The CAPE project develops a computer architecture for efficient Edge-Cloud.
Locally deployed powerful edge-cloud Infrastructure is needed to support AI-driven environments with networks of autonomous devices to maintain their context and individual and shared states. This allows the devices to work together in federation towards individual and shared goals. To scale sustainably, this infrastructure must be provided as a service and built on composable hardware foundations. Compute, memory, storage, and accelerators must dynamically self-configure to maximise utilisation, minimise waste, and adapt capacity and heterogeneity to local needs delivering cloud-class performance through decentralised, fine-grained deployment.
The EU-funded CAPE (European Open Compute Architecture for Powerful Edge) project addresses this gap by establishing edge micro data centres as a new, composable building block of the Edge–Cloud Continuum. CAPE combines open hardware, open-source software, and open standards to enable flexible, efficient, and sovereign edge computing across Europe.
Edge hardware platforms: Composable by design
CAPE rethinks edge servers as pools of dynamically composable resources rather than fixed-function systems. The platform is designed as autonomous, fine-grained edge micro data centres that can operate independently while federating with neighbouring nodes to form a distributed edge-cloud fabric. Compute, memory, storage, and accelerators can be flexibly combined to match workload requirements, significantly improving utilisation and reducing overprovisioning.
The project develops two complementary open hardware platforms based on COM-HPC technology: the embedded High-Performance Server (eHPS) and the Embedded Micro Data Center (EMDC). These platforms support heterogeneous compute nodes and target a wide range of edge scenarios, from industrial on-premise deployments to telecom environments.
Composable hardware is enabled through PCIe switching combined with Compute Express Link (CXL), which provides cache-coherent connections between processors, accelerators, and pooled memory. This allows resources to be dynamically pooled and reallocated at runtime, enabling each autonomous edge unit to adapt to fluctuating AI workloads while maximising hardware utilisation.
Energy efficiency is a core design objective. By maximising hardware utilisation and reducing stranded resources across distributed deployments, CAPE minimises overprovisioning and unnecessary energy consumption. Passive cooling, waste heat recovery, and building-level integration further reduce the environmental footprint of embedded edge infrastructure.
Software stack: Orchestrating the Edge-Cloud Continuum
To fully exploit hardware composability, CAPE develops an open-source software stack that hides infrastructure complexity while preserving control and transparency. The stack combines Kubernetes-based multi-cluster orchestration with AI-assisted automation. Orchestration preserves local autonomy, ensuring that each edge unit can operate independently while participating in federated coordination across the distributed infrastructure.
A central innovation is Infrastructure from Code (IfC), which derives infrastructure definitions directly from application code, reducing development effort compared to traditional Infrastructure as Code approaches. Applications can be deployed seamlessly across CAPE edge servers and external cloud environments, enabling elastic scaling and workload migration.
Unified management is provided by openMPMC, an open multi-platform management controller offering web-based operation and Redfish-compatible APIs for configuration, monitoring, and dynamic resource composition.
Standardisation and openness
CAPE is fully committed to open standards and open-source development. By building on technologies such as COM-HPC and CXL and actively contributing to European and international initiatives, the project avoids proprietary lock-in and fosters a sustainable ecosystem for edge computing. This openness directly supports Europe’s goals for digital sovereignty across the Edge–Cloud Continuum via the Open Federated Edge-Cloud Infrastructure Alliance.
Use cases: Impact in real-world environments
CAPE validates its architecture through three representative use cases. In smart grids, edge servers enable real-time monitoring and anomaly detection at substations. For deep learning at the edge, heterogeneous accelerators and CXL-based memory pooling support efficient AI inference under strict power constraints. In satellite payload data processing, CAPE enables near-real-time analytics through dynamic resource composition across edge and cloud.
Outlook
CAPE demonstrates how composability, open standards, and intelligent orchestration can transform edge computing into a flexible and energy-efficient foundation for future digital services. By tightly integrating hardware and software innovation, the project lays the groundwork for scalable, sovereign edge infrastructure in Europe.
Disclaimer
The CAPE project has received funding from the European
Union’s Horizon Europe research and innovation programme under grant agreement No 101189899, and was co-funded by the Swiss National Science Foundation (SNSF, No. 200429), and by Armasuisse Science and Technology.
Please Note: This is a Commercial Profile
This work is licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International.
Please note, this article will also appear in the 25th edition of our quarterly publication.
