our our suppliers to set public GHG well as high emissions—to identify and
sources in our value chain. The majority suppliers, which span many countries reduction targets and report their on setting GHG reduction and collaborate on carbon reduction
of these emissions are generated from and grid regions. In many regions where environmental data, which helps us renewable electricity targets. In 2023, we initiatives that support our own
the production of goods and services our suppliers operate, they face many of guide our priorities for our supplier hosted summits for both our technical emissions reduction target.
purchased for our operations, including the same challenges we do: insufficient sustainability program. infrastructure and consumer hardware
the upstream manufacturing and CFE capacity and limited device suppliers, where we discussed The semiconductor industry is a
assembly of servers and networking or nonexistent credible clean energy We engage some of our suppliers to Google’s net-zero ambitions, among prime example of an emissions
equipment used in our technical procurement mechanisms. In directly collect data, and also encourage other topics. hotspot due to the electricity and
infrastructure. For details on how we particular, the Asia-Pacific region, some suppliers to respond to CDP’s associated emissions from the
calculate Scope 3 emissions, see the which is a critical region for our Climate Change survey. In 2023, we invited Developing roadmaps to reduce manufacturing and powering of chips
Scope 3 GHG emissions section in the suppliers, is one of the most 312 suppliers to participate, a 40% emissions in electronics devices. We actively
Appendix. Some of our key Scope 3 challenging regions for contracting increase compared to 2022. At least 90% engage in consortia and industry
emissions reduction levers include and investing in carbon-free energy of our hardware suppliers, by spend, In 2023, we worked closely with our organizations to
supplier engagement, minimizing projects. 131 provided data. Of the suppliers that largest hardware manufacturing drive systemic change and support
embodied carbon, and sustainable we invited to respond to the CDP suppliers, by spend, to obtain scalable research and development
travel. The effectiveness of our supplier Climate Change survey in 2023, 60% decarbonization roadmaps, some of within the semiconductor industry (see
engagement strategy and the quality of reported having structured GHG which include GHG emissions reduction Figure 21).
Compared to 2022, our total Scope 3 supplier-specific data we collect can emissions reduction targets, 132 and targets, and we implemented a Google
emissions increased by 8% due to also be impacted by the varying 74% of those targets were science- Renewable Energy Addendum asking Through these organizations and our
increases in emissions generated from maturity of supplier climate programs based. these suppliers to commit to achieving direct supplier engagement, we’re
goods and services purchased for our and the supplier tier we’re working with 100% renewable energy match by collecting primary manufacturing data
operations, upstream emissions from (i.e., whether they’re a direct supplier 2029. 133 that allows us to more accurately model
purchased electricity, and emissions or further down our supply chain). our carbon footprint
related to data center construction. FIGURE 21

We expect our Scope 3 emissions will and networking equipment used in our Supplier engagement
continue to rise in the near term, in part technical infrastructure, and emissions
due to increased capital expenditures from materials used in the construction Imec’s Sustainable
We partner with many suppliers—from manufacturingSemiconductor Technologies
to indirect servicesSEMI’s Semiconductor
suppliers. Alland Systems
suppliers are Climate Consortium
required to sign our
and expected increases in our technical of data centers should be accounted for Google is a founding member of Imec’s program,
Google is an active member of SEMI’s
Supplier Code of Conduct, which states that suppliers should seek to minimize energy consumption and GHG emissions. , an industry grou
Catalyze which seeks to drive innovation needed to decarbonize the semiconductor industry at scale using tran
infrastructure investment to support in the year the assets We evaluate supplier performance in reporting, managing, and reducing their emissions, and incorporate these factors
long-term business growth and In 2023, Google into our supplier
of Catalyze, scorecards and
a decarbonization key procurement
program that aims totools.
accelerate access to
initiatives, particularly those related to renewable energy across the global semiconductor value chain by combining energy purchasing power and enabling supplier participation in renew
AI.

According to the GHG Protocol,

emissions from the upstream
manufacturing and assembly of servers
 Introduct AI for Our Our Appen

Google’s engagement with the

semiconductor industry
 Introduct AI for Our Our Appen

in the hardware supply chain through In 2023, we focused on reducing the reduce GHG emissions across the San Francisco Bay Area produced
life cycle assessments.
Sustainable travel
embodied carbon impact of growing AI technical infrastructure systems that savings of approximately 14,900 tCO2e
demand at our data centers. We support the digital economy. Our transportation team supports emissions—the equivalent of avoiding
Separately, as part of our efforts to
optimized space utilization by fitting sustainable commuting options to help more than 61 million vehicle kilometers
enable 5 GW of new carbon-free
more high- density machines within Consumer devices Googlers get to work—like offering (38 million vehicle miles) or taking, on
energy through
existing and new buildings. shuttles and encouraging carpooling, average, more than 3,500 fuel- based
investments in our key manufacturing We’re using recycled materials in our
Additionally, we’re running data- driven public transit, biking, and walking. cars off the road for a year. 136
regions, in 2023, we continued to invest consumer devices to lower our product
programs to guide carbon-aware fleet
toward this goal as part of a holistic manufacturing carbon footprint. For We’re working to reduce our impact
decisions, such as reusing or upcycling We strive to provide electric vehicle
strategy to increase the availability of example, the aluminum in the enclosure from business travel by exploring ways
technical infrastructure hardware. charging stations for 10% of the total
CFE across our hardware supply chain. of Pixel 5, launched in 2020, was made to make air travel less carbon-intensive
These programs integrate sustainability parking spaces at our San Francisco Bay
into the planning, deployment, and with 100% recycled content, and the Area and supporting the production of
Embodied carbon management of our data center aluminum in the enclosure of all Pixel headquarters, and we continue to work sustainable aviation fuel (SAF) at scale.
machine fleet. phones since Pixel 6 through Pixel 8 and toward this design standard for new
We work to minimize the carbon 8 Pro have also been made with 100% In 2023, Google joined the Avelia
development projects. As of 2023, we’ve
footprint of our data centers, offices, To further advance this work, we’re recycled content, reducing the carbon Sustainable Aviation Fuel program, which
installed more than 6,000 electric vehicle
and consumer working with groups like Building footprint of the aluminum portion of the offers SAF credits to corporate
charging ports at our offices in the United
devices by considering the embodied Transparency to advance the enclosures by over 35% compared to customers. Additionally, in 2024, Google
States and Canada.
carbon of the materials we use. development of tools to measure, 100% primary aluminum. 135 For more joined the United Airlines Ventures
model, and track the embodied carbon details, see the Circular economy Google offers commuter shuttles to Sustainable Flight Fund, a first-of-its-
Buildings and construction of building materials, and the iMasons section. kind effort to provide catalytic
many of its campuses to reduce
Climate Accord, an industry coalition individual vehicle commuting. For 2023, investment to drive SAF production.
We pursue adaptive reuse of existing
working to our shuttle buses in the
buildings to reduce the embodied
carbon from data center and office
construction. When constructing new
office buildings, we work to incorporate
low-carbon materials— like mass timber
—when appropriate. These materials
help reduce our carbon footprint by
storing or “sequestering” carbon within
the building material itself. At our newly
opened YouTube campus in San Bruno,
California,
we used low-carbon materials and mass
timber for the buildings’ structures,
which are estimated to have
approximately 50% fewer embodied
carbon emissions compared to
traditional materials like concrete and
steel, factoring in sequestration. 134
An electric shuttle bus outside of Google’s Bay View campus.
 Introduct AI for Our Our Appen

Managing residual emissions

We’re first focused on • For certainty, we strive to ensure that participate in this nascent market. As scientific studies in areas of carbon purchase of approximately 62,500
decarbonizing our operations projects have the potential positive with many emerging technologies, removals that would benefit from tCO2e of removal credits, which are
and value chain to reach our climate impact they claim to by governments and companies have a additional investigation, ranging from contracted for delivery by 2030 (see
net-zero emissions goal, but as rigorously critical and complementary role to play studying the effects of ocean alkalinity Figure 22).
assessing factors like additionality, in demonstrating promising carbon enhancement
the IPCC stated, “the
leakage, permanence, and verifiability. removal approaches on coastal ecosystems and the We recognize that this is just the
deployment of carbon dioxide and bringing them to a commercial beginning, and we look forward to
potential of enhanced weathering
removal to counterbalance Technology- and nature-based scale. In March 2024, Google projects in forests. accelerating our carbon removal efforts
hard-to-abate residual removals pledged to match the U.S. in the years to come. We’ll continue
emissions is unavoidable if Department of Energy’s Carbon Carbon removal procurement evolving our approach to
net-zero emissions are to be We’re supporting the advancement of Dioxide Removal Purchase program dollar counterbalancing our residual
for dollar: through our own initiatives, we As of the end of 2023, we signed three emissions.
achieved.” 137 both technology- and nature-based
removals toward their highest possible plan carbon credit offtake deals representing a
level of scale and certainty, working to to contract for at least $35 million of total
Our approach to carbon credits
address key challenges carbon removal credits over the next 12
months following the announcement. We
look forward
We aim to neutralize our residual enough to make a difference for the that these solutions face today. Coalition and a champion for their Carbon Dioxide Removal sector.
emissions with high-quality carbon planet. Ideally, some of the best
removal credits by 2030, and to do so in solutions could scale up to at least The main problem with technology- Another pressing challenge is that corporations may currently be reluctant to
a way that maximizes our positive half a gigaton per year of CO2e impact based solutions is that they currently
impact on global decarbonization. This and be available affordably in the lack scale— they’re often too expensive
approach represents an evolution of our foreseeable future. and typically only operate as small
strategy: starting in 2023, we’re no longer pilots.
maintaining operational carbon
neutrality. 138 We’re instead focusing on To help address this problem, in 2022,
accelerating an array of carbon solutions we pledged $200 million to Frontier, an
and partnerships advance market commitment that’s
that will help us work toward our net-zero accelerating the development of carbon
goal, and are aiming to play an removal technologies by guaranteeing
important role in advancing the future demand. We’re excited about
development and deployment of nature- completing our first carbon credit offtake
based and technology-based carbon deals through Frontier in 2023—including
removal solutions required to mitigate deals with Charm Industrial,
climate change. CarbonCapture, and Lithos Carbon—
and about the broader contributions
We prioritize two fundamental criteria the Frontier collective has made for
when considering the climate impact of the field of carbon removals, such as
these efforts: scale and certainty. publishing the buyer’s guide to
enhanced weathering.
• For scale, we consider whether the
solution can become big and affordable We’re also a member of the First Movers
 Introduct AI for Our Our Appen

to working with our partners to identify Contracted carbon

and scale the most promising removals portfolio
FIGURE 22
technology- and nature-based carbon
removal solutions and hope that other
companies will join us. Company Charm Industrial

Project type Biomass carbon

In addition to these partnerships, in removal and storage
2023, Google.org provided a $1 (BiCRS)
million grant to the Integrity Council Credit type Removal
on Voluntary Carbon Markets (ICVCM)
to help them orient the Project location United States

market toward various high-integrity Estimated contracted credits 22,600 tCO2e

solutions with adequate certainty to
Year deal was signed 2023
merit support.
This grant brings Google.org’s Expected timeframe for 2024–2030
delivery
cumulative contributions to
strengthening carbon markets to more Project details Charm Industrial collects
waste biomass that’s left
than $7 million as of the end of 2023—
over from agricultural
supporting organizations including The harvests or forest fire
management, and heats
Gold Standard, Rocky Mountain Institute,
it to a very high
the Voluntary Carbon Market Initiative, temperature in an oxygen-
deprived environment. The
and Climate Action Data Trust.
resulting bio- oil is then
injected into EPA- regulated
Beyond our purchases and wells, where it sinks and
solidifies permanently.
partnerships, Google is uniquely
Market commitment Frontier
positioned to help drive forward
advancements in research and
technology in this area. For example, we Note: At the time of publication,
the following information wasn’t
introduced our Google Carbon Removal
available: registry, project
Research Awards in 2023, which
identification number, project
provided more than $3 million in funding
name, and protocol used to
to universities and academic research
estimate removal benefits. We
institutions for don’t obtain an independent third-
party verification of company
data and claims related to our
contracted carbon removals.
 Introduct AI for Our Our Appen

RESOURCES

Some of our partnerships are helping bring clean energy and its benefits to historically underserved communities through grid-connected solar and home
energy upgrades.

SPOTLIGHT

Supporting environmental justice, climate justice, and a just transition

We’re using our expertise and advocacy to accelerate the transition to a low-carbon economy while also prioritizing equity and
support for vulnerable communities.
Clean energy procurement: We’re exploring how our to decarbonize. In 2023, Google was recognized for this work with a DEI Impact Award from Smart Energy Decisions and a
clean energy procurement can support historically Changemaker Award from RE100. We’ll continue to build on these programs to extend the benefits of our
underserved communities that haven’t typically energy investments to underserved communities.
benefited from the clean energy transition. Two
unique projects have emerged from our efforts to Climate adaptation: During Climate Week NYC in 2023, Google Cloud and Google.org ran a design sprint to
lower barriers for an equitable clean energy explore how geospatial technologies can help vulnerable communities understand and adapt to the impact
transition: of climate change. For more ways we’re supporting climate adaptation efforts, see Adaptation and resilience
in the Our products section.
• Through a first-of-its-kind partnership with EDP
Renewables North America, we’re creating a 500 Funding solutions: In 2022, Google.org launched the Environmental Justice Data Fund (EJDF), a collection of
MW community-based solar portfolio, adding clean grants that aim to help frontline communities that have been historically underserved and disproportionately
energy to the regional U.S. PJM power grid where we impacted by climate change and environmental injustice. The EJDF supports efforts focused on air quality, water
operate. A portion of the portfolio’s revenues will be quality, and environmental hazards. In 2023, G oogle.org expanded its
redirected as utility bill credits
to households facing a high energy burden.
• Through a partnership with Sol Systems, we’re
providing funding to three rural electric
cooperatives and one regional organization in
North Carolina and South Carolina to undertake
critical home pre-weatherization and safety
upgrades for low- and moderate-income
households. The partnership will deliver new solar
energy and battery storage resources, bringing
clean energy to one of the most challenging grids
 Introduct AI for Our Our Appen

fun g y solutions for vulnerable and underserved

din communities in the Asia-Pacific region.
g t For more details, see the Google.org grantees
for o section in the Appendix.
the
EJD l Green jobs skilling and alternative livelihoods: In 2022,
F. o we partnered with ASU to offer an online training
Go c program in sustainability—complementing Google
ogl a Career Certificates in Project Management—
e.o l teaching skills needed for entry-level sustainability
rg jobs. We’ve also supported initiatives that help
als o communities pursue alternative livelihoods as
o r economies transition away from extractive
sup g industries. We sponsored the Congo Power
por a program, which launched community power
ted n projects for people in the Democratic Republic of
the i the Congo and in conservation areas and national
Asi z parks that are of critical importance to the Congolese
an a and the planet. For more details, see our 2023
Ve t Supplier Responsibility Report.
ntu i
re o
Phil n
ant s
hro
py f
Net o
wor c
k’s u
AP s
AC e
Sus d
tai
na o
bilit n
y
See s
d u
Fun s
d, t
whi a
ch i
pro n
vid a
es b
gra i
nt l
fun i
din t
 Introduct AI for Our Our Appen

Water stewardship
We aim to replenish more water than we consume and help improve water quality and ecosystem health in the communities
where we operate

Our approach

Advancing responsible
water use
Data centers

Offices

Supply chain

Benefiting
watersheds and
communities

Supporting water
security with technology
 Introduct AI for Our Our Appen

Our approach to water stewardship

Our water stewardship

strategy is centered on FIGURE 23
How we support our commitment to water stewardship 2023 highlights Global challenges
enhancing responsible water Water stewardship projects: We’ve supported 74 Energy
water stewardship
efficiency vs.projects spanning
local water use: Utilizing
46 watersheds.
water coo
resource management across role in reducing energy use and energy-related emiss
our data centers and offices, resources—particularly
Water replenishment: Our water stewardship projects in regions
replenished an estimated wheregallons
1 billion water resour
of wa
sharing tools and technology Availability
Freshwater withdrawal by water scarcity: 69% of our freshwaterof water stewardship
withdrawals came projects: Water stewar
from watersheds wit
that can help address water to work to identify and develop opportunities across g
challenges, and improving Water risk framework: In 2023, we shared our water risk framework, which describes our data-driven
Navigating the local context of stewardship projects: Th
watershed health and challenges that must be assessed and navigated acro
ecosystems in water-stressed
communities (see Figure 23).
Advance responsible water use at Google
In 2021, we announced our goal to
Enhance our stewardship of water resources across Google offices and data centers
replenish 120% of the freshwater 139
volume we consume, on average, across
our offices and data centers by 2030,
and help restore and improve the quality
of water and health of ecosystems in the
communities where we operate.

To achieve our water replenishment Benefit watersheds and communities

target, we’re channeling our ongoing Collaborate to replenish our freshwater consumption and improve watershed health while supporting ecosystems and water-stressed communities
investments into projects located within
the watersheds where we operate. We
work with external partners
to implement these projects, delivering
volumetric water benefits and
improving other locally relevant aspects
of watershed health such as water
quality, community water access, and
Support water security with technology
biodiversity. Share technology and tools that enable everyone to predict, prevent, and recover from water stress
 Introduct AI for Our Our Appen

Advancing responsible water use

We prioritize responsible water Water plays an important role in our consumption and related carbon emissions when compared to air-based cooling. While it will take more time for electricity grids
use in our data centers and our data centers—cooling our servers, to decarbonize, we’ll continue using water cooling to improve our energy efficiency in certain geographies. Recognizing that this tradeoff will
regulating indoor temperatures, and
office operations around the increase our data center water footprint, we’re prioritizing responsible water use and water replenishment at new sites from the start.
keeping our products up and running. In
world. fact, water cooling has been shown to
Water risk framework
help reduce energy
We’re accelerating water reuse
practices across our offices and data In 2023, we shared our water risk framework, which describes our data-driven approach to responsible water use in our Google-owned data
centers, and tailoring site-specific centers.
solutions based on facility types,
locations, and local water contexts. We Building on our climate-conscious approach to data center cooling, this framework provides an actionable and repeatable process for
also use non-potable sources and evaluating local watershed
freshwater alternatives whenever health and water risks across our portfolio, focusing on water scarcity and the risk of a freshwater source being depleted. The
feasible. evaluation results help us determine when we should consider using air cooling technologies instead of water, and when we should consider
alternative water sources—such as reclaimed wastewater, industrial water, or seawater.
In 2023, the total water consumption at
our data centers and offices was 6.4 We apply this framework to every new data center site (including new sites before acquisition and future developments on
billion gallons (approximately 24 billion existing campuses) and aim to repeat these assessments across our existing Google- owned data center portfolio every three to five
liters or 24 million cubic meters)—the years to evaluate water risks that may require mitigation.
equivalent of what it
takes to irrigate 43 golf courses annually,
on average, in the southwestern United
States. 142 This represents a 14% increase
from 2022, primarily due to water
cooling needs at our data centers, which
experienced increased electricity
consumption year-over-year.

In 2023, 69% of our freshwater

withdrawals came from watersheds
with low water scarcity, 16% came
from watersheds with medium water
scarcity, and 15% came from
watersheds with high water scarcity.
143
Through our water replenishment
program, we focus on higher
volumetric water benefits in water-
scarce regions to ensure we’re
positively impacting the areas that
need it most.

Data centers
 Introduct AI for Our Our Appen

T t gle’s data centers and offices, which

h e provides us with a more holistic view of
e r water challenges and trends—see more
s p details in the Risk management section in
e r the Appendix.
i
a s This framework is already informing
s e our cooling technology selection. For
s - example, our new data center
e w campuses under
s i development in Mesa, Arizona, and
s d Canelones, Uruguay, will use air-cooling
m e technology because the source
e watersheds didn’t meet our responsible
n w use threshold for water cooling. And we
t a already employ air-cooling technology at
s t our data centers in Storey County,
e Nevada, and Dublin, Ireland.
a r
r Alternative water sources
e r
i In 2023, 22% of our total data center
i s water withdrawal (excluding seawater) Steam rises above the cooling towers at one of our existing data centers in The Dalles, Oregon

n k was reclaimed wastewater and other

non-potable water, and roughly one-
a a third of our data center campuses used SPOTLIGHT
d s air cooling or non- potable water
d s sources. Innovative water system at The Dalles
i e In 2006, Google opened our in The Dalles, Oregon.
t s We use reclaimed water at our data
i s centers in Singapore and in Douglas We’re now working on a new campus and available
its supporting
duringwater
the rainy
and sewer
season,
infrastructure.
which wouldAotherw
vital
o m County, Georgia, where the will pump excess surface water into an existing
its untreated
aquifer groundwater
for later use during
usage.drier months—funct
n e wastewater would otherwise be “savings account.” The surplus water
n discharged in the
t t Chattahoochee River. Our data centers in
o St. Ghislain, Belgium, Changhua County,
o Taiwan, and Eemshaven, Netherlands, use
a f industrial water. In Hamina, Finland, we
n use seawater to cool our data center, and
a then return it to the sea at close to its
a l original temperature.
n l
n Water efficiency in Singapore
u o In 2024, Google received a from PUB, Singapore’s National Water Agency, which recognizes
a f
l
G
e o
n o
 Introduct AI for Our Our Appen

Offices Our building design requirements for A central plant treats stormwater water per year, roughly 30%, compared
Supply chain
new construction include the gathered from retention ponds and with a building-by-building approach to
In 2023, Google had offices in nearly incorporation of water-efficient fixtures, wastewater collected from buildings, heating and cooling, per our predictive
60 countries around the world. To such as faucets, toilets, and irrigation producing recycled water that can be Water is also used in our supply chain in the
models.
manage our global environmental systems. For existing buildings, our used for cooling towers and irrigation. manufacturing of our consumer hardware
footprint, we’re design standards also include replacing devices and technical infrastructure
We continue to work toward responsible
developing responsible water use old fixtures with highly efficient ones in As of 2023, Bay View is still on track to be hardware. Addressing water challenges
water use in the water-stressed regions
practices that can be applied in every any space we move into and installing the largest development project in the throughout our supply chain requires high-
where we have offices, like the San
region where we operate and for the water meters with automatic leak world to achieve Water Petal certification quality water data. That’s why we ask
Francisco Bay Area. Our Gradient Canopy
many ways we use water in our offices, detection. from the LBC. Bay View’s adjacency to suppliers to disclose water- related data
building in Mountain View, California,
including for preparing food, cleaning and the San Francisco Bay makes water an via the CDP supply chain platform.
uses municipal and onsite- generated
sanitation, irrigating campus landscapes, important focus and we hope that others recycled water for toilets, irrigation, and
Water infrastructure innovation This reporting includes data on water
and keeping our workspaces cool. can learn from these innovative solutions. cooling, and our new YouTube offices
withdrawal from all sources, the portion
Beyond these standards, we drive water in San Bruno are expected to capture up
Sustainability design standards Another example is our all-electric of water permanently lost in withdrawal,
stewardship in our offices. to roughly 1.3 million gallons of
Moffett Park Thermal Plant in Sunnyvale, and water effluents discharged, among
stormwater per year, which will be treated
One of the main ways we conserve California, which opened in 2023. By other metrics. In 2023, we received an
A flagship example of this work is our Bay to meet on-site, non-potable water
water in Google offices is by adopting centralizing cooling towers and using 87% response rate (108 out of 124) from
View office in Mountain View, California, demands. The stormwater capture
sustainability design standards aligned heat recovery, the plant is expected to suppliers that we invited to participate in
which has a treatment system that’s system is expected to help reduce our
with leading save up to 2 million gallons of the Water Security survey via the CDP
designed to capture and reuse water on- municipal water use by approximately
third-party certifications, including Supply Chain program.
site via stormwater retention ponds and 18%. 144
the Living Building Challenge (LBC). constructed wetlands.
 Introduct AI for Our Our Appen

Our Bay View campus, as seen from across its stormwater retention pond. (Photo: Iwan Baan)
 Introduct AI for Our Our Appen

Benefiting watersheds and communities

Improving watershed and Our process for working toward our

ecosystem health is a critical water replenishment target includes:
part of our water stewardship TARGET
• Conducting water risk
strategy, and it’s integrated
assessments to identify priority
with our water replenishment
watersheds for water Replenish 120% of the freshwater volume we consume, on average,
goal. stewardship projects.
across our offices and data centers by 2030
We designed our strategy to prioritize a • Assessing local water challenges
Year set: 2021; Target year: 2030
range of watershed health issues, through data analysis and local
enabling us to invest in a diverse set of stakeholder engagement to guide us
impactful projects and partnerships that toward context- specific solutions.
are appropriate for each
local context. • Engaging project partners who are
2023 PROGRESS 6,000
working on the ground to identify
In addition to pursuing replenishment existing replenishment opportunities 1 billion gallons of water
projects with volumetric benefits to water that address shared water replenished, or 18% of our
systems, we also identify watershed challenges.
2023 freshwater
health projects with the potential to • Co-developing water replenishment
provide co-benefits for nature, consumption 4,000
opportunities with local partners if
biodiversity, and communities. none are readily available.
TREND
As of the end of 2023, we’ve supported Once we decide to move forward with We estimate that our water stewardship

Million
74 water stewardship projects spanning 46 a water stewardship project, projects replenished approximately 1
watersheds (see Figure 24). These 2,000
implementation can take up to a few billion gallons of water 147 in 2023 alone,
projects focus on land conservation and
years to complete, depending on the increasing our freshwater replenishment
restoration, water supply reliability,
activities. Projects often involve extensive from 6% in 2022 to 18% in 2023. This is the
ecosystem restoration, water quality
planning, community engagement, and result of adding 36 new
improvements, and increasing water,
permitting, in addition to the actual water stewardship projects to our
sanitation, and hygiene (WASH) access.
restoration or construction activities. replenishment portfolio—nearly doubling the 0

In 2023, these projects replenished an

After projects are operational, we use the portfolio size from 38 projects in 2022 to 74 18%
Volumetric Water Benefit Accounting projects in 2023. Legend
estimated 1 billion gallons of water 145
(VWBA) methodology to determine the
(approximately 3.9 billion liters or 3.9
estimated eligible replenishment benefits DETAILS 6%
million cubic meters). This figure Water replenished annually as a
Google can claim. For more details, see
represents 18% of our 2023 freshwater percentage of the amount of freshwater
the Water replenishment section in the 2022 2023
consumption for our data centers and we consume each year at our offices
Appendix.
offices. Once fully implemented, we and data centers (i.e., excluding
estimate that these 74 projects will have In 2023, we co-funded the seawater and reclaimed wastewater). 148
the capacity to replenish more than 2.8 development of VWBA 2.0, the next Replenishment benefits are counted
billion gallons of water annually. 146 iteration of water from projects that are active within
the watersheds that our operations rely on
replenishment quantification and that have confirmed volumetric benefits
methodologies.
 Introduct AI for Our Our Appen

from the reporting year. Total freshwater consumption for offices

and Google-owned and -operated data
centers

Estimated water replenished

Capacity of all contracted water

replenishment projects once fully
implemented
 Introduct AI for Our Our Appen

FIGURE 24
Global water replenishment project map
This map shows all 74 water stewardship projects we’ve supported as of the end of 2023.

North America
Europe, Middle East,
and Africa

Project category Total

Land conservation and restoration 14

Water supply reliability 24

Asia Pacific
Water, sanitation, and hygiene 8
(WASH)

Aquatic habitat restoration 18

Water quality 10

Total projects 74 Latin America

 Introduct AI for Our Our Appen

Supporting water security with

technology
Multi-benefit floodplain restoration along the Tuolumne and San Joaquin rivers in We’re using Google’s For example, we partnered to develop
California.
the Freshwater Ecosystems Explorer, a
products and technology to
free,
help others study and
SPOTLIGHT
easy-to-use geospatial platform that
respond to water security provides national, sub-national, and
challenges. basin-level data on freshwater
Water stewardship projects ecosystems. And in collaboration with
Below are some examples of impactful projects and partnerships from the 36 new water Global searches for “water security” OpenET, as well as government and
reached an all-time high in 2023,
stewardship projects we supported in 2023. academic research groups, we
demonstrating a growing awareness of supported the automation and scaling of
this critical issue. 149 evapotranspiration (ET) models to help
improve water
longleaf pine forest. This an ASR system
Arizona, United States: To further support initiative employed several tactics, burns. We provided critical funding on the island of Texel, and using smart with the Province of Groningen and other
water security in Mesa, Arizona, we including removing invasive species, to The Nature Conservancy and technology to prevent saltwater intrusion partners to develop water control structures to
contributed to the Salt River Project’s planting a diverse understory, and farmers in by monitoring water salinity and efficiently increase flood protections for salt marsh
efforts to restore the local watershed conducting prescribed the Upper Colorado and Upper Trinity routing freshwater. We’re also working habitats and aquacultures near our data center
and reduce wildfire risk. We partnered River Basins to implement —which will also improve local
with the Colorado River Indian Tribes Environmental Water Transfers that water quality.
(CRIT) and N-Drip to fund conversion of increased streamflow and developed India: Building on our portfolio of stewardship
flood irrigation to precision drip regenerative agricultural practices on projects in India, we partnered with CLEAN
irrigation, and to support farmers to 1,000 acres, aiming to reduce runoff International on lake restoration projects in Pune
enable water-saving crop rotations. and improve water quality. and Bengaluru that, combined, will restore 18
We
Chile: We expanded our partnership acres of lake and plant approximately 1,000
also partnered with the Quechan Tribe with Kilimo and are installing precision waterfront trees. We also expanded our
on a farmland initiative whose water irrigation management systems that will partnership with WaterAid by providing clean
savings will bolster efforts to improve use AI on approximately 1,980 acres water access in Hyderabad.
water security in the Colorado River (800 hectares) of farmland in Chile’s
Basin. Japan: We’re supporting Ramsar Network Japan
Maipo Basin to conserve water used for to create and improve wetland habitat in and
Illinois, United States: In the Chicago irrigated agriculture. around the Watarase Retarding Basin, which
River’s Wild Mile, we’re supporting Spain: We partnered with Fundación serves as critical habitat for the endangered
the creation of a freshwater mussel Naturaleza y Hombre to create Oriental Stork.
habitat to improve water quality and watering ponds and clear vegetation in
provide other important ecological Taiwan: We’re working with CLEAN International
the Sierra de Gata Biological Reserve. on installing rainwater harvesting systems and
benefits. We’ll also support green This effort will help improve forest
infrastructure projects at two sites storage tanks at schools in Taipei City, New
health and reduce fire risk in the Taipei City, and Changhua County. The schools
to improve stormwater watershed surrounding our Madrid
management in Chicago. will use water from the rainwater harvesting
office campus. systems for cleaning, landscape maintenance,
Texas, United States: We partnered Netherlands: We’re working with sanitation, and cooking.
with Texan by Nature on a watershed
Acacia Water on two projects: The Global Water Watch tool
health initiative to improve the
increasing the efficiency and capacity of
stewardship of an old-growth
 Introduct AI for Our Our Appen

We work to put environmental management.

information about the world’s forests,
fisheries, watersheds, and air into the At World Water Week 2023, we
hands of decision- makers, in support of highlighted the launch of Global
our aims to both address a key need for Water Watch, a new tool featuring
informed decision-making, near-real-time reservoir storage
and to support efforts to scale data that Google.org supported
models and solutions that make through the 2020 Google.org
water resource information visible Impact Challenge on Climate.
and actionable.

RESOURCES

G
Introduct AI for Our Our Appen

Circular economy
We aim to maximize the reuse of finite resources across our operations, products, and supply chains

Our approach

Building circular workplaces

Circular design and construction

Waste diversion

Reducing data center waste

Designing more sustainable

consumer hardware
devices
Recycled materials

Sustainable packaging

Product longevity

Working with suppliers

 Introduct AI for Our Our Appen

Our approach to the circular economy

We believe the world

FIGURE 25 How we’re working toward becoming a more 2023 highlights Global challenges
must accelerate the
circular Google
circular economy in order Waste diversion: We diverted 78% of our global data centers and
Infrastructure: Recovering
offices’ operational
and diverting
waste
resources
away from
fromdispos
dispo
to achieve net-zero
carbon emissions. Food waste reduction: We reduced food waste per Googler
Markets: Enabling
by 38%circularity
compared requires
to 2019,
supportive
151 building
fiscal
on anpolicy,
18%

Resold components: Since 2015, we’ve resold more than

Metrics 44 measurement:
and million hardware
Achieving
components
a circular
fromeconomy
our data will
cen
We set out to become a circular
measurement, metrics, and data to enable more efficient us
Google by maximizing the reuse Pixel software support: We announced our commitment to providing for Pixel 8, Pixel 8 Pro, and future Pixel
of finite resources across our Single-use plastics: Accelerating the switch from using single-
operations, products, and supply the economy. Additionally, transitioning away from plastic pa
chains—and enabling others to do
the same. Global searches for We design out waste and pollution
“circular economy” reached an all- We design for circularity from the start, making it easier for
time high in 2023, underscoring the materials used in the creation of our products, offices,
the materials used in the creation of our products, offices,
the growing awareness and and data centers to be reused in the future
and data centers to be reused in the future
demand for a shift toward a more
sustainable economic model. 150

Our circularity principles focus

on designing out waste and
pollution from the start, keeping We keep products and materials in use
products and We pursue strategies to extend the life of our products,
materials in use, and promoting offices, and data centers for as long as safely possible to maximize the use of materials that went into their creation
healthy materials and safer
chemistry (see Figure 25).

To this end, we’re pursuing

innovative strategies—from
adaptive reuse building projects
to recycled materials in our
We promote healthy materials and safer chemistry
consumer hardware devices. Our
impact on the circular economy
We design products, offices, and data centers with safer
spans our business operations We design products, offices, and data centers with safer
and materials for people and for the environment, enabling
them to be reused in the future
value chain—including our data
centers, offices, stores, products,
and suppliers.
 Introduct AI for Our Our Appen

Building circular workplaces

When we design, build, and world’s largest renovation typology We use the Rheaply platform, which helps locate and reuse office furniture and other materials to support real estate circularity. In
operate Google data centers, project to have attained the LBC 2023, Rheaply helped Google reuse more than 40,000 pieces of furniture and equipment internally, contributing to over
Materials Petal Certification.
offices, and stores around the 317,000 kilograms (700,000 pounds) of waste diverted from disposal, including landfill and incineration. We also sponsored the launch
world, we strive to deliver on of public Rheaply marketplaces in the San Francisco Bay Area, New York City, and Chicago to bring this solution to more users in those
cities.
our commitment to
accelerating the circular
economy. Waste diversion
We strive to divert solid waste from disposal (defined as diversion of waste from landfills or incinerators) and aim to minimize the amount
Circular design and of materials we use and maximize their lifespan within our ecosystem and the surrounding community.
construction In 2023, the global waste diversion rate for our offices was 77%. 154 We prioritize diverting
When the right opportunity comes up, we high-quality materials to maximize the success
pursue adaptive reuse projects,
renovating existing buildings to serve a
new purpose rather than demolishing
them for new builds. These reuse projects
tend to use fewer materials, produce less
construction waste, and have lower
embodied carbon than new construction
projects. Adaptive reuse projects also
help preserve community history, giving
historic structures new life as Google
offices and, in some cases, as community
spaces too.

For more than a decade, we’ve Our Hyperlink Bridge in Dublin, Ireland, connects three TRUE-certified buildings— Google Docks, Gordon House, and The Gasworks.

prioritized building materials that

demonstrate
safer chemistry to create healthy indoor
environments. One way we do this is by
leveraging third-party standards that
encourage transparency in the building
industry, like the Living Building
Challenge (LBC). As of the end of 2023,
Google’s Gradient Canopy building in
Mountain View, California, was the
largest new construction project to
attain the LBC Materials Petal
Certification. And as of the end of 2023,
our 237 Moffett Park Drive building in
Sunnyvale, California, remained the
 Introduct AI for Our Our Appen

o a treams, and remove contaminated

f m waste from our calculations.
i This effort contrasts with the typical
o n approach to calculating diversion rates,
u a which makes the assumption that
r t anything placed in recycling or compost
i bins has zero contamination. A view of Gradient Canopy’s interior canopy with clerestory windows.
w o
a n Our office circularity efforts extend
s beyond waste diversion when possible. In
SPOTLIGHT
t r 2023, our main campus in Dublin, Ireland
e a achieved TRUE Zero Waste Platinum
t certification, which recognizes the
Circular design and construction at Gradient Canopy
integration of circularity practices into
Our Gradient Canopy building in Mountain View, California is a flagship example
r e
e s building operations— including
c procurement, infrastructure, and
o o employee engagement. For details on
v f our TRUE precertifications, see the
e Certifications section in the Appendix.
r o Here are a few examples from the launch of impact
the building
humaninand
2023:
environmental health. We also en
y u Additionally, across our office portfolio As an example, many of the wooden doors and frame
r in 2023, we made efforts to refurbish Construction waste: At Gradient Canopy, weReclaimed
diverted materials:
more than Throughout Gradient Canopy,
90% of construction we
waste from
and recover computing devices and Healthy materials: Materials installed at Gradient Canopy were reviewed with manufacturers to e
e
they were free of LBC’s ingredients, which represent “worst-
f r other corporate electronics, diverting in-class” chemicals that negatively
f e nearly 90% of material from disposal
o c via reuse, donation, and recycling.
r y
t c
s l
. i
n
W g
e
a
a n
s d
s
e c
s o
s m
p
t o
h s
e t
i
c n
o g
n
t s
 Introduct AI for Our Our Appen

Food waste base year. We’re seeing measurable To further reduce food waste, we’re • Kitchen and cafe operations: We’re donated or properly composted.
progress toward this goal by using strengthening our efforts in three key reducing food waste across our back-of- We’re expanding partnerships with
We’re taking a holistic approach to different intervention levers to drive areas across our food program: house operations through improved municipal and community
address food waste, focused on the operational change (such as just-in-time culinary practices and demand organizations to develop composting
“three R’s”— reduce, reuse, and recycle. cooking) and spark behavior change • Sourcing and procurement: We’re working planning to ensure our food infrastructure and, in markets with
Reducing food waste generated in our (such as communications campaigns). In with our suppliers, distributors, and production levels are consistent with limited or emerging composting
kitchens is a key lever, as it decreases 2023, we reduced food waste per vendors to prevent food waste before user dining patterns. infrastructure, we’re exploring ways
the food waste that must be diverted Googler by approximately 38% compared it happens. We buy from innovative to process our food waste on-site.
• Excess food: We continue to install new
from landfills. That’s why in 2022, to 2019, 155 building on an 18% reduction businesses that make food products
waste containers and signage to help
we set a goal to cut food waste in half in 2022. We also have an ambitious goal from upcycled and imperfect Single-use plastics
separate food waste effectively, and
for each Googler by 2025, compared to to send zero food waste to the landfill by ingredients.
we’re working to ensure excess food
a 2019 2025. In 2023, we diverted 82% of food Global search interest in “plastic
is either
waste from landfill. pollution” reached an all-time high in
2023, underscoring the urgency of finding
solutions to this environmental challenge.
TARGET 156
We’re working with vendors and
suppliers to phase out products with
Divert all food waste from landfill by 2025 single-use plastics. These changes
Year set: 2022; Target year: 2025 involve rethinking how we handle, store,
prepare, and serve food and beverages.

We’re piloting and scaling plastic-free

2023 PROGRESS Legend
processes and products in our kitchens
Percentage of food waste diverted from landfill
82% of food waste diverted from landfill and we’re partnering with distributors
TREND 100 that are using more reusable and
Our food waste diversion rate decreased from 85% in 2022 to 82% in 2023 due, in part, to limited composting infrastructure in certain regions—including Asia Pacific—which continues to durable
impact our ability toto
containers divert food waste
transport goods.from
In our
75
kitchen and cafe operations. April 2023, we launched our Single-Use
50 Plastics Challenge, giving food companies
85%
25 that don’t use single-use plastics in
Percent

their packaging the opportunity to test

0
20222023
their products in Google’s U.S. campus
cafes and kitchens. In December 2023,
we announced the nine finalist
companies,
with solutions ranging from reusable
snack bags to edible cutlery, refillable
commercial cooking oil vessels, data-
DETAILS
powered
Annual food waste diverted from landfills and incinerators in kitchens and cafes at Google’s offices globally. We consider sending “zero food waste to landfill” as 99% diversion of food waste from bulk foodand
landfills dispensers, and
incinerators via composting, a
processing—which goes further than industry standards for Zero Waste to Landfill. more.

In our kitchen and cafe operations, we’re focused on ways to stop food waste before it starts.
 Introduct AI for Our Our Appen

Reducing data center waste

We design, build, and operate to Landfill. We integrated new data partners, and vendors to improve our
our data centers to maximize sources and assumptions to account waste data collection processes and
efficient use of resources and for parts and materials that enter our systems. We’re committed to
reverse logistics recycling value chain continuous improvement of our waste TARGET
materials. This includes
—inclusive of e-waste, racking management practices, particularly
exploring adaptive reuse
infrastructure, and packaging waste. We at our data centers, with the ultimate
projects for some of our data also refined our approach to diversion goal of becoming a more circular Google.
Achieve Zero Waste to Landfill for our global data center operations
centers (see Figure 26), and Year set: 2016; Target year: N/A
accounting for waste that’s thermally
focusing on data center processed (i.e., incinerated) when it For data center equipment, our
waste. leaves our data centers—reclassifying approach to circularity aims to
it as disposed instead of diverted, maintain servers for as long as
In 2016, we announced our aim to even when energy is recovered. possible by refurbishing, reusing, or
achieve Zero Waste to Landfill for our reselling components (following a 2023 PROGRESS
global data center operations. In 2023, following our updated waste rigorous security process)—and 29% of Google-owned and
Working to address waste at our data accounting methodology, we diverted recycling any -operated data center campuses achieved Zero Waste to Landfil
centers will require collaboration, 78% of operational waste from disposal components that can’t be reused. Since TREND
innovation, and high-quality data to across our global fleet of Google-owned 2015, we’ve resold more than 44 million Due to changes in methodology that have only been applied to data center waste data for 2023,
accurately measure performance and and -operated data centers, and 29% (8 hardware components from our data DETAILS
identify opportunities. out of 28) of our Google-owned and - centers into Annual operational waste for all Google- owned and -operated data center campuses globally. W
operated data center campuses met our the secondary market for reuse by other energy recovery, as waste disposal. 29%
Updating our approach to zero waste Zero Waste to Landfill goal. organizations, 157 including more than 7
In 2023, we refined our approach to million resold components in 2023 alone.
data center waste accounting to As we further refine our waste As of
8 out of 28 data center campus
better align governance, we’re working with our data the end of 2023, 29% of components
with circular economy principles and center site teams, used for server deployment,
more accurately track progress toward maintenance, and upgrades were
Zero Waste refurbished inventory.

FIGURE 26
Adaptive reuse examples for our data centers
Legend

Data center campuses that have achieved our Zero Wa

Data center campuses yet to achieve our Zero Waste to

The Dalles, Oregon The Dalles on a former U.S. Superfund site that previously housed an aluminum smelter. The development will also leverage an existing drainage channel for
In 2023, we started construction on a third campus in stormwater management.
 Introduct AI for Our Our Appen

Hamina, Finland na, Finland, into our second data center in Europe. In
W addition to the buildings, we also repurposed the existing
e electrical substation and a seawater cooling tunnel, which
we use to cool the data center.
c
o
n
v
e
r
t
e
d

d
e
c
o
m
m
i
s
s
i
o
n
e
d

p
a
p
e
r

m
i
l
l

i
n

H
a
m
i
 Introduct AI for Our Our Appen

Designing more sustainable consumer hardware devices

Since launching our first establish robust sustainability criteria We also set a target to use recycled or
consumer hardware devices, 158 for electronic devices. For more renewable material in at least 50% of
we’ve set information about our products, plastic used across our consumer TARGET
out to integrate sustainability including their recycled material hardware product portfolio by 2025,
content and packaging, see our Product prioritizing recycled plastic where we
considerations into every
Environmental Reports. can. 34% of the plastic Google used in Use recycled or renewable material in at least 50% of plastic used ac
aspect of our operations— products manufactured in 2023 Year set: 2020; Target year: 2025
from how we source our
materials, engineer and Recycled materials was recycled content. 163
represents a decrease in overall
This

package our products, run We’d set a goal to include recycled recycled content of plastics across
our materials in 100% of Google consumer our portfolio from 41% in 2022, 164
supply chain operations, and hardware products launching in 2022 and which was due to changes in our
every year after. 160 We first achieved this product mix—some product types
design our retail stores. 2023 PROGRESS Legend
goal in 2020 for our Nest, Pixel, and use Percentage of recycled or renewable material in plastic used in Google products m
Chromecast devices, and have less plastic than others, which can 34% of the plastic Google used in products
We aim to decrease our use of virgin
materials through our procurement of
maintained it for these products reduce opportunities to use recycled manufactured was recycled content 166
recycled materials and extend the life
launched each year since—through content. At least 20% of the material TREND 50
2023. 161 Google used in our new products The percentage of plastic used in our manufactured consumer hardware products that was recyc
of our products through software 40
launched and manufactured in 2023 was
updates and expanded repair options
Additionally, Fitbit devices launched in recycled content. 165 This effort includes 30
(see Figure 27).
2023 included recycled materials—the recycled material used in our devices,
20
We also support electronics standards first year Fitbit devices were included in such as aluminum, stainless steel, rare-

Percent
and certifications, including UL 110, IEEE this goal. 162 earth magnets, glass, and plastic parts. 10

1680.1, and the UL ECOLOGO 0

Program, 159 that
2021 2022 2023

DETAILS
Includes the minimum percentage of recycled or renewable plas
bio-based material. This target doesn’t include third-
party products such as the Nest x Yale Lock.
 Introduct AI for Our Our Appen

FIGURE 27 Our consumer hardware devices are built with sustainability in mind

Recycled materials Packaging Product longevity Low-carbon design E-waste

with recycled metals,
Focusedglass,
on and plastics,
using fiber- and now
Designed
based recycled
materials
to be andtin,
repairable as
makingwell
and used
our packaging
for as longmore easily Using
as possible recycled
recyclable
with feature aluminum
Offering
drops, programs
and OSin the
andenclosure
to recycle
security of Pixel
and trade-in old devices to help reduce e-waste
updates
phones to reduce the carbon
footprint, compared to using primary aluminum
 Introduct AI for Our Our Appen

operating systems, enhancing product

Sustainable packaging continue to work with suppliers to
Product longevity longevity, and reducing the amount of
For eligible devices from any brand at the
create plastic-free packaging end of their useful life, we offer free recycling
Our goal is to eliminate plastic from our solutions that are protective and potential devices that may end up as e- in every country where we ship consumer
We want our devices to be used and
hardware product packaging by 2025. aesthetic. waste. hardware devices. 175
valued for as long as possible, which we
We’re working to achieve this target by
work to achieve by guiding consumers
designing Pixel, Nest, Chromecast, and When comparing the packaging of our
Fitbit packaging to minimize the use of first Pixel phone to Pixel 8 and Pixel 8 Pro,
through the care, repair, reuse, and Working with suppliers
recycling of their products.
plastic. For new Google products we’ve made significant progress— Manufacturing waste
launched and manufactured in 2023, our eliminating plastic from Pixel 8 and Pixel Enabling security updates and bug
packaging was at least 99% plastic-free, 8 Pro packaging, and reducing packaging fixes helps promote product longevity. In 2023, we achieved UL 2799 validation for
169
and packaging for Pixel 8 and Pixel 8 weight and volume by In 2023, we announced our waste diversion at all final assembly
Pro uses 100% plastic-free materials. 170 at least 50%. 171 Smaller and lighter commitment to providing seven years manufacturing sites for our consumer
By focusing Pixel 8 Pro packaging means more of software support for Pixel 8 and Pixel hardware products. 176 Of these sites, 75%
on fiber-based materials, we’re also devices can be transported in a 8 Pro, including Android operating achieved either Zero Waste to Landfill
making our packaging more easily single shipping container and system upgrades and security updates. Gold
recyclable. We’ll therefore more natural resources can 174
And, for Google Nest products, we classification (representing a 95%–99% landfill
be conserved. provide critical bug fixes and patches for diversion rate) or Zero Waste to Landfill
at least five years after launch. Platinum classification (representing a 100%
TARGET landfill diversion rate). We’ll continue pursuing
We’re working to empower everyone UL 2799 validation for waste diversion at
Make product packaging 100% plastic-free by 2025 with more repair options to extend the our consumer hardware final assembly
Year set: 2020; Target year: 2025 useful lifespan of consumer hardware manufacturing sites, including for new sites as
technology because we believe that they’re added in the future.
Legend consumers should have the “Right to
2023 PROGRESS products launched andaccess
Percentage of product packaging that was plastic-free for new GoogleRepair”—including to theeach
manufactured same
year
Safer chemistry
For new Google products100launched and manufactured in 2023, our packaging
documentation, was
parts, and toolsat
thatleast 99% plastic-free 172
original equipment manufacturer (OEM) We’re working to eliminate antimicrobials and
75 repair channels have. Pixel repair use safer flame retardants and solvents
50
options are available in all countries across our consumer hardware products.
where we sell Pixel phones. We also offer We’re collaborating across the industry to
25 customer support, self- repair options, raise awareness of safer flame retardant
0 and an online community help center. In options available to the electronics sector.
202120222023 tandem with our trade-in program, these We’ve partnered with ChemFORWARD, a
efforts can extend the useful life of trusted nonprofit with chemical hazard data,
Pixel phones. on an approach to assess chemical safety
that goes above existing regulatory and
TRENDDETAILS In 2023, we announced that all
The increase from at least 96%B ased on the total weight of new Google Pixel, Nest, plastic-free packaging in 2022 toChromecast, and Fitbit retail packaging globally 99% in 2023 was industry
primarilynorms. We’ve performed
due to(excluding safer
adhesive materials and re
in our packaging material catalog. 173 Chromebook platforms will get regular chemistry evaluations for materials used in
automatic updates for 10 years. our products, and
ChromeOS Flex continues to offer an we’re expanding this work to drive industry
alternative for extending the useful life progress beyond the standard management
of aging devices by modernizing their of regulated chemicals.
 Introduct AI for Our Our Appen

Through our consumer hardware

product Restricted Substances
Specification and Manufacturer
Restricted Substances List, we
restrict many hazardous
substances and ensure our
suppliers have processes in
place to detect and prevent
them from entering the
manufacturing process. In 2023,
we
introduced new hazard-based
requirements, including
mandatory hazard ratings and
restrictions for certain solvents
and flame retardants. We also
maintain a Responsible Chemical
Management program, which
includes assessments, guidance,
and training resources to help
suppliers better mitigate
occupational and environmental
risks related to the chemicals they
use.

In 2024, we, along with other

companies, provided seed funding
for the Safer Chemistry Impact
Fund, which aims to advance
science- based, data-driven
solutions to systematically
eliminate hazardous chemicals and
replace them with verified, safer
alternatives.

RESOURCES
Percent
Introduct AI for Our Our Appen

Nature and biodiversity

We strive to restore and enhance nature and biodiversity through our campuses and technology

Our approach

Building for biodiversity

Developing guidelines for urban nature

Creating and restoring habitat

Managing and monitoring habitat

Restoring nature and making

it more accessible
Connecting people with nature

Sourcing responsibly

Developing technology to
address biodiversity loss
 Introduct AI for Our Our Appen

Our approach to nature and biodiversity

Preserving nature is critical for

FIGURE 28 Our four-pillar approach to nature conservation and 2023 highlights Global challenges
supporting biodiverse
ecosystems and healthy biodiversity enhancement
Habitat creation and restoration: As of the end ofLocal
2023,context,
we created
globalorscale:
restored
Navigating
approximately
the local67
compl
acre
communities, as well as for health while scaling global solutions requires careful
mitigating and adapting to Pollinator habitat: From 2021 to 2023, we created 20 acres of new habitat on our California campuse
Accounting for benefits of nature: Ecosystems are com
climate change. FSC-certified lumber: Approximately 99% of the new lumber used for our Gradient Canopy building (bo
the FSC. 177 Metrics and measurement: As the global consensus on
We strive to restore and enhance related metrics continues to evolve, organizations are
nature and biodiversity through a four- Building for biodiversity
pillar approach that starts with BWue iblduiildn gfo rf obiro dbivioerdsiitvy eorns-sititye at Google offices and Urban nature: Building nature in urban settings requir
Wcaem bpuilsde fso br yb dioedsiivgenrisnigty f onr -escitoelo agt yG aonodg hlee lopffi
incge tso a bnrdin g coordination with capital planning, design, and maint
designing our offices and campuses for
cnamtupreu sbeasc kb yto d ceistigensing for ecology and helping to bring nature back to cities
ecology and actively managing our
sites to maintain healthy ecosystems.
This includes everything from
developing guidelines for urban nature,
to increasing habitat creation,
restoration, and monitoring.
Restoring nature and making it more accessible
We also focus on restoring nature and We nurture the environment and connect people to nature, both on our campuses and in the communities where we operate
making it more accessible in the
communities where we operate,
sourcing responsibly, and developing
breakthrough technologies to
address nature and biodiversity loss and
assist various conservation initiatives (see
Sourcing responsibly
Figure 28). SWoe usorcuricneg r ersepsopnsoibnlys iabcrlyoss our supply chain, focusing on
Wpreo csuoruinrcge s ruestpaoi nasbiblely b aucilrdoinssg oaunrd s huaprpdlwy acrhea imn,a ftoecriuaslsin agn odn psuropcpuorr
tiningg s ubsiotadiinvaebrslee bfouoil di nsgys atn

Developing technology to address bDieovdeivloeprisnitgy t leocshsnology to address

Wbieo cdreivateer psliattfyo rlmoss,s open tools, and geospatial and AI-powered
solutions that help increase nature’s resilience to climate change aWned cardedartes psl athtfeo rcmauss, eosp eonf ntoatoulsre, a londss geospatial and AI-powered solutions that
and address the causes of nature loss
 Introduct AI for Our Our Appen

Building for biodiversity

We’re helping to bring nature control levees.

back into the built Creating and restoring
environment— for example, habitat
by restoring critical habitats We’re expanding native, biodiverse
like oak woodlands and habitats on our development sites
willow groves across our San through both large- and small-scale
Francisco Bay Area efforts that prioritize local species. As of
campuses. the end of 2023, we created or restored
approximately 67 acres of habitat and
planted roughly 4,500 native trees on
Developing guidelines Google’s campuses and the
for urban nature surrounding urban landscape, primarily
in the San Francisco Bay Area.
Our campus biodiversity efforts are
underpinned by our Urban Nature We made progress on monarch butterfly
Design Guide that establishes criteria habitat creation: from 2021 to 2023, we
for ecologically designed landscapes. We created 20 acres of new habitat on our
made this guide freely available for California campuses designed to support
landscape managers and the public. monarch butterflies and other pollinators,
aligned with Google’s 2021 monarch
Our ecology team also issues internal butterfly pledge. We’re already seeing
guidelines on topics such as pollinator benefits, with about 10 times more
habitat design, drought-tolerant native monarch caterpillars at our new habitat
landscaping, and native landscape sites, compared to control sites on our
maintenance to translate ecological campus without new monarch- friendly
science into action. plants. Our efforts to support pollinator
habitats earned us the 2023 Monarch
To make the nature-friendly urban design
Sustainer of the Year award from
approach broadly accessible, we helped
Pollinator Partnership.
support the Making Nature’s City Toolkit,
developed by the San Francisco Estuary In 2023, we also completed a native
Institute and the International Union for habitat restoration project along the
Conservation of Nature. West Channel near our Sunnyvale,
California, campus. This project restored
We’ve also published our research on
4.3 acres of aquatic, wetland, and
balancing nature with urban density in
riparian habitat and is expected to
an open-access academic paper—
benefit local and migratory water birds.
aiming to provide others with
Additionally, the project provides flood
replicable tools and techniques for
protection for adjacent neighborhoods
integrating nature into cities.
in Sunnyvale via
a combination of floodwalls and flood
 Introduct AI for Our Our Appen

c our sites to ensure we’re maintaining

M e habitat quality—including assessing our
a sites as part of ongoing maintenance,
w prioritizing wildlife-friendly management
n e practices, and refreshing planting areas.
a ’ At our Bay View campus in Mountain

g v
e
View, California, we’ve worked to protect
nesting cliff swallows,
i and we’ve piloted a rodenticide-

n i
n
free pest management approach. In
2023, we
g s documented more than 10 species at
t our Bay View campus that we’ve
a avoided exposing
a l to rodenticides as a result of this effort—

n l
e
including gray foxes, black-tailed jack
rabbits, and red-tailed hawks.
d d
We also monitor our sites for plant
n health and wildlife biodiversity. In
m a addition to monitoring bird biodiversity

o t
i
in New York City, we monitored birds
during the fall and winter of 2023 at our
n v Bay View campus in California and

i e identified more than 60 bird species

using the open water, wetland, oak
t l woodland, and grassland habitats—
o a including several species where we’ve
documented evidence of breeding on-
n
r d site. Earlier projects are also showing
i s success, such as the Charleston
c Retention Basin, where oak woodland,
n a willow groves, and wetlands have filled
g p into the basin since our initial planting
i in 2016.
n
h g We incorporated bird-friendly design
elements in all of our San Francisco
,
a Bay Area ground-up construction
b w office projects completed in 2023.
e We’re monitoring key
i buildings to evaluate the effectiveness
t m of our design approach in preventing
a collisions.
a n
t a
g
O e
n
 Introduct AI for Our Our Appen

Restoring nature and Sourcing Developing technology to

making it more accessible responsib address biodiversity loss
ly
To complement our on- add to menus across the globe.
initiative, and in Japan through a We’re focused on sourcing We build tools and technology
campus biodiversity efforts, partnership with Ramsar Network Japan.
we’re supporting ecological responsibly across our supply that enable partners, NGOs,
For more details, see the Water
restoration efforts beyond chain, aiming to minimize governments, and academics
stewardship section.
our campuses to deliver negative impact on global around the world to help address
benefits for communities and Connecting people with nature biodiversity by procuring nature and biodiversity loss.
ecosystems. sustainable building and
We engage with local communities and hardware materials and The public’s increased search interest in
“wildlife”—which saw a 138% increase over the
As examples, we’ve supported the lead programming that extends the supporting biodiverse food
benefits of nature to all. In partnership past two years 179 and reached an all-time high
restoration and enhancement of over systems. in 2023 180—likely indicates a growing desire to
750 acres of monarch butterfly habitat with the Santa Clara Valley Audubon
Society, the City of Mountain View, and understand and support conservation efforts.
across California—achieving Google’s For new campus developments, we’ve
2021 monarch butterfly pledge, and other organizations, in 2023 we used timber certified by the Forest As concerns for wildlife escalate, innovative
we’ve partnered with Southeastern continued hosting “Egret Office Hours” Stewardship Council (FSC)—the world’s technological tools empower us to observe
Grasslands Institute to restore at the Shorebird Way rookery to provide leading forest certification system for animal populations and habitats with
approximately 100 acres of native members of the public with the sustainable wood building materials. unprecedented detail, providing essential
prairie ecosystem near our data center opportunity to learn about local birds
insights for protection and conservation.
in Clarksville, Tennessee. and witness moments throughout the Approximately 99% of the new lumber
egret breeding cycle. used for our Gradient Canopy building Below are just a few examples:
Many of our water replenishment projects (both temporary and permanently
also deliver co-benefits for nature and In 2023, we completed the first sections installed) was procured from • Giant kelp monitoring: Giant kelp forests,
biodiversity. For example, in 2023 we of the Moffett Park Green Link at our responsibly managed forests certified crucial to Australia’s Great Southern Reef
supported the acquisition of a 20-acre Sunnyvale, California, campus—a by the FSC. 178 And at our new San ecosystem, are disappearing rapidly due to
parcel of land in California’s Central Valley publicly accessible landscape network Bruno campus, the mass timber climate change. That’s why we’ve partnered
that will enable floodplain reconnection that connects to regional walking and structures of the building are built with with leading scientific and conservation
and the restoration cycling corridors. Some of the key sustainably sourced FSC MIX-certified organizations to use AI-powered tools to
of native riparian forest habitat along components of the Green Link include timber materials. map the remaining kelp forests, identify
the Tuolumne River. We’re also advancing the restoration of native plant and
heat-resistant kelp varieties for restoration
wetland restoration efforts with a project tree species, as well as landscapes In partnership with our food vendors, we efforts, and raise awareness of this critical
in Belgium’s Sint-Onolfspolder nature designed to reduce demand on leverage procurement practices and environmental issue.
reserve via a partnership with local stormwater management menu design to replace monocrop
infrastructure. • EEAGER: The elusive beaver, a keystone
Natuurpunt’s Wetlands4Cities commodities with climate-resilient and
agrobiodiverse crops species and wetland architect, is finally
to ensure our food operations contribute shedding some of its secrecy thanks to
positively to global agricultural EEAGER—an image-recognition machine
biodiversity. For example, our culinary learning model pinpointing telltale dam
teams select specific regionally complexes amidst vast landscapes, saving
biodiverse beans, grains, and legumes to conservationists countless hours in their
quest to understand this industrious
 Introduct AI for Our Our Appen

engineer and protect their habitat.

• Eyes on Recovery: Australia’s
devastating bushfires were a
stark reminder of the fragile
balance of nature. AI-powered
cameras are capturing the
astonishing resilience of wildlife,
revealing signs of recovery and
offering conservationists a
powerful new tool to track wildlife
over time.
• Biodiversity on Britain’s railway: The
humble dormouse, nestled along
Britain’s sprawling railway
network, is finding new hope as
AI models illuminate their secret
lives and guide efforts to protect
them from dwindling numbers.
• Blue carbon: Similar to forests on
land, seagrass can take carbon
dioxide out of the atmosphere as
it grows—storing it through a
process called carbon
sequestration.
We’re working with partners to
explore novel AI applications to
measure the capacity
of seagrass ecosystems to
absorb and sequester carbon.

RESOURCES
Introduct AI for Our Our Appen

Appendix
About Google

Sustainability governance

Risk management

Stakeholder engagement and partnership

Multi-sector products

Ecosystems for collaboration

Environmental data

Certifications

Recognitions

Endnotes
 Introduct AI for Our Our Appen

About Sustainabili Risk

Googl ty management
e governance
As our founders explained in related risks and opportunities and
their first letter to shareholders, To learn more, see our data center Alphabet’s Board of Our Enterprise Risk understand their associated impact.
locations and our office locations. Directors has delegated to Management (ERM) team works
Google’s goal is to “develop We’ve increased our efforts to align our
services that significantly its Audit and Compliance with subject matter experts
climate risk assessment process more
improve the lives of as many Committee the primary across the enterprise to identify,
closely with the recommendations of the Task
people as possible.” responsibility for the assess, and report Force on Climate-Related Financial
oversight of many of the risks related to the company’s Disclosures (TCFD), leveraging the TCFD
We believe in technology’s potential risks facing our businesses. operations, financial categories
to have a positive impact on the performance, and reputation.
world. That The Audit and Compliance Committee
unconventional spirit has been a driving reviews and discusses with management As with financial, operational, and
force throughout our history, inspiring us any major risk exposures, including strategic risks, the team assesses
to tackle big problems and invest in sustainability risks, and the steps that environmental risks as part of the
moonshots, such as our long-term Alphabet takes to detect, monitor, and company’s overall
opportunities in AI. We continue this actively manage such exposures. risk management framework. The
work under the leadership of Alphabet risks and opportunities identified
and Google CEO Sundar Pichai. Our Sustainability Focus Area, an internal through this process support public
management team led by our SVP of disclosures and
Alphabet is a collection of businesses— Learning and Sustainability, provides inform Google’s environmental
the largest of which is Google. Google centralized management oversight of sustainability strategy. Our Chief
comprises two segments: Google sustainability and climate-related issues. Sustainability Officer and sustainability
Services and Google Cloud. Google teams work to address risks
Services’ core products and platforms The Sustainability Focus Area includes by identifying opportunities to reduce the
include ads, Android, Chrome, devices, the Chief Sustainability Officer and company’s environmental impact from its
Gmail, Google Drive, Google Maps, executives from across the company operations and value chain, and by
Google Photos, Google Play, Search, and with diverse skills, from teams such as improving climate resilience.
YouTube. Our devices include Fitbit operations, products, finance,
wearable devices, Google Nest home marketing, legal, communications, and
products, and Pixel devices. Our Google policy, among others. Climate-related risks
Cloud offerings include Google Cloud Climate-related risks and
Platform and Through the Sustainability Focus Area,
opportunities can span multiple time
Google Workspace. sustainability and climate ambitions
horizons and may have varying levels
are built into our company-wide goals,
of uncertainty regarding how climate
Our headquarters are located in plans of action, management policies,
trends, policy, and socioeconomic
Mountain View, California. We own and performance objectives, and how we
factors might evolve in the future.
lease office facilities and data centers monitor progress.
Google continues to build on
around the world, primarily in Asia, qualitative and quantitative risk
Europe, and North America. assessments to identify climate-
 Introduct AI for Our Our Appen

suppliers through the CDP Supply

of risks and opportunities and In 2023, we launched our data Chain Water Security questionnaire,
conducting climate scenario analyses. center water risk framework that inviting them to disclose their water
In an effort to drive completeness and allows us to evaluate the health of a management efforts.
consistency when reviewing these local community’s watershed and
categories, we adopted our ERM rating establish a data-driven approach
scales (i.e., impact, frequency, to advancing responsible water
likelihood, control effectiveness) to use across our data center
identify and prioritize areas of focus. portfolio. We use our decision-
making tool to evaluate the
Beginning in 2022, climate-related risks environmental responsibility of
and opportunities were analyzed across using water for cooling at new or
three time horizons—short term (through expanding data centers. This
2030), medium term (through 2040), and context-based water risk and
long term (through 2050)—for financial, impact methodology generates
operational, legal, and strategic risks. We more granular insights than can
considered acute and chronic physical be provided by other water-risk
risks (e.g., heat stress, water stress, and screening tools. It provides
extreme weather events), us with a framework to measure
as well as risks associated with and evaluate site-level water risks
transitioning to a low-carbon economy and the potential watershed
(e.g., energy costs, future regulations, impact to inform our decision-
and technology). We also assessed making process for new site
climate-related opportunities (e.g., selection, cooling system design,
developing low-carbon products and and ongoing operations.
services, improving energy efficiency,
and advancing energy technologies). To identify and assess water-
For more details, see our CDP Climate related risks in our supply chain,
Change Response. we’ve conducted a supply chain
water use analysis and a supplier
Water-related risks risk assessment using WRI’s
Aqueduct Water Risk Atlas, WWF’s
To identify and assess water-related Water Risk Filter, and WULCA
risks in our direct operations, AWARE. The key risks identified
Google annually undertakes a water included baseline water stress,
risk assessment of our data centers flood risk, access to safe drinking
and offices to identify water, and the level of sanitation
potential water-related risks that may and hygiene services.
present opportunities for water
stewardship action and risk mitigation. To safeguard the health of local
Indicators from available risk assessment waterways, we comply with all local
tools, including WRI Aqueduct Water Risk environmental regulations at our
Atlas 3.0 and WWF Water Risk Filter 6.0, Google-owned and -operated data
are blended with other metrics to centers and further stipulate that
evaluate risks related to scarcity, suppliers “[treat all wastewater] as
flooding, water quality, sanitation and required prior to discharge or
hygiene, reputation, and regulatory disposal.” We conduct regular
stressors. supplier audits to monitor
adherence to our code of conduct.
Additionally, we’ve engaged our
 Introduct AI for Our Our Appen

Stakeholder engagement and partnership

We recognize that suppliers evaluate their operations, we Conduct requirements. Lastly, we monitor
achieving our own Employees Suppliers perform our own ongoing due diligence and verify all corrective actions are
sustainability goals and Sustainability is part of Google’s Through our Supplier Responsibility and audits to verify compliance and to completed in the agreed-upon time
addressing the urgency of culture, with passionate employees Program, we’re working to build an understand our supply chain’s current frame, with a process for escalation if
actively driving environmental energy-efficient, low-carbon, circular and potential risks. necessary to the Supplier Responsibility
climate change and
initiatives through dedicated groups supply chain. We focus on the areas Steering Team, which comprises our
sustainability requires We investigate any issues identified Chief Compliance Officer and leaders
and engagement opportunities. where we can make an immediate and
engagement, collaboration, during an audit, and when we find that a from our data center, devices, and
lasting impact, such as helping our
and partnership across a For example, employees can join global suppliers improve their environmental supplier isn’t conforming to our extended workforce teams.
diverse set of stakeholders. and local internal community groups performance. expectations, we expect the supplier to
focused on sustainability. One collection provide a corrective action plan that In 2023, we audited a subset of our
That’s why we actively engage with a of sustainability- related internal Google’s Supplier Code of Conduct outlines the root cause of the finding, suppliers to verify compliance for various
wide range of stakeholders—including community interest groups had 22 includes requirements that enable us how and when they will resolve the environmental criteria (see Figure 29):
employees, suppliers, NGOs, chapters across 15 countries as of the to ensure issue, and what steps will be taken to
policymakers, customers, startups, end of 2023—focused on organizing local that those we partner with are prevent recurrence. We determine
researchers, academics, investors, and sustainability activities and raising responsible environmental stewards. whether the plan is acceptable based
more. These engagements and Along with having on our Supplier Code of
partnerships are essential for: awareness about environmental topics for
interested Googlers.
• Overcoming barriers to unlock new FIGURE 29
2023 audit conformance data for environmental criteria
opportunities: We work together Another sustainability-related internal
to
overcome obstacles and accelerate community group has more than 3,700 The lighter bars show the percentage of unique audited supplier facilities that had no findings for the listed criteria after
advancements in sustainability. Only their
members globally. This community
through collaboration can we develop audit. The darker bars show the percentage that had no findings after the corrective action plan (CAP) process was
group hosts weekly climate talks
completed.
and featuring internal
implement solutions on a global scale. and external speakers and e sustainability courses, internal
See below for more information biannual events highlighting Our newsletters, campaigns, and 96% 98%
• Shared learning: Our engagement
on howabout
we engage withproducts
some section, is a sustainability-related 20% Cool websites. Environmental
work also enables us to better more in the Our Air emissions 97% management
97% system
Roofs 88%
specific stakeholder groups.
successful example of an effort that originated opportunities—whereby
understand our stakeholders’ initiat 96% 98%
perspectives, elaborate on our Google employees are ive,
as a 20% project. Environmental permits
97% 98% and reporting
Product
allowed content
to use 20% ofrestrictions
their which
environmental strategy, and progress 90%
work time to explore you
against key targets, and it creates a Employees can also learn about sustainability can 95%
vital innovative ideas beyond their Hazardous substances
learn
two-way dialogue that informs our current roles. This enables
approach to the work. engineers with a diverse 92%
Wastewater and solid waste
range of technical skill sets
• Driving systemic change: These
to be deployed into
partnerships are crucial for advancing
addressing climate and
carbon-free energy technology
sustainability challenges,
investment, shaping effective throu
which is critical to unlocking
policies, and scaling up climate gh
innovation and solutions.
solutions around the world. onlin

87% 91%
Introduct AI for Our Our Appen

Legend Percentage in conformance

Resource efficiency before CAP Percent
age in
conform
ance
after
CAP
 Introduct AI for Our Our Appen

Policymakers United States

Engagement on U.S. federal sustainability, climate, and energy policy
Public policy and advocacy

Policy measures and corporate commitments will continue to play an important role in driving emissions Google had multiple engagements across the White House; Departments of State, Defense,
Executive branch
reductions in the next decade. See Figure 30 for our key positions on sustainability policy issues and Figure 31 for Energy, and Transportation; and the Environmental Protection Agency to discuss the role of
engagement
digital technology and AI in accelerating climate mitigation and adaptation.
a detailed list of our sustainability policy engagements in 2023.
In response to the FTC’s request for public comment on the Green Guides update, we
FTC Green Guides filed comments encouraging the FTC to prioritize true and fair representations in
FIGURE 30 Our sustainability policy positions environmental marketing claims—with scientific integrity and technical accuracy acting as
a north star.

Federal Energy
Climate Carbon-free Accelerating Device Regulatory Commission
In 2022, Google filed comments (initial and reply) on FERC’s “Improvements to Generator
energy climate-action with AI repairability Interconnection Procedures and Agreements (Interconnection) Notice of Proposed
(FERC) Generator
See Our position on climate See A policy roadmap for
Rulemaking.” FERC’s final rule, issued in June 2023, acknowledged Google input, including it
See Accelerating climate Interconnection
policy See Google & as a basis for their actions on issues around transparency and study delays.
24/7 carbon-free action with AI repairability final rule
energy

In September 2023, Google submitted comments to FERC supporting California Independent

Western energy System Operator’s (CAISO) proposal to expand its energy market across the Western
market expansion Interconnect—an important step toward a western Regional Transmission Organization.
FIGURE 31
Google’s policy engagements in 2023 FERC accepted CAISO’s proposal in late December, citing our comments in their final
order.
Global and cross-cutting initiatives FERC transmission In March 2023, Google submitted comments to FERC regarding the cost of
expansion and cost transmission projects, a significant challenge for our portfolio. Our comments
Google published a paper that highlights the importance of quickly commercializing advanced management advocated for policies that balance strong oversight of transmission project spending
Advanced clean clean electricity technologies alongside the rapid growth of solar and wind. The paper outlines with the need for timely grid expansion.
electricity technologies how corporate clean energy buyers can drive this progress by supporting favorable policies,
Engagement with coalitions and sustainability initiatives
signing long-term purchase agreements, and providing early-stage project funding.

We collaborated with partners across tech and traditional commercial and industrial
Accelerating climate We released a report with Boston Consulting Group which discusses AI’s potential to mitigate
sectors to drive customer-centric campaigns like the Electricity Customer Alliance (ECA)
action with AI global GHG emissions and provides recommendations for how policymakers can enable Electricity
and Western Freedom. This effort included ECA’s first-ever FERC filing in March 2023, which
deployment of AI for climate. customer coalitions
brought together state officials, trade associations, and think tanks to highlight the
Google participated in COP-28, hosted by the United Arab Emirates government and the importance of transmission infrastructure for meeting load growth, economic development,
United Nations. Our delegation—including Ruth Porat (President & Chief Investment Officer; and corporate sustainability goals.
UNFCCC 28th
Chief Financial Officer), Kate Brandt (Chief Sustainability Officer), Yossi Matias (VP of
Conference of the U.S. state engagement
Engineering & Research), along with a number of senior subject-matter experts—participated
Parties (COP-28) in nearly 200 engagements on the ground throughout the conference with public sector
Google participated in many regulatory proceedings and dockets across the United States,
leaders from across the globe, advocating for greater climate ambition and showcasing the Utility regulation
collaborating with coalition partners to promote the cost-effective adoption of clean
role AI can play in enabling mitigation and adaptation.
energy resources.
Google participated in the 2023 UNGA meetings in New York City, which marked the halfway
United Nations General Google led discussions with the National Association of Regulatory Utility Commissioners
point on the 15- year timeline to achieve the 17 UN SDGs. At UNGA, we showcased the role that Regulatory frameworks
Assembly (UNGA) and the National Association of State Energy Officials to discuss how Google’s 24/7 CFE
AI and digital technology can play in helping partners achieve and track progress toward the for decarbonization
goal can be a supportive framework to drive cost-effective grid decarbonization.
SDGs.

Tripling global We joined a letter calling for a global target to triple renewable energy capacity by 2030, which We affirmed our support for the “Right to Repair” movement, advocating for legislation like
renewable energy governments recognized in the COP-28 decision. We also joined a letter highlighting the role of Right to Repair the proposed bill introduced in Oregon and publishing a white paper outlining our approach to
capacity corporate clean energy buyers in supporting this objective. repair issues. This paper provides a set of principles for policymakers to consider when
creating regulations.
We announced our role as a founding sponsor of the Schneider Electric Catalyze program,
Founding sponsor of
 Introduct AI for Our Our Appen

Engagement with coalitions and sustainability initiatives

Europe
Engagement on European sustainability, climate, and energy policy In 2023, we worked with RE-Source to advocate for a stronger role for corporate clean
RE-Source Platform energy buyers within Europe’s evolving Electricity Market Design. These efforts led to the
Google engaged with EU policymakers through DIGITALEUROPE to inform the development adoption of new measures promoting PPAs and improving renewable energy access for
Energy Efficiency smaller buyers.
of a standardized energy and sustainability reporting framework for data centers and
Directive
establish measures to encourage greater reuse of waste heat.
We helped develop a methodology for measuring the net environmental impact of
European Green
information communication technology (ICT) solutions. We also participated in EGDC-
Digital Coalition
Google worked through DIGITALEUROPE and RE-Source to advocate for the inclusion of led panel discussions at COP-28 and the GeSI Digital with Purpose Summit.
Renewable Energy
time-stamping for Guarantees of Origin in the EU Renewable Energy Directive—enabling
Directive Sustainability
hourly CFE matching and greater transparency of clean energy claims. We announced a partnership with the EU Aviation Safety Agency (EASA) to develop
partnership with
standardized aviation emissions estimates. This will provide travelers consistent and reliable
European Aviation
We shared views on the future of European electricity market design in response to an EU flight emissions estimates available through airline partners.
Safety Agency
consultation, advocating for measures that support corporate clean energy procurement
European electricity We worked with the European 24/7 Hub, a collaboration with Eurelectric to create a platform
and drive European grid decarbonization. In partnership with the RE-Source Platform, we
market redesign European 24/7 Hub where energy buyers, suppliers, and policymakers can meet to learn more about 24/7 CFE
successfully advocated for a stronger role for corporate clean energy buyers in the EU’s
energy transition, with the adoption of several measures in the final legislation that and receive technical training and implementation guidance.
enhance the role of PPAs.

In response to the EU Commission’s consultation on the EU’s 2040 climate targets, we

Asia Pacific
EU 2040 climate targets reaffirmed our support of EU climate leadership. Our submission outlines how The Asia Clean Energy Coalition (ACEC), of which Google is a founding member, launched a
technology can drive climate action and shares policy recommendations for achieving ASEAN electricity grid
paper, supported by Google, advocating for accelerated regional grid integration within
climate neutrality by 2050. regionalization paper
ASEAN. This paper highlights the economic, energy security, and clean energy benefits of
regional electricity markets.
Google Climate Across five different events throughout the year, we hosted hundreds of leaders—
Summits in Brussels, including policymakers, members of Parliament, journalists, and decision-makers from We participated in the APEC Economic Leaders Week in San Francisco, California, to
Germany, France, business, NGOs, and academia—for a series of discussions focused on the role of Asia-Pacific Economic showcase the role of AI and digital technology in accelerating climate action in the Asia-
Sweden, and Switzerland technology and AI in addressing the climate crisis. Cooperation (APEC) Pacific region. We also participated in the APEC Energy Ministerial in Seattle, Washington, to
discuss the role of corporate clean energy purchasing in helping accelerate grid
decarbonization and meeting APEC member state climate goals.
We contributed to a report by our trade association, DIGITALEUROPE, highlighting how digital
DIGITALEUROPE Temasek Ecosperity and We engaged policymakers and private sector stakeholders in conversations about the critical
technology can accelerate energy sector decarbonization. The report features three Google
report on energy Asia Tech x Singapore role of data, digital infrastructure, and green computing in sustainability. These discussions
case studies (data center cooling, Nest, and Google/ENGIE on wind energy deployment) and
digitalization (ATxSG) Summit centered on the nexus between technology and sustainability, as global demands on
has been shared with EU policymakers and the International Energy Agency (IEA).
computing power and AI increase globally.

We sponsored and co-hosted a closed-door roundtable with Climate Group on

We worked constructively with the German government on their Energy Efficiency Bill.
Germany data center democratizing access to high-quality granular energy data to advance power system
Our technical deep dives into waste heat recovery at our data centers—a key feature of
energy efficiency Asia Action Summit decarbonization. Moderated by the IEA, this event brought together key stakeholders—
our first Google-owned Germany data center in Hanau—informed these discussions.
including energy producers, consumers, data providers, researchers, and governments—
to discuss opportunities in the Asia-Pacific region and develop a call-to- action for
In 2022, we commissioned Implement Consulting Group’s Digital Decarbonisation report, policymakers to facilitate greater access to energy data.
which demonstrated that digital solutions are essential to achieve 20–25% of the GHG
Implement Consulting We participated in AEBF and worked closely with the Asia Clean Energy Coalition (ACEC) to
reductions needed for a net-zero EU economy. In 2023, we expanded this study with ASEAN Energy Business
regional reports organize a session on accelerating power grid interconnectivity, as well as expressed support
country-specific reports for Sweden, Forum (AEBF)
for strengthening the current Memorandum of Understanding on the ASEAN Power Grid.
Germany, Belgium and Denmark—outlining how targeted digital solutions can accelerate
climate action in each country.
 Introduct AI for Our Our Appen

Trade associations and For example, we’re founding members of doesn’t mean that we endorse the
third-party groups the Chamber’s Task Force on Climate organization’s entire agenda, its events
Actions, and we’ve engaged within the or advocacy positions, or the views of
We belong to many sustainability- Task Force since its inception to support its leaders or members. We assess the
focused third-party groups through constructive engagement by the alignment of our trade association
which we engage on sustainability Chamber on climate policy to create participation with the goals of the Paris
policy issues around the world. See a low-carbon economy. We also Agreement, and engage within
Figure 32 for an overview of our participate in staff-level discussions on organizations to support advocacy for
participation in these groups, and the Business Roundtable’s Energy and climate policies needed to limit warming
consult our CDP Climate Change Environment committee. to 1.5°C and create a prosperous and
Response for additional details. competitive low- carbon economy. We’re
We respect the independence and in dialogue with our trade associations
We’re members of the U.S. Chamber of agency of trade associations and third to encourage alignment between our
Commerce, Business Roundtable, and parties to shape their own policy core public policy objectives and their
other business trade associations where agendas, events, and advocacy policy advocacy activities, including on
we’re engaged in climate and energy positions. Our sponsorship or climate change.
policy issues. collaboration with a third-party
organization

FIGURE 32
Select list of Google’s participation in sustainability-focused trade associations,
memberships, and groups

• Advanced Energy Buyers Group Leadership • Clean Energy Buyers Alliance

• Advanced Energy United • Carolinas
• Clean Energy Demand Initiative
• Advanced Power Alliance Clean
Energy • Clean Grid Alliance
• Alliance to Save Energy Business • Conservation Voters of South Carolina
• American Clean Power Associatio
• Corporate Eco Forum
Association n
• Data Center Coalition
• American Council on Renewable • Clean Air Task Force
Energy • DIGITALEUROPE
• Americans for a Clean Energy • Energy Alabama
Grid • Energy Storage Association
• Asia Clean Energy Coalition • Energy Systems Integration Group
• Business Alliance to Scale Climate • Energy Tag
Solutions
• Eurelectric
• B usiness Environment
• Fusion Industry Association
Leadership Council of the Center
for Climate and Energy Solutions • GeSI
• C ambridge Institute • Japan Climate Leaders Partnership A green wall at our inaugural Google European Sustainability Summit in Brussels.
for Sustainability • Keystone Energy Board
 Introduct AI for Our Our Appen

• Long Duration Energy Storage • SolarPower Europe

Council • Trellis Network
• Marktoffensive Erneuerbare • U.S. EPA Green Power
Energien Partnership
• N • Utah Clean Energy
o
r • We Are Still In
t • WindEurope
h

C
a
r
o
l
i
n
a

S
u
s
t
a
i
n
a
b
l
e

E
n
e
r
g
y

A
s
s
o
c
i
a
t
i
o
n

• Peak Load Management Alliance

• Princeton ZERO Lab
• RE100
• Renewable Northwest
• RE-Source Platform
• Resources for the Future
• Solar Energy Industries Association
 Introduct AI for Our Our Appen

Google partners with many organizations to accelerate progress toward shared

Partner sustainability goals. Examples of some key partnerships are listed below (see Figure Global Covenant of Google’s Environmental Insights Explorer was developed in partnership with GCoM through a shared
vision to support city climate action with useful and accessible data and insights. Today, GCoM is a
Mayors for Climate
33). strategic partner, sharing EIE data with its alliance of cities and local governments to accelerate climate
& Energy (GCoM)
action.

FIGURE 33 Google is a supporter of GRA’s campaign to triple renewable energy capacity globally by 2030 and have
Key partnerships Global Renewables
supported efforts to encourage high-impact corporate clean energy purchasing as a key strategy to
Alliance (GRA) accelerate progress toward this goal.

ICLEI Africa Google is a partner of the regional secretariats of ICLEI—Local Governments for Sustainability—in Africa,
Organization Details Europe, and the United States. Through these partnerships, ICLEI regional teams support sustainable
ICLEI Europe
ICLEI USA development projects in cities with data and insights from EIE.
24/7 Carbon-Free In 2021, Google helped launch the 24/7 Carbon-Free Energy Compact in partnership with Sustainable
Energy Compact Energy for All and UN-Energy to help grow the movement to enable zero-carbon electricity. We’ve partnered with the IEA on multiple energy-related projects. In 2022, we sponsored research by the
IEA on advancing decarbonization through clean electricity procurement and we launched a Search feature
Google is a founding supporter of and active participant in Ad Net Zero—a global initiative to help the International Energy
Ad Net Zero based on IEA data designed to inform people about the energy crisis in Europe and provide energy-saving
advertising industry tackle the climate crisis. Agency (IEA) tips. In 2023, we again leveraged IEA data for new Search features on electric vehicles and home heating.
In February 2024, Google participated in the IEA Ministerial Meeting and 50th Anniversary in Paris, France.
Bonneville Google has partnered closely with BEF since 2019 on the implementation of our water strategy, including
Environmental identifying and facilitating impactful water replenishment and watershed health projects globally, with a Google is a founding member and an active participant in the Governing Body of the iMasons
Foundation (BEF) variety of local organizations and partners. iMasons Climate Accord
Climate Accord, a coalition united on carbon reduction in digital infrastructure.

Business for Social Google has been a BSR member for many years and is one of a few select Spark members. We participate Since 2018, Google has been working with ReFED—a nonprofit with a mission to catalyze the food system
Responsibility (BSR) in a number of BSR collaboration initiatives, and one of our senior leaders sits on its board. toward evidence-based action to stop wasting food—supporting their technical teams and exploring
ReFED
ways to convene businesses. In 2022, to activate industry-wide change, Google provided anchor funding
C40 and Google launched the 24/7 Carbon-Free Energy for Cities program to empower cities around to kickstart the ReFED Catalytic Grant Fund, which selected as grantees 10 organizations working to
C40 Cities the world to run entirely on clean energy. In 2023, the program was expanded into eThekwini, South accelerate and scale food waste solutions.
Africa with grant support from Google.org. C40 is a strategic partner of Google’s Environmental Insights
Google has supported TNC on watershed projects in Chile and the United States. Separately, Google.org
Explorer.
supported a three-phased approach to catalyze active reforestation of kelp at impactful scales.
The Nature
Coalition to End Wildlife In 2018, Google and other companies launched the Coalition to End Wildlife Trafficking Online, Google.org also provided a grant to TNC to develop a machine-learning-powered timber-tracing API to
Conservancy (TNC)
Trafficking Online collectively creating a wildlife policy framework for online trade and an industry-wide approach to stop deforestation in the Amazon at scale. A team of Google engineers worked full-time for six months
reduce online wildlife trafficking. with TNC to develop this product as part of the Google.org Fellowship Program.

In addition to reporting our carbon footprint to CDP since 2009, Google partnered with CDP to host its In 2023, Danfoss, Google, Microsoft, and Schneider Electric—together with the Danish Data Center
Net Zero Innovation
CDP annual conference, host a hack-a-thon, and launch CDP scores in Google Finance, making corporate Industry—launched the Net Zero Innovation Hub for Data Centers, a pan-European consortium located in
Hub for Data Centers
carbon disclosure information more widely available. Denmark designed for cross- industry collaboration on decarbonizing the data center industry and
ensuring grid stabilization.
Google was actively involved in the creation of CEBA in 2018. A Google representative continues to serve Together with Insomniac Design and CyBourn, Google supported the design and build of the NZPDU proof
Clean Energy Buyers
as the Board Chair of this organization. In 2022, Google.org provided a $1 million grant to the Clean Energy
Association (CEBA) of concept, which is based on recommendations from the Climate Data Steering Committee. Hosted on
Buyers Institute to support CEBA’s international expansion. Net Zero Public Data
Google Cloud and launched in 2023 at COP28, the NZDPU proof of concept aims to be a freely available,
Utility (NZPDU)
Climate Neutral Data Google helped establish the CNDCP, a coalition of European data center operators who commit to a set of global repository of company-level climate transition-related data, allowing users to easily access and
Centre Pact (CNDCP) voluntary sustainability targets to set them on a path toward climate neutrality. interpret a core set of data that’s crucial to realizing the net- zero transition.

Google joined the Ellen MacArthur Foundation’s Network in 2015 and, as a Network Partner, has jointly co- United Nations Food
Ellen MacArthur Since 2015, Google and the UN FAO have partnered on the monitoring of forests, natural
authored thought leadership white papers and case studies covering safer chemistry, building and Agriculture
resources, livelihoods, and the environment.
Foundation (EMF) deconstruction and reuse, electronics, and AI’s role in the circular economy. Organization (UN FAO)

Since 2012, Google has partnered with EDF to map air quality using Street View cars in the United States, United Nations In collaboration with UNEP and the European Commission Joint Research Centre, Google launched the
Environmental Defense Europe, and Southeast Asia—as well as to detect methane leaks in U.S. cities. We’ve launched a Environment Program Freshwater Ecosystems Explorer—a platform that enables all countries to freely measure and monitor
Fund (EDF) partnership with EDF’s MethaneSAT to help power their satellite data analysis, map leaks from oil and gas (UNEP) freshwater resources (toward Sustainable Development Goal 6.6.1), as well as when and where surface
infrastructure around the globe, and put methane insights into the hands of scientists and decision- water is changing.
makers.
World Business A member of the WBCSD since 2019, Google actively participates in initiatives related to improving
Google supported the launch of the European 24/7 Hub with Eurelectric, which provides education on the Council for Sustainable well-being for both people and the planet—including shifting diets, influencing consumer behavior
European 24/7 Hub change, and supporting regenerative agriculture.
“what, why, and how” of 24/7 CFE for buyers and suppliers in Europe. Development (WBCSD)

Google is an active member of the EGDC—a group of technology companies committed to supporting Google partners with WEF on various initiatives, including: the First Movers Coalition, which Google joined at
European Green Digital
the green and digital transformation of the EU, harnessing the emission-reducing potential of digital WEF’s annual meeting in 2022; Google’s Chief Sustainability Officer participates in WEF’s Chief
Coalition (EGDC) solutions for all other sectors, and supporting green and digital transformations in the EU. World Economic Forum Sustainability Leaders Community and is a co-chair of WEF’s Global Future Council on the Future of Net Zero
(WEF) Living; Google is an active member of WEF’s Tech for Climate Adaptation initiative, which launched the
In 2021, Google joined ERI and the UN Race to Zero Campaign, the largest ever alliance committed to
Exponential Roadmap report Innovation and Adaptation in the Climate Crisis in 2023; Google
halving emissions by 2030 toward net-zero emissions by no later than 2050. In 2023, Google co-hosted
Initiative (ERI) is a member of WEF’s Alliance for Clean Air; and, Google.org is a member of WEF’s Giving to Amplify Earth
Solutions House during Climate Week NYC with ERI and Futerra to engage innovators and solutionists
Action (GAEA) initiative that convenes public, private, and philanthropic partnerships for climate and
around climate action.
nature.
Google has supported WRI since 2007. Some key WRI projects include developing a near real-time land
 Introduct AI for Our Our Appen

In 2022, Google committed $200 million to Frontier, an advance market commitment that will
Frontier accelerate the development of carbon removal technologies by guaranteeing future demand. As one
of its founding members, we’re helping to guide overall strategy and governance.
 Introduct AI for Our Our Appen

Multi-sector products

While many of our products energy and resource usage across their company), is helping brands gain a guide, which provides
focus on reducing emissions operations and supply chains. For deeper understanding of sustainable a comprehensive overview of how to Google Earth
example, Google Cloud, in partnership build software that uses energy more
within their respective sourcing practices across supplier Google Earth has democratized geospatial
with NGIS (a geospatial solutions networks. By combining efficiently.
sectors, others offer information for a wide range of users. It
the power of our cloud computing, AI, renders a 3D representation of Earth by
a more holistic approach.
and geospatial analytics, we’re helping superimposing satellite images, aerial
These solutions empower our companies get real-time, global, reliable photography, and GIS data onto a digital
customers and partners to information into operations at a local globe, allowing people to explore our planet
make climate- conscious supplier level. This effort includes from endless vantage points.
decisions that support both helping companies like Unilever build
adaptation and mitigation a more holistic view of the forests, Businesses utilize its layers to analyze
across various industries. water cycles, and biodiversity that potential renewable energy sites and optimize
intersect its supply chain. And Regrow— logistics, while governments rely on it for
a technology company that helps its urban planning and emergency response.
Google Cloud customers measure, manage, and Researchers track deforestation patterns,
reduce on-farm emissions—is using explore ocean depths, and visualize complex
Google Cloud offers organizations
Earth Engine and advanced machine scientific models. Citizens use Google Earth
solutions to drive impact for their
learning models to monitor 1.2 billion to travel the world virtually, discover local
business and sustainability. We help
acres of land globally. landmarks, and gain a new perspective on our
organizations harness AI for improved
sustainability measurement to build planet. Its vast collection of images and data
Grow paints a comprehensive picture of the Earth’s
resilience, AI-powered insights to use
We help organizations use AI to find surface and its history.
energy and resources more efficiently
new growth opportunities and markets
in operations and supply chains to
in the low- carbon transition. For Timelapse in Google Earth is a global,
reduce costs, and AI tools to unlock new
example, we launched SpatiaFi with zoomable video that provides a clear picture
growth opportunities and markets
our Cloud partner Climate Engine to of Earth’s dynamic change since 1984,
while accelerating sustainability impact.
help the banking sector harness the illustrating the planet’s transformation.
power of geospatial analytics to Timelapse shows climate change in action,
Measure
support climate finance. as well as beautiful natural phenomena that
We help organizations use AI-powered
insights to monitor their progress toward unfold over decades. Timelapse is being
Build used in partnership with other technologies
sustainability targets in order to build
We help developers reduce the carbon and programs to empower everyone to take
business resilience. For example, our
footprint of their cloud-based climate action across our planet’s cities,
Cloud partner Watershed’s software
applications. We’ve created a suite of oceans, and forests.
platform is used by companies to manage
products in our Carbon Sense Suite so
climate and ESG data, produce audit-
customers can accurately measure, Google Earth can even illuminate potential
ready metrics for reporting, and drive
report, and reduce their cloud carbon solutions to ecological problems—revealing
real decarbonization.
emissions (through Carbon Footprint) suitable terrain for solar arrays, providing
with recommendations for carbon context for conservation efforts, and
Optimize
reduction actions (through Region facilitating discussions about responsible
We help organizations work more A view of our planet from Google Earth ©2020 Landsat / Copernicus
Picker and Active Assist). We’ve also
efficiently by using AI to streamline
put together the Go Green Software
 Introduct AI for Our Our Appen

development. Its insights empower governments worldwide access to up-

decision- makers to create more to-date insights on how our planet is
sustainable strategies for the future. changing.

Google Earth Engine Google Earth Engine empowers a

wide range of stakeholders with
Launched in 2010, scientists, sustainability insights. Powered by
researchers, and developers have been Google Earth Engine, the United
using Google Earth Engine—Google’s Nations FAO supports countries
planetary-scale platform for Earth with forest monitoring to reduce
science data and analysis—to detect deforestation-related emissions.
changes, map trends, and quantify OpenET uses Google Earth Engine
differences on the Earth’s surface. to power science-based
evapotranspiration (ET) models
The tool offers over 1,000 Earth that provide vital agricultural water
observation datasets with powerful cloud usage data in the western United
computing to show timely, accurate, States, aiding efficient water
high-resolution insights about the state of management. MapBiomas
the world’s habitats and ecosystems— contributes to Brazilian
and how they’re changing over time. In conservation with detailed land-
2023, over 100,000 users regularly used use mapping powered by Google
Google Earth Engine’s data analytics Earth Engine, and the UN
and computing for research and Environment Programme World
educational purposes. In 2022, Google Conservation Monitoring Centre
Earth Engine partnered with Google (UNEP-WCMC) uses the platform
Cloud Platform to enable commercial to monitor ecological integrity
use, giving businesses and and assist nations with
biodiversity planning. Google
Earth Engine’s unique ability to
process massive geospatial
datasets at scale provides these
organizations with the insights
needed to drive environmental
impact.
 Introduct AI for Our Our Appen

Ecosystems for collaboration

Data Commons Nations’ Intergovernmental Panel on By investing early in technical project support, Google for supported startups through a 10-week
Climate Change (IPCC) to the Brazilian technologies aimed at tackling Startups Accelerator programming curriculum complete with guidance
Every moment, all around the world, Institute focuses on product design, customer from Google and industry leader
governments, organizations, and many sustainability challenges like
of Geography and Statistics to the acquisition, and leadership development mentors. These 11 participating
others U.S. Department of Commerce—and climate change through for participating founders. startups report they’ve
makes this initiatives such as Google
are generating data on topics as widely data available to policymakers, raised over $46 million in funding.
varied researchers, for Startups and Startups for • Google for Startups Accelerator:
as temperature, agricultural production, nonprofit organizations, journalists,
or students,
groundwater levels. and anyone trying to better understand people to ask questions like, “Which Sustainable Development, we Climate Change: Since launching the program
societal issues and find solutions. countries in Africa have had the greatest have the potential to move the in 2021, we’ve hosted six climate accelerators
This sustainability data, even if publicly increase in electrification?,” “What are across five continents, which have
available, is fragmented across
needle on sustainability and
Today, Data Commons is one of the the greenhouse gas emissions from collectively supported 68 startups that report
thousands of silos, in many formats and these places?,” and “How do these
positively impact our planet.
world’s largest public knowledge graphs they’ve raised over $800 million in funding.
schemas, and across a multitude of on sustainability and includes data about places compare with the United States The first two cohorts included 21 climate-
databases rendering it difficult to climate, health, food, crops, shelter, and Germany?” Google for Startups focused startups in North America. In 2023,
access and use. emissions, and more. Our APIs are open Accelerators in addition to hosting another North American
and free to anyone to build new tools. accelerator, we expanded this program to
In 2017, we started the Data Commons Our sustainability-focused accelerator
For enterprise customers, this data is Europe, in partnership with Google Cloud, as
project, which aggregates data from programs work to identify, support,
available via Data Commons well as to Latin America and the Middle East
a wide range of publicly available and scale startups that are building
on the BigQuery Analytics Hub, via and Africa.
sources, like from governments and technologies to combat climate change
Python Collaborative Notebooks, and even
nonprofits, into a unified database to • Google for Startups Accelerator: Circular
via embeddable widgets you can include and build a more circular economy.
make it more accessible and useful. Economy: This program focuses on supporting
on your article or webpage.
startups in North America and the Asia-
Google for Startups Accelerators are
Data Commons was developed by Pacific region working on challenges related
The ability to analyze a multitude of 10- week programs designed to bring
Google to organize sustainability data to advancing a more circular economy—from
publicly available datasets with Data the best of Google’s programs,
from hundreds of publicly available food waste to fashion, recycling, and reuse—
Commons products, people, and technology to
sources—from the United via advanced technologies including AI. In
has allowed it to be used to monitor and Seed to Series A technology startups.
2023, we
address a variety of sustainability In addition to mentorship and
challenges, including helping track
progress toward
the UN Sustainable Development Goals,
understanding the state of climate
progress at COP-28, and exploring how
farming regions will be impacted by
climate change.

Last year, Data Commons started

harnessing the power of AI—specifically
Data Commons is using AI to make the world’s public data more accessible and helpful Googlers working and collaborating in an office in New York City.
large language models—to create a
natural language interface that allows
 Introduct AI for Our Our Appen

Startups for Sustainable

Development
Through our Startups for
Sustainable Development
program, we’re working with
impact-driven startups using
technology to address one or
more of the UN’s 17
Sustainable Development
Goals, from
eradicating poverty and hunger
to improving healthcare and
advancing climate action.
Startups receive long-term
support to scale their impact,
including mentoring from expert
advisors, connections to funding
partners, and access to cutting-
edge research and technology.

The program now supports more

than 500 startups in over 70
countries, working with a network
of over 180 partner organizations.
 Introduct AI for Our Our Appen

and ambition of methane mitigation solutions through

Google.org grantees ideas submitted. Climate adaptation
expanded support of the Global
• As part of the Google.org Impact Challenge
was the most represented submission on Climate Innovation, $5 million will support
topic, matching an urgency to support Methane Hub. the development of WasteMAP—a platform
Google.org— Google’s philanthropy—
supports nonprofits, civic entities, and those most affected by climate change, consolidating satellite information on
other organizations that address while nearly half of the submissions landfill methane emissions. WasteMAP will
humanitarian issues worldwide. We directly targeted impacts in the Global empower local governments and nonprofits
empower these organizations with a South. to pinpoint methane sources, allowing them
unique blend of funding, in-kind to take action and mitigate emissions
donations, and technical expertise from In 2023, as part of this program, before they become hazardous.
Google employees—supporting Google. org provided $5 million to
• To expand our support of methane reduction
scalable, data-driven solutions to global World Wildlife Fund (WWF) to support
beyond the Google.org Impact Challenge,
challenges with the potential to produce ManglarIA—“AI for Mangroves.”
$3.25 million will support the UNEP
meaningful change. Through this project, WWF is using AI
International Methane Emissions
to analyze data about mangrove
Observatory to develop a new AI-based
Helping to pilot and scale AI forests in Mexico. This work will
system for detecting methane emissions.
provide conservationists around the
solutions for sustainability This system will automatically detect, alert,
world with the
and notify governments and corporations
tools and information they need to
A key focus of Google.org is about their methane emissions, empowering
conserve and restore these vital
supporting efforts to accelerate them to take action.
ecosystems in a changing climate. The
climate action. We do this by
Woodwell Climate Research Center was In 2023, Google.org announced the 15
leveraging technology and data,
another program recipient, receiving $5 projects selected for the AI for the Global
especially AI and machine learning, to
million in 2023 to develop a near real- Goals program, a $25 million commitment
support the creation of free, open-
time tracking system for Arctic to support innovative AI-driven approaches
sourced tools and datasets for the global
permafrost thaw with the Permafrost that accelerate progress on the UN SDGs.
community. In 2023, we granted more
Discovery Gateway. This project will help Among the 15 supported organizations, three
than $67 million to social impact
researchers, nonprofits, and others projects accelerated environmental progress
organizations for projects that leverage AI
forecast seasonal permafrost thaw and specifically. This effort includes DHI A/S,
to build a more sustainable world,
estimate associated carbon release. which received $2 million to develop machine
including through two open-call funding
Additionally, learning models to accurately monitor
programs launched in 2022—the
some Google employees will provide wetlands globally, in partnership with UNEP.
Google.org Impact Challenge on Climate
full-time pro-bono support for six
Innovation and AI for the Global Goals.
months through Beyond our work with the Google.org Impact
the Google.org Fellowship Program, Challenge on Climate Innovation and AI for the
The Google.org Impact Challenge on
Climate Innovation provided $30 million working alongside the Woodwell team to Global Goals, we provided an additional $30
to support breakthrough projects that help build scalable machine learning million in 2023 to other AI-based sustainability
use information, innovation, and AI to workflows and data tools. and climate action projects.
accelerate climate action. Selected
Technology will be a vital tool to help
organizations received funding to scale
meet global goals like the Global
their activities, along with access to
Methane Pledge, which aims to reduce
Google’s technical expertise to help
methane emissions by 30% by 2030. In
them maximize their impact. Over 800
2023, Google.org provided more than $8
organizations applied to the program,
million to organizations working on
and we were impressed with the quality
 Introduct AI for Our Our Appen

Supporting grassroots continue encouraging other

innovation private actors to join these
efforts. Our open-call programs
As of 2022, Google.org provided more clearly demonstrate strong
than interest from the
$9 million in funding to the Windward social impact and social
Fund to launch the Environmental enterprise sectors to use AI to
Justice Data Fund (EJDF). This fund tackle the climate crisis. Corporate
seeks to address the disproportionate philanthropy has the opportunity
climate and environmental impact on to play
frontline communities that have been an important role in catalyzing
historically underserved. Announced in innovation and impact by
2023 at COP-28, Google.org expanded providing the risk capital and
its funding for the EJDF by more than non-financial resources that
$3.5 million. This additional support is organizations
dedicated to addressing air quality, need for bold experimentation.
water quality, That’s why we partnered with the
and environmental hazards faced by WEF’s Giving to Amplify Earth
frontline communities. Action initiative and joined a
community of like-minded
The Asia-Pacific region—particularly its
corporate philanthropies
low-lying and small island countries—is
and leading foundations to help
exceptionally vulnerable to climate
launch the Corporate Philanthropy
change. Local organizations are at the
Challenge for People and Planet,
forefront
with the shared goal of mobilizing
of developing sustainable practices and
$1 billion in catalytic capital for
combating impacts, such as heat waves,
climate and nature solutions by
rising sea levels, and biodiversity loss.
2030.
Google.org’s support for the Asia Venture
Philanthropy Network’s (AVPN) APAC
Sustainability Seed Fund began in 2022
with a $3 million grant to support 13 local
organizations focused on underserved
communities in the region.
In 2023, we expanded this support with an
additional $5 million grant for the APAC
Sustainability Seed Fund 2.0. This work is
also supported by the Asian Development
Bank (ADB), which acts as the strategy
and outreach partner for the fund.

Calling on others

Google.org actively champions increased

corporate philanthropy for innovation in
the nonprofit and social impact sectors,
and we’ll

The Google.org Impact Challenge on Climate Innovation supports brea

 Introduct AI for Our Our Appen

Environmental data

Report scope • We recalculated Scope 1 and Scope 2 Ǣ In addition, we began calculating as described below. Due to rounding,
The reporting period for our
Recalculation of (market-based) emissions to include emissions generated from some of our reported values for prior
environmental data covers our fiscal previous environmental fugitive emissions from refrigerant remaining goods and services years don’t directly match the related
leakage. purchased for our operations, and Independent Accountants’ Review
year January 1, 2023,
metrics have included these Reports from those years.
• Scope 3 emissions:
through December 31, 2023. Most of our emissions in the newly
Ǣ e revised our methodology to presented
environmental data covers Alphabet Inc. Our internal recalculation policy, Greenhouse gas emissions
and which W calculate Category 1.
its subsidiaries. All reported data is emissions related to our consumer
follows guidance from the Greenhouse
global and annual unless otherwise devices manufacturing, as we We continually review emissions
Gas Protocol, informs how we apply GHG emissions reporting standards
specified. migrated from a spend-based calculation methodologies and are
updates made in the current reporting
methodology to a Life Cycle committed to implementing best GHG emissions are calculated according
period to metrics from prior reporting
Data measurement and periods—including our
Assessment-based methodology. practices.
2019 base year for our emissions Ǣ W e made improvements to the to the Greenhouse Gas Protocol standards
reduction quality
uncertainty target. Updates may include of data used in estimating our
Assurance and guidance developed by the WRI and
structural changes, calculation emissions associated with the WBCSD, including A Corporate
All reported values represent the best Accounting and Reporting Standard
methodology updates, the inclusion of manufacturing our equipment We obtain limited third-party
data available at time of publication. used in our technical (Revised Edition), Scope 2 Guidance, and
additional activity data, assurance from an independent auditor
Where actual data isn’t available, we infrastructure, that include for certain environmental metrics, Technical Guidance for Calculating Scope
improvements in the accuracy of
may use estimates. We base our emission factors or activity data, and supplier- specific data; as well as including select GHG emissions, energy, 3 Emissions (collectively, “the
estimates and methodologies on the correction of errors. To maintain the LCAs and LCA emission factors and water metrics as indicated in our Greenhouse Gas Protocol”).
historical experience, available consistency over time so that used to calculate emissions Environmental data tables. Ernst &
information, and on various other meaningful metric comparisons can related to data center Young LLP reviewed these metrics Our inventory
assumptions that we believe to be be made, it may be necessary to construction.
reasonable. recalculate
our historical metrics, including Ǣ W e included two additional within the Schedules of Select We use the operational
base Scope
Environmental approach to control

All environmental data found in this different measurements. The precision year emissions, to the extent a change the impact was deemed significant. See
3 categories to our GHG emissions
report is subject to measurement of different measurement techniques is significant. our Environmental data table endnotes
inventory—Category 3 and Category 5—
uncertainties resulting from may also vary. for more information on which metrics
and added these two categories to our
limitations inherent in the In line with our recalculation policy, in and years were recalculated. The changes reported historical inventories.
nature and the methods used for 2023 we recalculated certain primarily included:
Ǣ T o enhance transparency, we now present
determining such data. The selection of previously reported metrics, including
different but acceptable measurement “Category 1: Purchased goods and services”
our GHG emissions for our 2019 base
techniques can result in materially separately in our Environmental data tables.
year and interim years presented where
 Introduct AI for Our Our Appen

In previous years, emissions generated Indicators for the fiscal year ended
from manufacturing consumer devices December 31, 2023 and the Schedule of define our organizational
were presented in “Category 2: Capital Base Year GHG Emissions (including boundary, which means that we
goods” while emissions from our food Recalculation) for the fiscal year ended account for all emissions from
program were presented in “Other December 31, 2019. For more details, operations over which we have
categories.” Now, both are included in see our 2024 (FY2023) Independent control. We define operational
Category 1. Accountants’ Review Report. control as having
the authority to introduce and
implement operational policies
Methodology over an asset, and we report all
The below methodologies apply to our energy and emissions for
GHG emissions, as well as certain Alphabet Inc. and its subsidiaries’
other carbon, energy, water, and data centers, offices, and other
waste metrics for all years presented assets under our operational
in our Environmental data tables. control (“Global Facilities”).
These metrics have been rounded
Our Scope 1 and Scope 2
emissions include four of the
seven GHGs addressed by the
Kyoto Protocol—carbon dioxide
(CO2),
 Introduct AI for Our Our Appen

methane (CH4), nitrous oxide (N2O), and we estimate refrigerant leakage by Factors from Cross Sector Tools, the including teleworking “Category 2: Capital goods” includes
hydrofluorocarbons (HFCs). Other taking an average of GWP values from 2023 EPA Center for Corporate Climate upstream emissions generated from
GHGs, including perfluorocarbons • Category 11: Use of sold products
known refrigerants within our portfolio Leadership GHG Emission Factors Hub, goods and services we purchase for our
(PFCs), sulfur hexafluoride (SF6), and • Category 12: End-of-life
and leakage rates at Global Facilities. the 2023 DEFRA UK Government GHG operations, including manufacturing
nitrogen trifluoride (NF3), aren’t Conversion Factors, the 2023 IEA treatment of sold products
and assembly of servers and networking
included in our inventory, The emission factors used to calculate Emission Factors, the 2024 EPA eGRID equipment used in our technical
as they’re not emitted as a result of our None of the Scope 3 categories have
Scope 1 emissions include the 2017 Emission Factors, and the 2023 Climate infrastructure, as well as emissions from
operations. We report emissions both associated biogenic CO2 emissions.
WRI/WBCSD GHG Protocol Emission Registry Default Emission Factors. materials used in the construction of
in the For all
Factors from Cross Sector data centers and offices. We use the
unit of metric tons per gas (i.e., tCO ,
tCH , hybrid
2 4 Tools, the 2023 EPA Center for
Corporate The emission factors used to calculate
Scope reported Scope 3 categories, we report
tN2O, and tHFCs) and in the standardized bill, we estimate natural gas consumption period.
unit of metric tons of carbon dioxide using square footage of Global Facilities Climate Leadership GHG Emission 2 (market-based) emissions are the same as
equivalent (tCO2e), with the exception and internally developed natural gas Factors Hub, 2023 Climate Registry The emission factors used to calculate Scope 2 (location-based) with the addition of
of biogenic emissions which are intensity factors by office type, based Default Emission Factors, and the 2023 Scope 2 (location-based) emissions emission factors specific to energy attribute
reported as tCO2 only. on data from the reporting period. Where Department for Environment, Food and include the 2017 WRI/WBCSD GHG certificates. Outside of Europe, residual
actual refrigerant leakage data isn’t Rural Affairs (DEFRA) UK Government Protocol Emission emission factors aren’t available from third-
We round all reported emissions values available, GHG Conversion Factors. party sources to account for voluntary
to the nearest hundred, except for purchases, and this may result in double
Scope 3 emissions (which we round to Scope 2 GHG emissions counting between electricity consumers.
the nearest thousand) and emissions
per gas (which we Scope 2 GHG emissions are indirect Scope 3 GHG emissions
round to the nearest hundred, unless the emissions from: purchased electricity;
total is less than 50 tons in which case natural gas use and refrigerant leakage Scope 3 GHG emissions are indirect emissions
we report to the nearest one). in our leased offices; purchased steam, from other sources in our value chain, such
hot water, and chilled water from district as our suppliers, the use of our consumer
We source the global warming energy systems. The location- based devices, and business travel. In 2023, we
potentials (GWP) for each GHG from the method reflects the average carbon included two additional Scope 3 categories
IPCC Fourth Assessment Report, intensity of the electric grids where our to our GHG emissions inventory—Category 3
Appendix A: Global Warming Potentials operations are located and thus where and Category 5—and added these categories
(AR4), and IPCC Fifth Assessment Report our electricity consumption occurs. The to our reported historical inventories. We
(AR5) in select instances. market- based method incorporates our calculate our Scope 3 GHG emissions using
procurement choices, namely our the Greenhouse Gas Protocol’s Technical
Scope 1 GHG emissions renewable energy purchases via Guidance for Calculating Scope 3 Emissions
contractual mechanisms (version 1.0), in the following categories
Scope 1 GHG emissions are direct like PPAs. identified as relevant:
emissions from sources such as
company vehicles or generators at our We use actual data (such as third- • Category 1: Purchased goods and services
offices and data centers. They party invoices, monthly utility bills, or • Category 2: Capital goods
represent direct emissions from owned meter readings) to calculate Scope 2
• Category 3: Fuel- and energy-related
Global Facilities, including fuel use from emissions. Where actual data isn’t
available, we estimate electricity activities not included in Scope 1 or Scope 2
back-up generators, fuel consumption
from our operated vehicles and aircraft, consumption, natural gas • Category 4: Upstream transportation
methane consumption, and activity from district and distribution
and nitrous oxide from biogenic fuel energy systems using square footage of • Category 5: Waste generated in operations
sources, natural gas usage, and Global Facilities and internally
• Category 6: Business travel
refrigerant leakage. Where actual data developed intensity factors by office
isn’t available, for example from a utility type, based on data from the reporting • Category 7: Employee commuting,
 Introduct AI for Our Our Appen

emissions according to their minimum method, as defined by the based Scope 2 total. We use the
boundaries listed by the Greenhouse Greenhouse Gas Protocol. For average-data method, as
Gas Protocol. For certain categories, manufacturing and assembly of
we’ve also included activities which the servers and networking equipment
Greenhouse Gas Protocol deems used in our technical
optional. infrastructure, we collect supplier
GHG emissions data from our
In our Environmental data tables, we contract manufacturers, component
present certain emissions from Category suppliers, and fabless suppliers
2, Category 11, and Category 12 as an through the CDP Supply Chain
aggregated subtotal—“Other Program. These suppliers represent
categories”—for business reasons, as our key “Tier 1” manufacturing
described further below. suppliers with whom we have a
direct relationship. Per GHG
“Category 1: Purchased goods and Protocol, a “Tier 1” supplier
services” includes upstream provides or sells goods or services
emissions generated from directly to a company, while a “Tier
manufacturing consumer devices, 2” supplier provides or sells goods
our food program, and additional or services to the company’s “Tier
goods and services purchased for 1” supplier. Where actual supplier
our operations. We use the hybrid emissions data isn’t available, we
method, as defined by the estimate supplier emissions using
Greenhouse Gas Protocol. To spend data, U.S. Environmentally-
calculate full supply chain emissions Extended Input-Output industry-
generated from manufacturing average GHG intensities by
consumer commodity type, or other supplier
devices, we perform third-party- GHG data. We calculate data
verified Life Cycle Assessments (LCAs) center construction emissions by
in accordance with ISO 14040 and ISO using an LCA analysis to derive
14044. To calculate emissions construction emissions data and
generated from our food program, we then applying this to our
use LCA emission factors from WRI construction activity. In our
and annual procurement volumes Environmental data tables,
from our offices. Where actual we present emissions beyond our
procurement volume data isn’t “Tier 1” manufacturing suppliers
available, we extrapolate in “Other categories.”
calculated emissions to our other
offices using building admittances. To “Category 3: Fuel- and energy-
calculate emissions generated from the related activities not included in
remaining goods and services Scope 1 or Scope 2” includes
purchased for our operations we upstream emissions from
estimate supplier emissions using purchased fuels (e.g., natural
spend data and industry-average GHG gas, diesel, and gasoline) and
intensities by commodity type. purchased energy (i.e., electricity,
steam, heating, and cooling), as
well as emissions from
transmission and distribution
losses from purchased energy,
calculated using the market-
 Introduct AI for Our Our Appen

defined by the Greenhouse Gas collected from data from other transportation emission factors from the 2023 EPA “Category 6: Business travel” includes emissions
Protocol. We calculate upstream our providers, using emission factors providers and the weight shipped. For Center for Corporate Climate Leadership from business-related air, rail, bus, personal
emissions from purchased fuel, steam, from the 2023 Global Logistics warehousing emissions, we collect GHG Emission Factors Hub for U.S. activity vehicle, taxi, rideshare, shuttle, and rental car
heating, and cooling and emissions Emissions Council (GLEC) framework energy and refrigerant leakage data and the 2023 DEFRA UK Government travel, including emissions from relocation
from transmission and distribution of or EPA SmartWay carrier performance directly from the warehouses and GHG Conversion Factors for non-U.S. travel. We use a combination of the distance-,
steam, heating, and cooling, using data. Where logistics provider weight calculate emissions using LCA activity. fuel-, and spend-based calculation methods,
2023 DEFRA UK Government GHG and distance data isn’t available, electricity and fuel emission factors as defined by the Greenhouse Gas Protocol.
Conversion Factors. We calculate we estimate emissions based on reported from the Sphera Professional database We collect all travel data through either our
emissions from upstream electricity by 2023 and refrigerant online booking system or a third- party travel
country using the 2023 IEA Emission emission factors from the 2023 EPA agency. We calculate emissions from air, rail,
Factors. We calculate emissions from Center for Corporate Climate taxi, rideshare, non-U.S. personal vehicle, and
electricity transmission and Leadership GHG Emission Factors Hub. non-U.S. shuttle travel using 2023 DEFRA UK
distribution losses using the 2023 IEA Where actual warehouse energy data Government GHG Conversion Factors. We
Emission Factors and, for the United isn’t available, we estimate the energy calculate emissions from car rental, U.S.
States, the 2024 EPA eGRID Emission using the 2018 Commercial Buildings personal vehicle, and U.S. shuttle travel using
Factors. For upstream electricity, we Energy Consumption Survey (CBECS) emission factors from the 2023 EPA Center
calculate emissions using Scope 2 data and the warehouse square footage for Corporate Climate Leadership GHG
market-based data (i.e., by using allocated to Alphabet. Where actual Emission Factors Hub.
the remaining electricity not addressed warehouse refrigerant leakage data
by renewable energy). isn’t available, we estimate refrigerant “Category 7: Employee commuting, including
leakage based on the average leakage teleworking” includes emissions from the
“Category 4: Upstream transportation rate from available data. transport of our full-time employees between
and distribution” includes emissions their homes and their worksites
generated primarily from “Category 5: Waste generated in by passenger car (i.e., carpool, dropoff, taxi,
transportation and warehousing of operations” includes emissions from solid rideshare, or single-occupied vehicle), rail,
our consumer products and data waste generated at our offices and data bus, motorcycle, and gas-powered scooter.
center equipment. We calculate this centers that is either composted, We calculate this category’s emissions to also
category’s emissions to also include recycled, landfilled, or incinerated (with include the optional activity of teleworking. We
the or without energy recovery). We use the distance-based method, as defined by
optional activities of (1) upstream calculate this category’s emissions to the Greenhouse Gas Protocol. We survey our
emissions of transportation, and (2) also include the optional activity of employees to determine typical commuting
transportation of data center equipment waste transportation, which is and teleworking patterns and apply these
to decommission locations (which is an embedded in the emission factors we patterns to our global employee population.
optional activity under Category 5). We use. We use a combination of the We use a mode-specific commuting
use a combination of the fuel-based, waste-type-specific method and the distance obtained from the American Public
distance-based, and site-specific average-data method, as defined by Transportation Association’s 2021 Fact Book
methods, as defined by the Greenhouse the Greenhouse Gas Protocol. and the U.S. Department of Transportation’s
Gas Protocol. For transportation The waste generation data comes from 2022 National Household Travel Survey. We
emissions, we collect data a combination of data from invoices and calculate employee commuting emissions
from our logistics providers. These WTW on-site measurements. Where actual using mode-specific emission factors from
GHG emissions are calculated based on waste data isn’t available for a specific the 2023 EPA Center for Corporate Climate
fuel use or weight-distance data and facility, we estimate waste tonnage Leadership GHG Emission Factors Hub. We
routing associated with a shipment. using waste container size account for the home energy emissions
Where actual logistics provider and pickup frequency, actual waste data generated by our full-time employees
emissions data isn’t available, we from similar facilities, or historical waste working remotely by applying the estimation
estimate WTW emissions using weight data from the same facility. We use
and distance data by shipment waste type- and disposal type-specific
 Introduct AI for Our Our Appen

methodology outlined in EcoAct’s 2020 Protocol, we report biogenic emissions Our biogenic CO2 emissions are consumption using company
Homeworking Emissions white paper to separately from other Scope 1 GHG generated from our operated square footage and internally
our annual average teleworking emissions. vehicles and generators that developed intensity factors
workforce. consume biofuels. We calculate based on data from the
biogenic emissions using reporting period.
“Category 11: Use of sold products”
emission factors from the 2023
includes downstream emissions We calculate electricity purchased
EPA Center for Corporate Climate
generated by Google’s flagship from renewable sources (%) on a
Leadership GHG Emission
consumer devices sold in the reporting calendar-year basis for our global
Factors Hub.
period. Flagship consumer devices are operations by dividing the
products that can provide their main megawatt-hours of renewable
Other carbon and energy metrics
functionality without connection to electricity procured (i.e., through
another product. For example, this PPA contracts, on-site
We calculate our carbon intensity
generally doesn’t include accessories
metrics as defined by GRI
such as cases. We calculate emissions
Disclosure 305-4a-c. Carbon
from these activities using laboratory
intensity metrics are based on
power draw measurements, data on
gross global combined Scope 1 and
use patterns, common industry
Scope 2 (market- based) emissions.
assumptions on product lifetimes, and
We round reported carbon intensity
LCA electricity emission factors from
per unit of revenue and per full-
the 2023 Sphera LCA for Experts
time equivalent (FTE) employee
database. In our Environmental data
values to the nearest hundredth,
tables, we present emissions from use
and reported carbon
of sold products in “Other categories.”
intensity per MWh of energy
consumed values to the nearest ten
“Category 12: End-of-life treatment of sold
thousandth.
products” includes downstream emissions
associated with the end-of-life
We calculate total energy
treatment of Google’s flagship
consumption as defined by GRI
consumer devices sold in the reporting
Disclosure 302-1e-f. Total
period. We use the average-data
energy consumption includes
method, as defined by the Greenhouse
all fuel and natural gas
Gas Protocol. We calculate end-of-life
consumption; purchased
emissions through our LCA process,
electricity, steam, heating,
using emission factors from the 2023
cooling; and all
Sphera LCA for Experts database. Our
electricity generated on-site from
annual assessments continue to identify
renewable sources. We round
this category to be one that
reported energy consumption
doesn’t have significant life-cycle impact.
metrics to the nearest hundred.
We continue to develop programs to
extend the life of our sold products and We calculate total electricity
also to ensure efficient management of consumption as defined by GRI
end-of-life materials. In our Disclosure 302-1c(i) and 302-1f.
Environmental data tables, we present Total electricity consumption
emissions from end-of-life treatment of includes
sold products in “Other categories.” both purchased and self-generated
electricity. Where actual natural gas
Biogenic emissions or electricity consumption for
In accordance with the Greenhouse Gas facilities isn’t available, we estimate
 Introduct AI for Our Our Appen

renewable energy generation, and exceeds our load in a given hour and them in million gallons. Water withdrawal is based on actual metered or
renewable energy in the electric grids We calculate Contracted CFE as a
region, the contracted CFE consumed by invoiced data when it’s available. At offices
where our facilities are located) by the percentage of our load that’s matched We calculate water consumption by
Google where actual metered or invoiced data isn’t
total megawatt-hours of electricity with CFE on an hourly basis due entirely subtracting water discharge from water
is capped at the load; this means the CFE available, we estimate water withdrawal
consumed. This metric includes to the CFE that we purchase, and withdrawal.
percentage in this hour would be 100% using facility square footage and internally
all renewable energy purchased, without consideration of the CFE
and that “consumed” Contracted CFE developed water withdrawal intensity factors
regardless of the market in which we already on the grids where we operate.
can never exceed 100%. The “excess by office type based on data from the
consumed the renewable energy. To If Google’s total Contracted CFE
CFE” from the projects under contract reporting period.
achieve our 100% renewable energy that generate MWhs of clean electricity
match goal, we first consider both our over and above what Google consumes Water discharge is based on actual metered or
on-site renewable energy generation in a particular hour is not counted invoiced data when it’s available. Where
and the renewable electricity toward our Google CFE percentage, actual potable water discharge isn’t available,
already in the electric grids where our however it still contributes to we apply an industry-standard 90% discharge
facilities are located (using the residual decarbonization of the broader grid. flow factor to a facility’s water withdrawal
mix where to estimate water discharge. For irrigation
data is available). We then procure We calculate Consumed Grid CFE as a water, we apply a 0% discharge flow factor to
renewable energy through PPAs and percentage of our load in a given a facility’s water withdrawal to estimate water
utility renewable energy tariffs. We have market that’s matched with CFE from discharge. We apply this estimation process to
a few facilities located in geographies the grid after the application of all offices and to potable and irrigation water
where we’re not currently Contracted CFE. For hours when withdrawal at data centers used for domestic
able to source large volumes of Contracted CFE equals or exceeds our purposes (i.e., water not used for IT cooling)
renewable energy, so we make up for this load, Consumed Grid CFE is equal to where actual discharge data isn’t available.
by procuring surplus renewable energy in zero.
regions where it’s abundant. For If our Contracted CFE is less than our Water replenishment
example, by procuring larger amounts of load in an hour, then the Consumed Our water replenishment metrics are
wind energy in places like Europe, we Grid CFE is calculated by applying the based on the volumetric water benefits
compensate for our lack of renewable hourly Grid CFE percentage to the from water stewardship projects in our
energy purchases in the Asia-Pacific remaining load, and then water replenishment portfolio. We engage
region. dividing that product by the total load in our independent third-party volumetric
that hour. benefit quantification partner LimnoTech,
We calculate Google CFE as a
who applies industry standard
percentage that measures the degree For more details on how we calculate methodologies and assumptions to
to which our electricity consumption carbon-free energy percentages, please calculate two metrics following the
on a given regional grid is matched with see 24/7 Carbon-Free Energy: Volumetric Water Benefit Accounting (VWBA)
CFE on an hourly basis. This is Methodologies methodology. We calculate water replenished
calculated using both CFE under and Metrics. by estimating the total volumetric water
contract by Google (Contracted CFE)
benefits of our
as well as CFE coming from the overall
Water metrics current water replenishment portfolio
grid mix (Grid CFE). Grid CFE refers to
during the year. We calculate contracted
the percentage
Global operational water water replenishment capacity by estimating
of carbon-free energy sources
We report all operational water metrics for the annual expected volumetric water
consumed within a Regional Grid. Grid
Alphabet Inc. and its subsidiaries’ data benefits of our water replenishment project
CFE is applied to Google’s load for any
centers, offices, and other assets under portfolio
hour where Google’s Contracted CFE is
our operational control. Our reported throughout each project’s implementation
less than the load. Grid CFE values are
water metrics exclude seawater. We and respective duration.
calculated by a third party, Electricity
round water metrics to the nearest
Maps.
hundred thousand gallons and report
 Introduct AI for Our Our Appen

Once projects are funded and Google-owned and -operated data

completed, volumetric water benefits centers and Google- owned warehouses. Forward-looking
are first accounted for in the year the
project begins delivering them and in
We round reported waste diversion
metrics to the nearest one percent.
information
subsequent years—provided there’s References to information in this
reasonable evidence that the project is report should not be construed as
maintained and continues to function a characterization regarding the
as intended, which is confirmed via an materiality of such information to
annual review. If a project has multiple our financial results or our
funders, operations. While certain matters
the volumetric water benefit is adjusted discussed in this
to reflect our proportional financial report may be significant, any
contribution compared to the total significance should not be read as
project cost. The specific calculations necessarily rising to the level of
applied to each project depend on the materiality used for the purposes
project’s objectives, activities of complying with applicable
implemented, and available information. securities laws
and regulations. The information in
Waste metrics this report may contain
projections, future estimates,
We report all waste metrics for Alphabet plans, expectations, goals, and
Inc. and its subsidiaries’ data centers, other forward- looking statements.
offices, and other assets under our Forward-looking statements are
operational control. based on current expectations and
assumptions that are subject to
We calculate waste generated by certain risks and uncertainties,
quantifying solid waste generated at our which could cause our actual
offices and data centers that’s either results to differ materially from
composted, recycled, landfilled, or those reflected in the forward-
incinerated (with or without energy looking statements. Any changes
recovery). The waste generation data in methodology may result in
comes from a combination of data from material changes to our
invoices and on-site measurements. calculations and may result in the
Where actual waste data isn’t available current and previous periods,
for a specific facility, we estimate waste including our base year, to be
tonnage using waste container size and adjusted. Except as required by law,
pickup frequency, actual waste data we undertake no obligation to
from similar facilities, or historical waste correct, revise, or update any
data from the same facility. We round information included in this report.
reported waste generation metrics to
the nearest hundred.

We calculate waste diversion by

quantifying the percentage of total
waste generated that is diverted from
disposal (defined as diversion of waste
from landfills or incinerators, with or
without energy recovery). The waste
diversion rate for data centers includes
 Introduct AI for Our Our Appen

Targets data table

This table summarizes our targets and progress as of December 31, 2023, compared to the prior year. See the Environmental data tables that follow for more details.

Target
Topic Target Unit 2022 2023 Learn more
year

Achieve net-zero emissions across all of our operations and value chain by 2030

% global average
Carbon-free Run on 24/7 carbon-free energy on every grid where we operate by 64% 64% 2030 See page 35
Net-zero carbon energy 2030
carbon-free
energy

Reduce 50% of our combined Scope 1, 2 (market-based), and 3 30% 48%

Carbon reduction % reduction since 2019 2030 See page 31
absolute emissions (compared to our 2019 base year) by 2030 181 increas increas
e e

Replenish more water than we consume and help improve water quality and ecosystem health in the communities where we operate

Water stewardship
Water Replenish 120% of the freshwater volume we consume, on % freshwater replenished 6% 18% 2030 See page 46
replenishment average, across our offices and data centers by 2030

Maximize the reuse of finite resources across our operations, products, and supply chains

% of data centers at See

Data centers Achieve Zero Waste to Landfill for our global data center 29% 183 N/A See page 53
Zero Waste to page 53
operations
Landfill 182

Circular economy Offices Divert all food waste from landfill by 2025 % food waste diverted 85% 82% 2025 See page 52

Use recycled or renewable material in at least 50% of plastic % recycled/

41% 34% 2025 See page 54
Consumer used across our consumer hardware product portfolio by 2025 renewable
material
hardware
devices
Make product packaging 100% plastic-free by 2025 % plastic-free packaging 96% 99% 2025 See page 55
 Introduct AI for Our Our Appen

Environmental data tables

1 W e recalculated select metrics related to our 2019 base year emissions, including
Scope 1 emissions and Scope 2 market-based emissions, and obtained
GHG EMISSIONS
limited third-party assurance from an independent auditor for those select
metrics. For more details, see our 2024 (FY2023) Independent
Emissions inventory Unit 20191 2020 2021 2022 2023 Accountants’ Review Report and the endnotes below.
2 2019-2021 Scope 1 and 2019 Scope 2 market-based emissions were
Scope 12 tCO2e 81,900 55,800 64,100 91,200 79,400 recalculated primarily due to changes to the reporting boundary to
include fugitive emissions from refrigerant leakage.
3 Due to recalculations, the following GHG emissions metrics were adjusted:
2019 Impact of PPAs and market-based emissions factors; 2019-2022
Scope 2 (location-based) tCO2e 5,116,900 5,865,100 6,576,200 8,045,400 9,252,900 Scope 3 (total) emissions, 2019-2022 Total emissions; and all 2019-2021
Carbon intensity metrics.
Impact of PPAs and market-based emissions factors 3
tCO2e -4,281,400 -4,953,700 -4,753,100 -5,553,200 -5,829,500 4 See endnote 2 above.
5 To enhance transparency, we now present Category 1: Purchased goods
and services emissions separately. In previous years, emissions
Scope 2 (market-based)4 tCO2e 835,500 911,400 1,823,100 2,492,200 3,423,400
generated from manufacturing consumer devices were presented in
Category 2: Capital goods while emissions from our food program
were presented in “Other categories.” Now, both are included in
Category 1.
Scope 3 (Category 1: Purchased goods and services) 5
tCO2e 2,676,000 2,249,000 2,828,000 3,762,000 4,038,000
6 2019-2022 Scope 3 (Category 2: Capital goods) emissions were
recalculated primarily due to methodology updates for data center
Scope 3 (Category 2: Capital goods)6 tCO2e 2,378,000 1,886,000 1,983,000 1,645,000 1,605,000 construction emissions to use LCAs and LCA emission factors.
7 In 2023, we expanded our Scope 3 reporting boundary to include two
Scope 3 (Category 3: Fuel- and energy-related activities not additional emissions categories and recalculated 2019-2022 to include
tCO2e 381,000 462,000 686,000 916,000 1,186,000
included in Scope 1 or Scope 2)7 this additional Scope 3 activity in our reported historical inventories:
Category 3: Fuel- and energy-related activities and Category 5: Waste
generated in operations.
Scope 3 (Category 4: Upstream transportation and tCO2e 508,000 464,000 484,000 556,000 584,000
8 2019-2020 Scope 3 (Category 4: Upstream transportation and
distribution)8
distribution) emissions were recalculated primarily due to
Scope 3 (Category 5: Waste generated in operations)9 tCO2e 11,000 8,000 9,000 9,000 10,000 improvements in data accuracy.
9 See endnote 7 above.
Scope 3 (Category 6: Business travel)10 tCO2e 369,000 97,000 26,000 211,000 283,000 10 2020 Scope 3 (Category 6: Business travel) emissions were recalculated
primarily due to improvements in data accuracy.
11 2020 Scope 3 (Category 7: Employee commuting, including teleworking)
Scope 3 (Category 7: Employee commuting)11 tCO2e 173,000 116,000 111,000 151,000 113,000
emissions were recalculated primarily due to refinements to our
calculation methodology.
Scope 3 (Other categories)12, 13 tCO2e 2,258,000 2,318,000 2,761,000 2,784,000 2,993,000 12 We present certain Scope 3 emissions from Category 2, Category 11, and
Category 12 as an aggregated subtotal “Other categories” for business
Scope 3 (total)14 tCO2e 8,754,000 7,600,000 8,888,000 10,034,000 10,812,000 reasons, as described further in the Methodology section.
13 2019-2022 Scope 3 (Other categories) emissions were recalculated
primarily due to improvements to the quality of data used in estimating
our emissions beyond our “Tier 1” manufacturing suppliers of our
Total emissions: Scope 1, 2 (market-based), and 3 (total) 15
tCO2e 9,671,400 8,567,200 10,775,200 12,617,400 14,314,800 equipment used in our technical infrastructure, which include supplier-
specific data.
14 See endnote 3
above. 15 See endnote
Biogenic emissions tCO2 21,900 5,400 3,800 17,900 18,700 3 above.

Assured for 2023

Introduct AI for Our Our Appen
 Introduct AI for Our Our Appen

ENERGY 1 D ue to recalculations, the following GHG

GHG EMISSIONS emissions metrics were adjusted: 2019 Impact of
PPAs and market- based emissions factors; 2019-
Carbon intensity1 Unit 2019 2020 2021 2022 2023 Energy consumption3 Unit 2019 2020 2021 2022 2023 2022 Scope 3 (total) emissions; 2019-2022 Total
Carbon intensity per tCO2e/ Fuel MWh 366,400 181,800 205,200 374,800 301,200 emissions; and all 2019-2021 Carbon intensity
5.67 5.30 7.3 9.13 11.40
unit of revenue million USD metrics.
3
($) Purchased electricity4 MWh 12,226,200 15,125,700 18,238,400 21,685,300 25,252,600 2 “ Total renewable energy allocated” includes
renewable electricity generation from contractual
Carbon intensity per FTE Purchased heat5 MWh 150,500 124,900 119,300 219,100 278,500
tCO2e/FTE 8.48 7.62 13.0 14.76 19.02 instruments (i.e., EACs), which have been used in
employee
0 Purchased steam MWh 17,600 17,600 22,600 23,500 14,500 the calculation of Scope 2 market-based
Carbon intensity per emissions per the Greenhouse Gas Protocol
tCO2e/MWh 0.0717 0.0624 0.101 0.1155 0.1352 Purchased cooling MWh 34,900 34,800 45,600 54,800 53,000 Scope 2 Quality Criteria.
3 The following Energy consumption metrics were
On-site renewable MWh 6,300 7,200 8,800 9,600 10,700
recalculated primarily due to changes to the
electricity
reporting boundary with the inclusion of
GHG EMISSIONS 2023 Total energy consumption MWh 12,801,900 15,492,000 18,639,900 22,367,100 25,910,500 additional activity:
2019-2022 Purchased steam and 2019-2022
Scope 2 Scope 2 Purchased cooling. Due to recalculation, 2019-
GHG emissions by type Unit Scope 1
(market-based) (location-based)
2022 Total energy consumption was adjusted.
CO2 tCO2e 54,800 3,388,000 9,183,600 4 “ Purchased electricity” is electricity sourced from
an electrical grid and purchased from a local
CH4 tCO2e 200 4,800 16,900 Electricity consumption Unit 2019 2020 2021 2022 2023 electric utility company. This metric differs slightly
Total electricity consumption MWh 12,237,200 15,138,500 18,287,100 21,776,200 25,307,000 from “Total electricity consumption,” which
tCO2e 200 11,300 33,100 6
N2O
includes both purchased and self-generated
HFCs tCO2e 24,200 19,300 19,300 electricity.
5 “ Purchased heat” includes both natural gas in
Total tCO2e 79,400 3,423,400 9,252,900
leased facilities and district heat in applicable
ENERGY 2023 facilities.
CO2 tCO2 54,800 3,388,000 9,183,600 6 See endnote 4
Energy consumption Unit Renewable sources Non-renewable sources Total above. 7 See
CH4 tCH4 6 200 700
Fuel MWh 68,300 232,900 301,200 endnote 4 above. 8
N2O tN2O 1 38 100 See endnote 5 above.
Purchased electricity7 MWh 15,672,000 9,580,600 25,252,600 9 This figure represents primarily PPAs, and includes
HFCs tHFCs 15 13 13 some generation capacity from targeted
Purchased heat 8
MWh 0 278,500 278,500
renewable energy investments where we also
Purchased steam MWh 0 14,500 14,500 receive EACs.
Purchased cooling MWh 0 53,000 53,000
On-site renewable MWh 10,700 0 10,700
electricity
Scope 2 Scope 2
GHG emissions by region Unit Scope 1 Total energy consumption MWh 15,750,900 10,159,600 25,910,500
(market-based) (location-based)
North America tCO2e 54,600 1,855,700 6,864,600
Europe, Middle East, & tCO2e 11,300 59,900 752,400
Africa
Latin America tCO2e 1,200 16,200 142,000
Asia Pacific tCO2e 12,300 1,491,600 1,493,900
Global total tCO2e 79,400 3,423,400 9,252,900
North America MWh 18,535,300 11,875,400

Introduct
Europe, Middle East, & AI
MWhfor 3,547,200 Our
3,456,200 Our Appen
Africa
Latin America MWh 424,900 336,200

Asia Pacific MWh 2,799,600 4,100

Global total MWh 25,307,000 15,671,900

Assured for 2023
76

RENEWABLE ENERGY

Energy consumption Unit 2019 2020 2021 2022 2023

Renewable energy
contracts (cumulative) MW 5,400 5,700 7,200 11,600 14,9009

Renewable electricity MWh 9,715,000 12,069,200 14,109,40 16,693,600 19,089,200

(PPAs) 0
Renewable
electricity (on-site) MWh 6,300 7,200 8,800 9,600 10,700

Renewable electricity MWh 2,515,900 3,062,100 4,168,900 5,073,000 6,207,100

(grid)
Total renewable electricity
MWh 12,237,200 15,138,500 18,287,100 21,776,200 25,307,000
purchased
Electricity purchased from
% 100 100 100 100 100
renewable sources
 Introduct AI for Our Our Appen

CARBON-FREE ENERGY (CFE) 1 2022 CFE across Google offices (hourly) was
DATA CENTER GRID REGION CFE 2023 recalculated due to methodology updates.
2 2 023 is the first year we’re disclosing Regional
Grid CFE % average CFE across Google data centers. Data for
Global average CFE Unit 2019 2020 2021 2022 2023 Regional Google Contracted Consumed
Country Regional grid7 Unit
CFE CFE Grid CFE prior years is not disclosed.
CFE across Google data centers (hourly) % 61 67 66 64 64 13 3 T he eastern North America regional CFE
Indonesia Perusahaan Listrik Negara (PLN) % 13 0 13 percentage includes the following grid regions:
CFE across Google offices (hourly) % - - - 541 56 43 DUKE, IESO, Hydro- Québec, PJM, Santee Cooper,
Ireland EirGrid % 43 0 43
SOCO, and TVA.
CFE across Google data centers and offices (hourly) % - - - 64 64
Israel Israel Electric Corporation (IEC) % 5 0 5 4 T he central North America regional CFE
5
percentage includes the following grid regions:
Italy Terna % 52 0 52 52 ERCOT, MISO, and SPP.
Regional average CFE across Google data centers Unit 2019 2020 2021 2022 2023 2 5 T he western North America regional CFE
Japan Kansai Electric Power Company % 30 0 30 30 percentage includes the following grid regions: SRP,
(KEPCO)
North America - East 3
% - - - - 47 BPA, CAISO, NVE
Japan TEPCO Power Grid (TEPCO) % 16 0 16 and PACE.
North America - Central 4
% - - - - 91 16
6 A grid region (or regional grid) corresponds to the
North America - West 5
% - - - - 63 Netherland Tennet % 80 61 19
58 area over which a single entity manages the
s
operation of the electric power system and
Europe, Middle East, & Africa % - - - - 83
Poland Polskie Sieci Elektroenergetyczne % 31 0 31 31 ensures that demand and supply are finely
Latin America % - - - - 91 (PSE) balanced. In the United States, this generally
0 means the ISO or RTO in regions that have these
Asia Pacific % - - - - 12 Qatar Kahramaa % 0 0 0
regional market structures. If no such structure
Global CFE across Google data centers % - - - - 64 Saudi Arabia National Grid SA % 0 0 0 0 exists, then Google defines the grid region as
the electricity-balancing authority where our
Singapore Energy Market Authority of % 4 0 4 4
data centers are located. Outside of the United
Singapore
States,
South Eksom % 16 0 16 the grid region most often refers to the
Africa 16 geographic boundary of a country, because
DATA CENTER GRID REGION CFE 2023
most grid system operators operate at the
South Korea Power Exchange (KPX) % 35 0 35
35 national level. Certain regions that span multiple
Google Contracted Consumed Grid CFE % Korea
Country Regional grid 6
Unit countries are well interconnected and could be
CFE CFE Grid CFE Spain Red Eléctrica % 76 0 76 76
considered as one grid; however, our grid mix
Australian Energy Market Operator 33 calculations already include import and export
Australia % 33 0 33 Switzerland Swissgrid % 92 0 92 92
(AEMO), New South Wales (NSW) considerations and therefore take into account

Australian Energy Market Operator Taiwan Taiwan Power Company % 18 1 17 power

Australia (AEMO), Victoria % 40 0 40 flows from neighboring grids. In the future, we may
40 United Bonneville Power Administration % 84 0 84 18 update our definition as we work with grid
Belgium Elia % 82 35 47 States (BPA) operators to better understand how
California Independent System 84
Operador Nacional do Sistema 74 United % 55 0 55 transmission constraints or
Brazil Elétrico (ONS) % 90 0 90 Operator (CAISO)
States congestion impact CFE measurement within and
across grid regions.
Canada Hydro-Québec % 100 0 100 United Duke Energy Carolinas (DUKE) % 63 11 52 55
90 7 See endnote 6 above.
The Independent Electricity System States
Canada Operator (IESO) % 87 0 87 59
100 Electric Reliability Council of Texas
United (ERCOT) % 79 61 18
Chile Sistema Interconectado Central % 91 75 16 87 States 41
Midcontinent Independent System
Denmark Energinet % 92 43 49 United % 95 93 2 34
61 Operator (MISO)
States
Finland Fingrid % 98 77 21
86 United NV Energy (NVE) % 26 0 26 26
States
France Réseau de Transport d’Électricité % 94 0 94 92 29
(RTE) United PacifiCorp East (PACE) % 29 0 29
Germany Germany % 90 73 17 94 States 41
United PJM Interconnection (PJM) % 52 19 33
Great National Grid ESO % 92 76 16 64 48
States
Britain
CLP Power Hong Kong Limited 65 United Salt River Project (SRP) % 48 0 48 24
Hong Kong (CLP Power) % 28 0 28 States
South Carolina Public Service
India Northern India Regional Grid % 29 0 29 United % 29 6 23

India Western India Regional Grid % 14 0 14

Introduct AI for Our Our Appen

28 30

29 47

14 53
 Introduct AI for Our Our Appen

1 P UE is a standard industry ratio that compares

the amount of non-computing overhead energy
DATA CENTER ENERGY EFFICIENCY (PUE)1, 2 WASTE
(used for things like cooling and power
distribution) to the amount of energy used to
Country Location Unit 2019 2020 2021 2022 2023 Waste generation Unit 2019 2020 2021 2022 2023
power IT equipment. A PUE of 2.0 means that for
every watt of IT power, an additional
Belgium St. Ghislain PUE 1.09 1.08 1.08 1.09 1.09 Waste generated Metric tons 48,100 28,900 28,200 38,200 41,100
watt is consumed to cool and distribute power
Chile Quilicura PUE 1.09 1.08 1.09 1.09 1.09 to the IT equipment. A PUE closer to 1.0 means
Waste diversion Unit 2019 2020 2021 2022 2023 nearly all the energy is used for computing.
Denmark Fredericia PUE - - - 1.12 1.10 2 W e report individual campus PUE only for campuses with
Data center waste diversion % 90 81 78 86 783 at least twelve months of data. All reported PUE
Finland Hamina PUE 1.09 1.09 1.09 1.09 1.09 rate values are rounded to the hundredths place.
Office waste diversion rate % 71 71 64 75 77 4 3 I n 2023, we adjusted our methodology for
Ireland Dublin PUE 1.12 1.09 1.09 1.09 1.08
calculating waste generation and diversion for our
Total waste diversion rate % 77 77 77 84 78 data centers—for more details, see Reducing data
Netherlands Eemshaven PUE 1.09 1.09 1.08 1.07 1.08
center waste in the Circular economy section.
Singapore 1st facility PUE 1.15 1.14 1.13 1.13 1.13 These changes are reflected in our reported total
Data center hardware
Unit 2019 2020 2021 2022 2023 “Waste generated,” “Data center waste diversion
refurbishment and reuse
Singapore 2nd facility PUE - - - 1.21 1.19 rate,” and “Total waste diversion rate” for 2023,
Refurbished inventory used but not for prior years.
Taiwan Changhua County PUE 1.13 1.13 1.12 1.12 1.12
for server deployment, % - 10 13 21 29 4 I n 2023, we adjusted our methodology for
maintenance, and upgrades calculating waste generated and diversion for
United Berkeley County, South PUE 1.11 1.11 1.10 1.10 1.10
our offices, integrating new data sources for
States Carolina
Components resold into the Million reused furniture and recycled e-waste. These
10.0 8.2 4.9 5.0 7.0
United Council Bluffs, Iowa (1st PUE 1.12 1.11 1.12 1.12 1.11 secondary market componen changes are reflected in our reported total
States facility) ts “Waste generated,” “Office waste
United Council Bluffs, Iowa (2nd PUE 1.09 1.09 1.09 1.08 1.08 diversion rate,” and “Total waste diversion rate”
States facility) for 2023, but not for prior years.

United The Dalles, Oregon (1st PUE 1.11 1.10 1.11 1.10 1.10 Assured for 2023
States facility)
United The Dalles, Oregon (2nd PUE 1.07 1.07 1.06 1.07 1.07
States facility)
United Douglas County, Georgia PUE 1.12 1.10 1.09 1.09 1.09
States
United Henderson, Nevada PUE - - - 1.11 1.08
States
United Jackson County, Alabama PUE - - 1.13 1.12 1.10
States
United Lenoir, North Carolina PUE 1.10 1.09 1.09 1.09 1.09
States
Loudoun County, Virginia
United PUE - - 1.10 1.09 1.08
(1st facility)
States
Loudoun County, Virginia
United PUE - - 1.13 1.09 1.08
States (2nd facility)

United Mayes County, Oklahoma PUE 1.10 1.12 1.10 1.10 1.10
States
United Midlothian, Texas PUE - - - 1.16 1.13
States
United Montgomery County, PUE - - 1.10 1.11 1.10
States Tennessee

United New Albany, Ohio PUE - - - 1.14 1.10

States

United Papillion, Nebraska PUE - - - 1.13 1.09

 Introduct AI for Our Our Appen

5 2023 is the first year we’re disclosing Freshwater

withdrawal by water scarcity (per our updated framework). Data for prior years is not disclosed.
6 2022 was the first year of implementation of our water replenishment program. WATER

Global operational water Unit 2019 2020 2021 2022 2023

Water consumption Million 3,412.4 3,748.9 4,561.8 5,564.7 6,352.0

gallons
Water discharge Million 1,748.3 1,939.8 1,734.8 2,034.9 2,301.3
gallons
Water withdrawal Million 5,160.7 5,688.7 6,296.6 7,599.6 8,653.3
gallons

Freshwater withdrawal by
Unit 2019 2020 2021 2022 20235
water scarcity

Low water scarcity % - - - - 69

Medium water scarcity % - - - - 16

High water scarcity % - - - - 15

Water replenishment Unit 2019 2020 2021 20226 2023

Water replenished Million - - - 271.0 1,035.9

gallons
Contracted water
 Introduct AI for Our Our Appen

1 B ased on the average annual irrigation of golf

WATER USE BY WATER USE BY
2023 Golf course 2023 courses in the southwest U.S. of 459 acre-ft or
DATA CENTER LOCATION DATA CENTER LOCATION Golf course
equivalents around 150 million gallons. Source: “How Much
equivalents
(estimated) Water Does Golf Use and Where Does It Come
(estimated
Location Unit Withdrawal Discharge Consumption 1 Location Unit Withdrawal Discharge Consumption From?” U.S. Golf Association, November 2012.
)5
2 T reated wastewater from the Sweetwater Creek
Ashburn, VA Million gallons 57.9 3.3 54.6 Lockbourne, OH Million gallons 31.8 8.5 23.3 Sidestream Reuse Facility.
<1 <1
Potable water 57.9 Potable water 31.8 3 A ir-cooled facility; no water used for cooling.

Mayes County, OK Million gallons 1,037.1 222.0 815.1 4 Industrial water supplied by North Water.
Berkeley County, SC Million gallons 847.2 83.8 763.4
5.1 5.4 5 S ee endnote 1 above.
Potable water 847.2 Potable water 1,037.1
6 See endnote 3 above.
Middenmeer, Million gallons 7.1 2.1 5.0 7 W ater drawn from the Nimy-Blaton-Peronnes shipping
Council Bluffs, IA Million gallons 1,334.9 354.8 980.1
6.5 Netherlands <1 canal and treated on-site.
Potable water 1,334.9
Potable water 7.1 8 See endnote 3
above. 9 See endnote
The Dalles, OR Million gallons 383.7 81.3 302.4 2.0 Midlothian, TX Million gallons 164.3 28.5 135.8 <1 3 above.
Potable water 383.7
Potable water 164.3
Douglas County, GA Million gallons 418.8 73.2 345.6
Montgomery County, TN Million gallons 342.0 53.4 288.6 1.9
Potable water 34.5
Potable water 342.0
Reclaimed 384.3 2.3
wastewater2 Montreal, Canada6 Million gallons 0.04 0.03 0.01 <1
<1
Dublin, Ireland 3
Million gallons 0.6 0.5 0.1 Potable water 0.04
Potable water 0.6
New Albany, OH Million gallons 152.0 24.9 127.1 <1
Eemshaven, Million gallons 296.4 64.4 232.0 Potable water 152.0
Netherlands
Potable water 2.1 Papillion, NE Million gallons 164.2 29.5 134.7
1.5 <1
Non-potable water 4
294.3 Potable water 164.2

Quilicura, Chile Million gallons 190.7 85.5 105.2

Frankfurt, Germany Million gallons 2.2 1.8 0.4 <1
Potable water 190.7
Potable water 2.2 <1
St. Ghislain, Belgium Million gallons 348.9 104.7 244.2
Fredericia, Denmark Million gallons 27.0 6.3 20.7
Potable water 27.0 Potable water 22.3 1.6
<1
Non-potable water7 326.6
Hamina, Finland Million gallons 3.0 2.7 0.3
Potable water 3.0 <1 Sterling, VA Million gallons 81.0 25.4 55.6
<1
Potable water 81.0
Henderson, NV Million gallons 273.8 115.0 158.8 1.1
Potable water 273.8 Storey County, NV8 Million gallons 1.9 1.7 0.2
Potable water 1.9 <1
Inzai, Japan Million gallons 11.2 4.4 6.8 <1
Potable water 11.2 Sydney, Australia9 Million gallons 0.9 0.8 0.1
Potable water 0.9 <1
Jackson County, AL Million gallons 159.8 17.6 142.2 <1
Potable water 159.8 Other data center Million gallons 698.5 57.9 640.6
locations
Lancaster, OH Million gallons 15.4 7.7 7.7 Potable water 31.1 4.3
Potable water 15.4 <1 Non-potable water 296.6
Reclaimed wastewater 370.8
Leesburg, VA Million gallons 246.8 73.6 173.2
Potable water 246.8 1.2 Data centers total Million gallons 7,657.2 1,556.6 6,100.6
Potable water 5,984.6
40.7
Lenoir, NC Million gallons 358.1 21.3 336.8 Non-potable water 917.5
Potable water 358.1 2.2
Reclaimed wastewater 755.1

Assured for 2023 79

 Introduct AI for Our Our Appen

Certifications Recognitions

CERTIFICATION DETAILS

ISO 50001: In 2023, we maintained our ISO 50001 certification for all Google-owned and -operated data
Energy management centers in Europe that met our operational threshold for power usage. We were the first major
internet company to achieve a multi-site energy management system certification to ISO 50001,
which we first obtained in 2013.
ISO 14001:
Environmental
management We maintain an ISO 14001 certification for our mobile phones, computer, and tablet consumer
hardware in the United States.

As of 2021, Google is a signatory of the Climate Neutral Data Centre Pact, a pledge of data
centers in Europe to achieve climate neutrality by 2030. The Pact sets targets in five areas:
Climate Neutral Data
energy efficiency, renewables, water, circular economy, and heat recovery. In 2023, we
Centre Pact (CNDCP) successfully completed third-party verification that our five Google-owned and -operated data
centers in the EU met the five pact targets in line with the
Self-Regulatory Initiative.
EU Code of Conduct
on Data Centre Energy Below is a selection of sustainability-related recognitions received in 2023. While most
In 2023, our five Google-owned and -operated data centers in the EU became “Participants” in the focus on environmental topics exclusively, some also recognize broader achievements.
Efficiency EU Code of Conduct on Data Centre Energy Efficiency.

Leadership in Energy As of the end of 2023, over 300 Google office facilities have achieved LEED certification, CDP (A score) Gartner
CDP (included) (“strong” 5/5 score)
and Environmental including 71 with a Platinum rating and 165 with a Gold rating. In 2023 alone, we achieved LEED
Design (LEED) certification for 16 Google office facilities, including six with a Platinum rating and six with a Gold
Sustainability Magazine
rating. For a list of some of Google’s LEED- certified projects, see the U.S. Green Building (ranked #1)
Council’s project library. Dow Jones
International Living (included)
Future Institute (ILFI) In 2023, one Google building achieved ILFI certification: our Gradient Canopy building in Mountain
Technology Magazine
View, California, achieved ILFI’s LBC Materials Petal certification—the largest new construction Dow Jones
project to date to attain this certification. (included) (ranked #1)

Green Business In 2023, we achieved TRUE Zero Waste certification for 14 buildings across nine campuses, EPA Green Power Partnership
TIME
Certification Inc. (GBCI) including Platinum- level certification for seven buildings in Dublin, Ireland, and precertification for Partner Rankings (ranked #2)
(ranked #3 overall)
TRUE Zero Waste seven other buildings around the world. As of early 2024, we achieved TRUE precertification across
18 of our office campuses in Mountain View, California, which comprise more than 100 buildings. As
 Introduct AI for Our Our Appen

o ied portfolio in the world.

f

e
a
r
l
y

2
0
2
4
,

w
e

h
a
v
e

t
h
e

l
a
r
g
e
s
t

T
R
U
E

p
r
e
c
e
r
t
i
f
 Introduct AI for Our Our Appen

Endnotes
1 This calculation is based on internal data, as of May 2024.
Google’s analysis of traffic patterns before and after methodology (Reig et al., 2019). components, films, coatings, and adhesives.
2 “The Carbon Footprint of Machine Learning Training Will
recommended adjustments to traffic signals that were 14 Based on total weight of new Google Pixel and Fitbit retail 28 See endnote 14
Plateau, Then Shrink,” Computer, vol. 55, July 2022.
implemented during tests conducted in 2022 and 2023. packaging (excluding adhesive materials and required plastic above. 29 See endnote
3 According to Google’s own analysis of our more efficient
Emissions reductions estimates are modeled using a stickers) as shipped by Google. To meet the request of some 14 above.
servers, power infrastructure, and cooling systems,
Department of Energy emissions model. A single fuel-based retail partners, stickers and/or security tags are applied to 30 G oogle Trends provides access to a largely unfiltered
compared with data center industry averages based on
vehicle type is used as an approximation for all traffic, and some packaging variations and may contain plastic. sample of actual search requests made to Google, allowing
2023 data. Uptime Institute’s annual data center survey
it is not yet adjusted for local fleet mix. These data points 15 Based on total weight of new Google Pixel retail packaging us to display interest in a particular topic from around the
from 2023 noted that the primary contributor to the
are averaged from coordinated intersections, and are (excluding adhesive materials and required plastic stickers) globe or down to city- level geography. Google Trends is
flatlining of the industry average PUE is a richer
subject to variation based on existing scenarios. We expect as shipped by Google. To meet the request of some retail anonymized (no one is personally identified), categorized
geographical mix of surveyed data centers, with an
these estimates to evolve over time and look forward to partners, (determining the topic for a search query) and aggregated
increasing
sharing continued results as (grouped together).
number of data centers in the Asia, Middle East, Africa, we perform additional analysis. stickers and/or security tags are applied to some 31 B ased on Google Trends data, when comparing global
and Latin America regions. Facilities in these regions tend 9 Alphabet’s percentage of electricity purchased from packaging variations and may contain plastic. Google Search interest from 2004 through 2023.
to be smaller renewable
in capacity and located in warmer climates—both factors sources methodology is a custom calculation and is based 16 U nique, signed-in Google users that were provided 32 T his is based on a 1% sample of all signed-in users who
which typically require greater energy consumption. on a global approach. Percentage of renewable energy is information to make a more sustainable choice by at least were active on Google Search in the United States over the
4 According to the Uptime Institute’s 2023 Global Data Center calculated on a calendar-year basis, dividing the volume of one sustainable product feature. course of a year (May 1, 2023 to April 30, 2024). This
Survey, renewable electricity (in calculation only included users
the global average PUE of respondents’ data centers was actions and their effects. These factors contribute to a megawatt-hours) procured for our global operations (i.e., capacity may vary from the signed amounts based on changes
around 1.58. The Institute noted that the primary contributor range of possible outcomes, within which we report a renewable energy procured through our PPA contracts, on- during construction or project terminations.
to the flatlining of the industry average PUE is a richer central value. site renewable energy generation, and renewable energy in 12 The Google Renewable Energy Addendum applies to the electricity
geographical mix of surveyed data centers, with an 7 The estimated population covered is based on the the electric grids where our facilities are located) by the consumed by suppliers in the manufacturing of Google technical
increasing number of data centers in the forecasted flood risk area, using the WorldPop Global Project total volume of electricity consumed by our global operations. infrastructure and consumer hardware products.
Asia, Middle East, Africa, and Latin America regions. Facilities Population dataset. The numerator includes all renewable energy procured, 13 We contracted a third-party to estimate replenishment benefits
in these regions tend to be smaller in capacity and located 8 Reductions in stops estimates are based on early data points regardless of the market in which the renewable energy was using the Volumetric Water Benefit Accounting (VWBA)
in warmer climates—both factors which typically require from consumed. Additional details on Alphabet’s criteria and
greater energy consumption. methodology can be found in the “Achieving Our 100%
5 “Reduce Carbon and Costs with the Power of AI,” Boston Renewable Energy Purchasing Goal and Going Beyond”
Consulting Group, January 2021. disclosure.
6 Google uses an AI prediction model to estimate the 10 A grid region (or regional grid) corresponds to the area
expected fuel or energy consumption for each route over which a single entity manages the operation of the
option when users request driving directions. We identify electric power system and ensures that demand and supply
the route that we predict are finely balanced. In the United States, this generally
will consume the least amount of fuel or energy. If this route is means the ISO or RTO in regions that have these regional
not already the fastest one and it offers meaningful energy market structures. If no such structure exists, then Google
and fuel savings with only a small increase in driving time, defines the grid region as the electricity-balancing
we recommend it to the user. To calculate enabled emissions authority where our data centers are located. Outside of
reductions, we tally the fuel usage from the chosen fuel- the United States, the grid region most often refers to the
efficient routes and subtract it from the predicted fuel geographic
consumption that would have occurred on the fastest route boundary of a country, because most grid system operators
without fuel-efficient routing and apply adjustments for operate at the national level. Certain regions that span
factors such as: CO2e factors, fleet mix factors, well-to- multiple countries
wheels factors, and powertrain mismatch factors. We then are well interconnected and could be considered as one
input the estimated prevented emissions into the EPA’s grid; however, our grid mix calculations already include import
Greenhouse Gas Equivalencies Calculator to calculate and export considerations and therefore take into account
equivalent cars off the road for a year. The cumulative figure power flows from neighboring grids. In the future, we may
covers estimated emissions prevented after fuel-efficient update our definition as
routing was launched, from October 2021 through we work with grid operators to better understand how
December 2023, while the annual figure covers estimated transmission constraints or congestion impact CFE
emissions prevented from January 2023 through December measurement within and across grid regions.
2023. Enabled emissions reductions estimates include 11 The total GW figure represents primarily PPAs, and includes
inherent uncertainty due to factors that include the lack of some generation capacity from targeted renewable energy
primary data and precise information about real-world investments where we also receive EACs. Actual generation
 Introduct AI for Our Our Appen

17 See endnote 7 that are active more than one day over the course of
above. 18 See endnote the year. Topics related to sustainability include climate
6 above. 19 See change, sustainable food choices, sustainable
endnote 10 above. 20 consumer goods, sustainable transport and travel,
See endnote 9 above. electric vehicles, sustainable energy, home energy
21 See endnote 11 efficiency and electrification, recycling and waste
above. 22 See endnote management, and others.
15 above. 33 B ased on Google Trends data, when comparing U.S.
23 Having previously used “before” and “by” interchangeably, Google Search interest in this disaster type from 2004
we’ve refined the language of our carbon reduction target through April 2024.
to “by” 2030 to enhance clarity and improve consistency with 34 Based on global Google Trends data
our targets. This adjustment, made in consultation with the from 2023. 35 See endnote 34 above.
Exponential Roadmap Initiative, doesn’t alter our ambition, 36 Based on Google Trends data of global search
target strategy, or approach to our emissions reduction interest between January 2004–April 2024.
efforts. 37 Based on Google Trends data, when comparing
24 We consider “Zero Waste to Landfill" for our data center global Google Search interest from 2004 through
operations to mean that more than 90% of waste is diverted April 2024.
from landfill and incineration, in line with industry standards. 38 B ased on Google Trends, when comparing U.S.
For more details, see Reducing data center waste in the Google Search interest from 2004 through 2023.
Circular economy section. 39 See endnote
25 In 2023, we adjusted our methodology for calculating 34 above. 40 See
waste generated and diversion for our data centers—for endnote 5 above.
more details, see Reducing data center waste in the Circular 41 According to the Emissions Database for Global
economy section. These changes are reflected for 2023, but Atmospheric Research’s 2023 “GHG emissions of all
not for prior years. 2022 world countries” report, 2022 GHG emissions for the
performance (38%) is based on our previous methodology European Union’s 27 Member States were
and is not comparable year-over-year. Had we not made approximately 7% of global 2022 GHG emissions, falling
these methodology changes, we estimate that our reported within the 5–10% range estimated by Boston
2023 percentage of data centers to reach Zero Waste to Consulting Group.
Landfill would have been 43%, or a 5% increase compared to 42 Using satellite imagery, large-scale weather data,
our reported 2022 percentage. and flight data, we trained a contrails prediction
26 Based on total plastic weight of Google Pixel, Nest, model. For this trial, we partnered with American
Chromecast, and Fitbit products manufactured in 2023. This Airlines to integrate contrail likely zone predictions
does not include plastics in printed circuit boards, labels, into the tablets that their pilots used in flight so
cables, connectors, electronic components and modules, they could make real time adjustments in altitude to
optical components, electrostatic discharge (ESD) avoid creating contrails.
components, electromagnetic interference (EMI) components, We evaluated the model’s performance using
films, coatings, and adhesives. satellite imagery, comparing the number of contrails
27 Based on total plastic weight of Google Pixel, Nest, and produced in flights where pilots used predictions to
Chromecast products manufactured in 2022. This does not avoid contrails, to the number of contrails created in
include plastics flights where pilots didn’t use contrail predictions.
in printed circuit boards, labels, cables, connectors, For more details, see the Project Contrails website.
electronic components and modules, optical components, 43 E stimated energy savings are calculated based on
electrostatic discharge (ESD) components, electromagnetic the average percentages for heating and cooling
interference (EMI) savings found in real-world studies of the Nest
Learning Thermostat in the U.S. and U.K., and
generalized for Nest thermostat usage worldwide,
assuming user
 Introduct AI for Our Our Appen

opt-in for available energy-saving features. To calculate the

59 B ased on the average annual irrigation of golf courses in workdays in a 105 These emissions savings represent the cumulative impact of
total Nest savings, we applied the savings percentages to
the southwest U.S. of 459 acre-ft or around 150 million month. 80 See endnote PPAs and market-based emissions factors from 2012 to 2023
the actual heating and cooling hours of all Nest
gallons. Source: “How Much Water Does Golf Use and Where 8 above. (i.e., the difference between our Scope 2 location-based
thermostats in use in 2023.
Does It Come From?” U.S. Golf Association, November 81 “The Contribution of Global Aviation to Anthropogenic emissions, which don’t take into account our CFE
44 “Data Centres and Data Transmission Networks,” IEA, July
2012. Climate Forcing for 2000 to 2018,” Atmospheric procurement, and our Scope 2 market-based emissions). For
2023, accessed May 2024. This metric covers global final
60 S ee endnote 24 Environment, January 2021. Calculated using Supplementary the amounts from each year, see the Environmental data
electricity demand. Although this statistic is based on a
above. 61 See endnote data to compare the global table in the Appendix for 2019-2023, the
2022 dataset, we believe this is the best and most
16 above. 62 See warming potential (GWP100) of contrails to the total global 2019 Google Environmental Report data table for 2013-
applicable proxy for comparison to Google’s 2023 data
endnote 31 above. warming potential of the three primary aviation pollutants 2018, and the 2018 Google Environmental Report data
center energy consumption. More recent statistics include
63 This sector categorization—energy; transportation; (CO2, NOx, and contrails). table for 2012.
activity like cryptocurrency mining, which is not relevant to
buildings; forests, land use, and agriculture; and industrial— 82 See endnote 42 above. 106 “Greenhouse Gas Equivalencies Calculator,” U.S.
Google.
references the work of leading climate expert organizations, 83 “Aviation Contrail Climate Effects in the North Atlantic from Environmental Protection Agency, March 2024, accessed May
45 The IEA estimates that global data center electricity
including the IPCC’s Clsimate Change 2023 Synthesis Report, 2016 to 2021,” Atmospheric Chemistry and Physics, vol. 22, iss. 2024.
consumption is 240–340 TWh, or around 1–1.3% of global final
Project Drawdown’s Drawdown Foundations, and the U.S. 16, August 2022. 107 S ee endnote 3 above.
electricity demand (roughly 25,000 TWh). In 2023, Google’s
EPA’s Sources of Greenhouse Gas Emissions. 84 We estimated a cost efficiency range based on our trial’s 108 According to Google’s platform-neutral measurement for
data centers consumed over 24 TWh of electricity, translating
64 According to the IPCC’s Climate Change 2023 Synthesis Report, results in combination with published contrail literature. CPU resources analyzed over a five-year period from 2018–
to 7-10% of the estimated 240–340 TWh consumed by data
the energy sector (34%) and transport (15%) accounted for a There is inherent uncertainty for this cost range due to 2023.
centers globally, and less than 0.1% of the estimated 25,000
combined 49%—or roughly half—of global emissions in 2019. estimations of the impact of contrails to the climate 109 See endnote 4 above.
TWh of total global electricity demand. See endnote 44 for
65 See endnote 8 above. system. For more detail, see the Project Contrails website. 110 Energy Attribute Certificates (EACs) are tradable
more information on the IEA statistic.
66 The Solar API estimates the rooftop solar potential of 85 “Global Forest Resources Assessment 2020 - Key instruments issued to a unit of generation (generally, one
46 “Aligning Artificial Intelligence with Climate Change
buildings around the world using high-resolution, 3D models Findings,” Food and Agriculture Organization of the United MWh) which are used to aggregate and track energy
Mitigation,” Nature Climate Change, June 2022. We believe this
of individual roofs from our aerial imagery in Google Nations, 2020. attributes. Depending on the system that issues them and
continues to be an appropriate proxy for 2023 activity.
Maps. We’ve counted the number of individual buildings 86 “Land Area - Hungary,” The World Bank Data, accessed May the market where they are used, corporate buyers may
47 See endnote 44
for which we have data, and which can be queried via a 2024. 87 “Feeding the Future Global Population,” Nature purchase them bundled with or unbundled from the
above. 48 See endnote 9
lat-long in Google Maps Platform. Communications, underlying generation to secure the property rights to
above.
67 “Japan’s Greenhouse Gas Emissions Fall 2.5% in FY22/23 to January 2024. energy attributes. EACs are often interchangeably referred to
49 PUE is a standard industry ratio that compares the amount of
Record Low,” Reuters, April 2024. 88 “2023 Was the World’s Warmest Year on Record, by Far,” as Renewable Energy Certificates (RECs).
non- computing overhead energy (used for things like
68 See endnote 6 NOAA, January 2024. 111 For example, research by Princeton University, TU
cooling and power distribution) to the amount of energy used
above. 69 See endnote 89 See endnote 34 above. Berlin, and the International Energy Agency has shown
to power IT equipment. A PUE of 2.0 means that for every
43 above. 90 “NASA Clocks July 2023 as Hottest Month on Record Ever that local and hourly
watt of IT power, an additional watt is consumed to cool and
70 The enabled emissions reductions are calculated based on Since 1880,” NASA, August 2023. carbon-free energy matching goals have a greater impact
distribute power to the IT equipment. A PUE closer to 1.0
these energy savings, applying standard emission factors 91 See endnote 7 above. on grid- level decarbonization than the prevailing approach of
means nearly all the energy is used for computing.
for fossil fuels, and using U.S. EPA AVERT marginal 92 “Spreading Like Wildfire: the Rising Threat of Extraordinary 100% annual renewable energy matching.
50 See endnote 3 above.
emissions for the 95% of 112 C arbon-free energy is any type of electricity generation
that doesn’t directly emit carbon dioxide, including (but not
limited to) solar,
51 S ee endnote 2 electricity savings that occur in the U.S., with an adjusted Landscape Fires,” UNEP, February wind, geothermal, hydropower, and nuclear. Sustainable biomass
above. value for 2022.
52 This estimate is based on our internal analysis comparing the new Blackwell GPUs (4 megawatts) to the power the 5% of electricity savings outside the U.S. GraphCast predictions. Note that these results are shown for
the BFLOAT16 / INT8 model step time ratio measured on required with older GPUs (15 megawatts) and found the new 71 “Sixth Assessment Report: Chapter 9: Buildings,” IPCC, pressure level PL1000, where a majority of the wind farms are
the MLPerf 3.1 GPT-3 175B model. The results (11,798ms / GPUs use roughly 25% of the power (or 75% less) of the 2022. 72 See endnote 43 above. located.
8,431ms = 139%) can be interpreted as a 39% speed older ones. 73 See endnote 70 above. 77 “Sixth Assessment Report: Chapter 10: Transport,” IPCC, 2022.
improvement and, in turn, training efficiency. 56 This calculation is based on internal data, as of March 74 “Solar PV,” IEA, July 2023, accessed May 78 See endnote 6 above.
53 “Gemini 1.5: Unlocking Multimodal Understanding Across 31, 2024. 57 See endnote 49 above. 2024. 75 See endnote 66 above. 79 This is based on estimated daily car rides at the intersections where
Millions of Tokens of Context,” Google, 2024. 58 See endnote 4 above. 76 Tapestry refined the baseline GraphCast models, Green Light has been implemented, multiplied by the average
54 “TPU v4: An Optically Reconfigurable Supercomputer for outperforming HRES by 15.1% with a 96-hour lead time, and
Machine Learning with Hardware Support for by 18.4% with a
Embeddings,” In Proceedings of the 50th Annual 120-hour (5-day) lead-time (both location and wind
International Symposium on Computer Architecture, June weighted). HRES is the weather model of the ECMWF
2023. (European Centre for Medium-Range Weather Forecasts)
55 According to “What Nvidia’s new Blackwell chip says about and is considered the state- of-the art model for weather
AI’s carbon footprint problem,” Nvidia compared the total predictions. Testing was performed by generating the
power needed to train the latest ultra-large AI models using forecasts for 2021 and comparing the accuracy with
 Introduct AI for Our Our Appen

93 S ee endnote 31 above. and carbon capture and storage (CCS) are special
94 “Canada’s Record-Breaking Wildfires in 2023: A Fiery Wake-up cases considered on a case-by-case basis, but are
Call,” Natural Resources Canada, May 2024, accessed May often also considered carbon-free energy sources.
2024. 113 For more details about our energy policy and
95 “Is climate change increasing the risk of disasters?,” WWF, ecosystem advocacy work, see the Public policy and
April 2024. advocacy section in the Appendix.
96 See endnote 7 114 The total GW figure includes generation capacity
above. 97 See endnote from targeted renewable energy investments.
7 above. Actual amounts funded and generation capacity
98 “Climate Change Impacts on Air Quality,” U.S. developed may vary from the amounts anticipated
Environmental Protection Agency, April 2024, accessed when the agreements were signed.
May 2024. 115 See endnote 11 above.
99 Includes only Google-owned and -operated data center 116 S olar panels equivalency based on “How Much
locations. Some of our locations have more than one data Power is 1 Gigawatt?,” Office of Energy Efficiency
center campus. and Renewable Energy, August 2023, accessed May
100 See endnote 11 2024.
above. 101 S ee endnote 117 This estimated spend is based on contracts
9 above. 102 See signed to purchase clean energy for our
endnote 23 above. operations, and includes some targeted renewable
103 Although 2020 was the most recent emissions inventory energy investments where we also receive EACs.
available at the time the target was set, 2020 was deemed Actual spend may vary from these estimates
to not be representative of a typical year, because based on changes in
operations were impacted by the COVID-19 pandemic. The renewable electricity output from operational
next most recent year with representative data, 2019, was projects, the number of contracts signed, project
selected as the base year. terminations, and energy market prices
104 See endnote 9 above.
 Introduct AI for Our Our Appen

118 See endnote 11 above.

the reduction percentage from the base year, the total local context. electromagnetic interference (EMI) components, films,
119 Contracted CFE (%) represents the percentage of our load
base year emissions, the target year, and the target 142 See endnote 59 coatings, and adhesives. For products launched in 2022 and
that’s matched with hourly electricity generation from clean
status. above. 143 S ee 2023: Pixel, Nest, and Chromecast devices are made with
energy projects contracted by Google, without
133 S ee endnote 12 above. endnote 141 above. recycled materials ranging between 10% and 60% based on
consideration of the CFE already on the grids where we
134 This third-party estimate is based on a 2022 whole-building 144 Expected stormwater collection is calculated based on respective product weights.
operate. For more details, see the Other carbon and energy
life cycle assessment, which includes requirements based on the roof areas and estimated peak flow rates into the 162 F or products launched in 2023: Fitbit devices are made
metrics section in the Appendix.
the LEED v4 Building Life Cycle Impact Reduction credit stormwater cistern, weather data for expected rainfall, with recycled materials of at least 5% based on product
120 Grid CFE (%) refers to the percentage of carbon-free
requirements, as well as full cradle-to-grave life cycle stages and expected use for non- potable water demands. The weight.
energy sources consumed within a Regional Grid. Grid CFE
over a 75-year analysis period. The analysis includes the estimated reduction in municipal water use is calculated 163 B ased on total plastic weight of Google Pixel, Nest,
is applied to Google’s load for any hour where Google’s
foundation, structure, enclosure, and interior materials. based on expected stormwater captured and reused Chromecast, and Fitbit products manufactured in 2023. This
Contracted CFE is less than the load. These values are
135 C arbon footprint reduction claim based on third-party- onsite compared to total expected site water does not include plastics in printed circuit boards, labels,
calculated by a third party.
verified life cycle assessment. Recycled aluminum in the consumption per year. cables, connectors, electronic components and modules,
121 See endnote 10 above.
enclosures is at least 9% of applicable product based on 145 See endnote 13 optical components, electrostatic discharge (ESD)
122 Our CFE percentage measures the degree to which our
weight. This 100% recycled content claim excludes Pixel 5a. above. 146 See endnote components, electromagnetic interference (EMI) components,
electricity consumption on a given regional grid is
136 Based on data collected from shuttle commuting trips in 13 above. 147 See films, coatings, and adhesives.
matched with CFE on an hourly basis. This is calculated
the San Francisco Bay Area, as compared to a scenario in endnote 13 above. 164 Based on total plastic weight of Google Pixel, Nest, and
using both CFE under contract by Google as well as CFE
which these employees had each used a private, fuel-based 148 Freshwater consumption from Other Bets is excluded Chromecast products manufactured in 2022. This does not
coming from the overall grid mix. CFE coming from the
car for commuting. Assumptions include average Googler from our water replenishment target. Non-Google include plastics
overall grid mix is based on data obtained from a third
commute distance, miles per gallon assumptions from the businesses under Alphabet Inc. are referred to collectively in printed circuit boards, labels, cables, connectors,
party. For more information, see our 2021 white paper,
Bureau of Transportation, and emissions per gallon of fuel as Other Bets, and Other Bets operate as independent electronic components and modules, optical components,
“24/7 Carbon-Free Energy: Methodologies and Metrics.”
assumptions from the Greenhouse Gas Protocol. Equivalency companies. electrostatic discharge (ESD) components, electromagnetic
123 The central North America regional CFE percentage
estimate is based on workdays in the year, using data from 149 S ee endnote 31 interference (EMI) components, films, coatings, and
includes the following grid regions: ERCOT, MISO, and SPP.
“Greenhouse Gas Equivalencies Calculator,” U.S. above. 150 See endnote adhesives.
124 R egional Google CFE across Google data centers refers to
Environmental Protection Agency, March 2024, accessed 31 above. 165 B ased on total weight of new Google Pixel and Fitbit
the percentage of carbon-free energy sources consumed by
May 2024. 151 P ercent reduction in food waste per Googler was calculated products launched and manufactured in 2023, including U.S.
Google’s data centers within a given global region, following
137 “Climate Change 2022: Mitigation of Climate Change as food waste generated in kitchens and cafes at Google’s configuration of in-box accessories.
the same approach for calculating Google CFE (%) metrics.
Summary for Policymakers,” IPCC, 2022. global offices per unique building badge swipes, against a 2019 166 S ee endnote 163
For more details, see the Other carbon and energy metrics
138 In 2007, we set a goal to achieve operational carbon base year. above. 167 S ee endnote
section in the Appendix.
neutrality, and we met this goal every year from 2007 to 152 This figure represents the cumulative hardware 164 above. 168 See
125 T he eastern North America regional CFE percentage
2022 via three steps. First, we worked to reduce our total components resold from 2015 to 2023. For the amounts endnote 163 above. 169
includes the following grid regions: DUKE, IESO, Hydro-
energy consumption by pursuing energy efficiency from each year, see the Environmental data table in the S ee endnote 14 above.
Québec, PJM, Santee Cooper, SOCO, and TVA.
initiatives. Next, we matched 100% of Appendix for 2019-2023 and the 2019 Google 170 S ee endnote 15
126 C ontracted CFE (%) represents the percentage of our load above.
the global electricity consumption of our operations with Environmental Report data table for 2015-2018.
that’s matched with hourly electricity generation from clean
purchases of renewable energy. Finally, we purchased high- 153 Updates for at least seven years from when the device first 171 B ased on packaging comparisons of Pixel 8 and Pixel 8 Pro
energy projects contracted by Google, without
quality carbon avoidance credits to address any remaining became available on the Google Store in the U.S. See phones vs. a first generation Pixel phone. Based on total
consideration of the CFE already on the grids where we
emissions, for example through the capture and destruction g.co/pixel/updates for details. weight of new Google Pixel retail packaging (excluding
operate. For more details, see the Other carbon and energy
of highly potent greenhouse gasses like methane. Going 154 I n 2023, we adjusted our methodology for calculating adhesive materials and required plastic stickers) as shipped
metrics section in the Appendix.
forward, we continue to support a range of efforts to curb waste generated and diversion for our offices, integrating by Google. To meet the request of some retail partners,
127 Consumed Grid CFE (%) measures the percentage of our stickers and/or security tags are applied to some packaging
highly potent greenhouse gasses, for example through our new data sources for reused furniture and recycled e-waste.
load in a given market that’s matched with CFE from the variations and may contain plastic.
work on MethaneSAT and grants to key organizations 155 S ee endnote 151
grid after the application of contracted CFE. For more
working on methane mitigation solutions. For more details, above. 156 S ee endnote 172 S ee endnote 14
details, see the
see Methane leaks in the Our products section and the 31 above. 157 See above. 173 See endnote
Other carbon and energy metrics section in the 14 above. 174 S ee
Google.org grantees section in the Appendix. endnote 152 above.
Appendix. 128 Residual grid mix removes from the grid endnote 153 above.
158 Our consumer hardware products include Pixel, Nest,
electricity mix the
Chromecast, and Fitbit devices. 175 Learn more about Google’s recycling program on our site.
159 U L110 and IEEE 1680.1 are multi-attribute, consensus-based 176 This includes all final assembly manufacturing sites globally
for Google consumer hardware products with at least one
year of data.
proportion of renewable energy contracted to other 139 W e define freshwater as naturally occurring water from sustainability standards for mobile phones and for Learn more about the UL 2799 Zero Waste to Landfill
parties that surface or computers validation.
have claims to that electricity through EACs, and avoids to methodology updates. based on our CFE methodology. 132 P er CDP’s guidance documents, reported targets are considered
double- counting. 130 T his estimate is based on anticipated contracted energy 131 “How the Tech Sector Can Drive Renewable Energy “structured” if they include the scope(s) of emissions covered,
129 2022 CFE across Google offices (hourly) was recalculated due from the projects that were terminated and then applied Opportunities in APAC,” ENGIE Impact, accessed May 2024.
 Introduct AI for Our Our Appen

groundwater sources that isn’t salty, and is suitable for and displays, respectively. Google uses a third party to 177 T he percentage of FSC-certified wood used at
consumption if clean or processed. Freshwater excludes validate conformance and independently certify to these Gradient Canopy is calculated based on material
seawater and reclaimed wastewater. standards. See UL Spot Database. costs for new wood purchased for this project
140 S ee endnote 13 above. 160 Doesn’t include third-party products such as the Nest x Yale during construction.
141 To define water scarcity levels, Google assesses Lock. 161 F or products launched in 2020 and 2021: Nest, Pixel, and 178 See endnote 177 above.
operational water risks for data centers and offices. For data Chromecast devices are made with recycled plastic ranging 179 B ased on Google Trends data, when comparing
centers, we assess water scarcity and depletion by applying between 9% and 68% based on weight of plastic used in each global Google Search interest in 2022–2023 to 2020–
our Data Center Water Risk Framework, and assign a low, respective product launched during this timeframe. The 2021.
medium, or high water scarcity level. For our office following items are excluded: plastics in printed circuit 180 See endnote
operations, we assess water scarcity using the WRI boards, labels, cables, connectors, electronic components and 31 above. 181
Aqueduct Water Risk Atlas and the WWF Water Risk Filter, and modules, See endnote 23
where appropriate we adjust the assigned level of water optical components, electrostatic discharge (ESD) above. 182 S ee
scarcity based on components, endnote 24
above. 183 S ee
endnote 25
above.
 Introduct AI for Our Our Appen

Glossary Photo details

AI: artificial intelligence LBC: Living Building Challenge On the cover:
Google Earth image of Nome, Alaska.
CDP: Formerly known as the Carbon LCA: life cycle assessment
What’s inside (page 1):
Disclosure Project
LEED: Leadership in Energy and Interior of Google’s Bay View office. Photo credit: Iwan Baan.
CFE: carbon-free energy Environmental Design Introduction (page 2):
Golden Hills wind farm in California (43 MW for Google)
CO2e: carbon dioxide m: meter
A letter from our Chief Sustainability Officer and our
equivalent CPU: central ML: machine learning Senior Vice President of Learning and Sustainability (page 3):
Photo of Kate Brandt by Vaughn Ridley/Sportsfile for Web Summit via Getty Images
processing unit EAC: energy MW: megawatt
AI for sustainability (page 9):
attribute certificate MWh: megawatt-hour EDF’s new satellite, MethaneSAT, will map, measure, and track methane from oil and gas with
unprecedented precision, offering a comprehensive view of methane emissions. Photo credit:
EIE: Environmental Insights Explorer NGO: non-governmental MethaneSAT LLC.
EPA: U.S. Environmental Protection organization PPA: power Our products (page 14):
By building AI models on the emissions profile of different vehicle types, fuel-efficient routing in
Agency EV: electric vehicle purchase agreement PUE: power Google Maps analyzes traffic, terrain, and the vehicle’s engine to find the most efficient route.
ft: foot usage effectiveness Mitigation (page 16):
The Environmental Insights Explorer estimates the technical solar potential of all buildings in a region
FSC: Forest Stewardship SDGs: Sustainable Development Goals based on total sunshine exposure, weather patterns, and roof dimensions.

Council GHG: greenhouse gas tCO2e: metric tons of carbon Adaptation and resilience (page 23):
Governments, aid organizations, and individuals can use Flood Hub to take timely action and
dioxide equivalent
GPU: graphics processing unit prepare for riverine floods, seeing locally relevant flood data and forecasts up to 7 days in
TPU: Tensor Processing advance.
GT: gigaton
Our operations (page 27):
Unit TWh: terawatt-hour
GW: gigawatt Our Bay View campus, as seen from across its stormwater retention pond. Photo credit: Iwan Baan.

GWP: global warming WBCSD: World Business Council Net-zero carbon (page 29):
for Sustainable Development El Romero solar farm in Chile (80 MW for Google)
potential IEA: International
WRI: World Resources Institute Water stewardship (page 42):
Energy Agency A rainwater retention pond outside our data center in Berkeley County, South Carolina.
UNEP: United Nations
Circular economy (page 49):
IPCC: Intergovernmental Panel on Environment Programme The restored glue-laminated wooden interior of our Spruce Goose hangar, with the sculpture
Climate Change nicknamed the “Ghost of the Goose” in view. Photo credit: Connie Zhou.
UNFCC: United Nations
ISO: International Organization for Framework Convention on Nature and biodiversity (page 56):
Climate Change The design of the Charleston Retention Basin on our Mountain View Campus allows visitors to
Standardization
experience, learn about, and enjoy the expanded riparian habitat.
X: Alphabet’s “moonshot factory”
kWh: kilowatt-hour Appendix (page 60):
Norther Offshore wind farm in Belgium (92 MW for Google)
Environmental
Report
2024

Want to learn more about our latest efforts?

To stay up to date on our news and

progress, see our Sustainability blog.

Publication date: July 2024