Skip to content 
In January 2025, dbt Labs announced the acquisition of SDF Labs, a company known for its deep expertise in advanced SQL parsing and execution. This was a major leap forward for the analytics engineering community.
This technology now powers the new dbt Fusion engine, which represents the most substantial upgrade to dbt Core’s internals since inception.
As the data transformation landscape continues to evolve, dbt Labs is leading the charge with a next-generation engine designed to dramatically enhance performance, scalability, and developer experience.
In this article, we will explore what makes dbt Fusion a game-changer, highlight its standout features, and share practical steps to help your team prepare for a smooth and successful transition.
What is the new dbt Fusion Engine?
The new dbt engine marks a major evolution in analytics engineering, integrating SDF Labs’ advanced SQL comprehension technology directly into dbt Core. This upgrade introduces a powerful multi-dialect SQL compiler and software toolset that transforms how dbt parses, understands, and executes SQL. The result is dramatically faster model compilation, more accurate error detection, and significantly improved lineage tracking.
dbt Fusion is a paradigm shift in SQL processing. Built on a high-performance Rust-based architecture, it enables dbt to deliver faster, more intuitive feedback, allowing data teams to write, test, and deploy models at a pace orders of magnitude faster than before. Beyond speed, this upgrade enhances developer productivity, improves governance, and helps reduce data platform costs. At the same time, it reinforces dbt Labs’ commitment to building a more reliable and scalable transformation framework for modern data teams.
dbt Fusion key features and improvements
Deep SQL comprehension
SDF enables dbt to parse and understand SQL code natively, allowing for real-time error detection, intelligent autocomplete suggestions, and more accurate model validations.
Enhanced performance
With its Rust-based architecture, SDF offers significantly faster model compilation and execution times, streamlining the development process and reducing time-to-insight.
Improved lineage and metadata
It provides detailed table and column-level lineage, offering better visibility into data transformations and aiding in compliance and governance efforts.
Cost efficiency
By emulating data platforms locally, SDF allows for code validation without consuming warehouse resources, leading to cost savings and more efficient resource utilisation.
Development with dbt before Fusion
In the traditional dbt workflow, SQL queries are treated as plain text and passed directly to the data warehouse without validation. The warehouse is responsible for checking the correctness of the query. If there's an error, it sends that feedback back to dbt, which then displays it to the user.
As a result, data engineers must rely on the warehouse for feedback, which means:
- Slower development cycles due to round trips to the warehouse;
- Higher compute usage, hence less efficient spend.
Development with Fusion
Since SDF is a multi-dialect SQL compiler, it can validate SQL queries locally, without having to send them to the data warehouse. This means engineers receive immediate feedback during development, allowing them to catch errors earlier. This capability is available in Visual Studio Code for even faster developer feedback experience.
Key benefits include:
- Instant visibility into code issues and validation errors;
- Reduced reliance on warehouse compute, saving time and resources;
- Local development without the need for an online warehouse.
Lineage
Fusion introduces advanced column-level lineage that maps how data flows through models with remarkable precision, also available inside Visual Studio Code.
It categorises dependencies into three distinct types: copy, transform, and inspect.
- Copy dependencies occur when a column’s values are passed downstream without modification, even if filtered or aggregated (marked by copy);
- Transform dependencies involve changes to the data, such as aggregations or function applications, where the downstream column is derived from one or more upstream columns (marked by mod);
- Inspect dependencies indicate that an upstream column has been referenced (e.g. in WHERE, GROUP BY, or JOIN clauses) to influence logic, but has not directly contributed values to the downstream column (marked by scan).
By capturing these nuanced relationships, Fusion enables highly detailed lineage graphs that help teams understand data transformations at a granular level, improving debugging, governance, and impact analysis.
Fusion represents the future of the dbt, delivering faster performance, more efficient use of data platform resources, and improved lineage tracking, all without any changes to the existing code in your dbt project.
By upgrading to the new engine, teams can immediately benefit from these enhancements while maintaining full compatibility with their current workflows.
Transition to SDF
Transitioning to the new engine requires careful planning. Here are steps to ensure a smooth migration:
- Upgrade to dbt Core v1.10 or later:
Ensure your projects are running on the latest version of dbt Core to take advantage of SDF's features and receive the latest updates and fixes. For dbt Cloud users, make sure your environment is set to the latest release track to stay compatible with the new engine. - Address deprecation warnings:
Review and resolve any deprecation warnings in your project. dbt Core v1.10 introduces warnings for behaviours that will be unsupported in future versions. Addressing these now will prevent issues post-migration.
Benefits for data teams
The integration of SDF into dbt is more than a technical upgrade; it is a strategic enhancement that empowers data teams to work more efficiently and effectively. By providing deeper insights, faster performance, and improved governance capabilities, SDF positions dbt as a central component in modern data stacks.
Licence
dbt Core remains open-source with no licence change to consider. It will keep receiving new versions moving forward; you can find the roadmap post for ongoing investment in dbt Core here.
The new dbt Fusion Engine is a blend of open source, proprietary, and source-available components. The source-available portions are licensed under the Elastic Version 2 (ELv2) Licence.
Under ELv2, users are free to use the code and its binaries, provided they follow three key rules:
- You may not offer the software as a managed service to others;
- You may not bypass or disable any licence key mechanisms or features protected by them.
- You may not remove or hide any licensing, copyright, or attribution notices.
The ELv2 licence is designed to prevent direct commercial competition using the Fusion codebase. This model supports dbt Labs’ commitment to building a sustainable business while continuing to invest in innovation. Read more on ELv2.
Getting started with dbt Fusion
The introduction of dbt Fusion marks a significant milestone in dbt's evolution, offering a host of features designed to meet the growing demands of data teams. By understanding its capabilities and preparing accordingly, organisations can harness the full potential of this powerful engine.
For a more detailed overview and official guidance, refer to dbt Labs' blog post "How to Get Ready for the New dbt Engine".
Stay tuned for our upcoming detailed migration guide and cost-saving calculator to help you estimate the impact Fusion could have on your workflows and warehouse spend.
Visit the Infinite Lambda blog for more insights from the data and AI world. Explore our Case Studies to see how we leverage cutting-edge technology to empower modern organisations and data teams.
