Empowering Edge Computing Through Composability

The CAPE project is dedicated to developing a computer architecture for efficient Edge-Cloud solutions. Local edge-cloud infrastructure is crucial for supporting AI-driven environments with networks of autonomous devices. These devices need to maintain their context and individual and shared states to work together towards common goals. To ensure sustainable scalability, this infrastructure must be provided as a service and built on composable hardware foundations. This involves dynamically configuring resources like compute, memory, storage, and accelerators to maximize utilization, minimize waste, and adapt to local needs for cloud-class performance through decentralized deployment.

The EU-funded CAPE project aims to address this need by establishing edge micro data centers as a new building block of the Edge-Cloud Continuum. By combining open hardware, open-source software, and open standards, CAPE enables flexible and efficient edge computing across Europe.

CAPE reimagines edge servers as pools of dynamically composable resources rather than fixed-function systems. These autonomous edge micro data centers can operate independently while federating with neighboring nodes to create a distributed edge-cloud fabric. By flexibly combining compute, memory, storage, and accelerators to match workload requirements, CAPE significantly improves resource utilization and reduces overprovisioning.

The project is developing two open hardware platforms, the embedded High-Performance Server (eHPS) and the Embedded Micro Data Center (EMDC), based on COM-HPC technology. These platforms support heterogeneous compute nodes and cater to various edge scenarios, from industrial on-premise deployments to telecom environments.

Composable hardware is made possible 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 maximizing hardware utilization.

Energy efficiency is a core design focus for CAPE. By maximizing hardware utilization and reducing stranded resources across distributed deployments, CAPE minimizes overprovisioning and unnecessary energy consumption. Passive cooling, waste heat recovery, and building-level integration further reduce the environmental impact of embedded edge infrastructure.

To fully leverage hardware composability, CAPE is developing an open-source software stack that simplifies infrastructure complexity while maintaining control and transparency. The stack combines Kubernetes-based multi-cluster orchestration with AI-assisted automation to enable seamless deployment of applications across edge servers and external cloud environments.

A key innovation in the software stack is Infrastructure from Code (IfC), which derives infrastructure definitions directly from application code, streamlining development compared to traditional approaches. 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.

CAPE is committed to open standards and open-source development, building on technologies like COM-HPC and CXL. By actively contributing to European and international initiatives, the project avoids proprietary lock-in and fosters a sustainable ecosystem for edge computing. This approach supports Europe’s goals for digital sovereignty across the Edge-Cloud Continuum through the Open Federated Edge-Cloud Infrastructure Alliance.

The CAPE project validates its architecture through three key use cases: smart grids, edge AI, and satellite data processing. In smart grids, edge servers enable real-time monitoring and anomaly detection at substations. For edge AI, heterogeneous accelerators and CXL-based memory pooling support efficient AI inference under power constraints. In satellite data processing, CAPE enables near-real-time analytics through dynamic resource composition across edge and cloud.

Overall, 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 integrating hardware and software innovations, the project paves the way for scalable, sovereign edge infrastructure in Europe.

Please note, the CAPE project has received funding from the European Union’s Horizon Europe research and innovation program under grant agreement No 101189899, co-funded by the Swiss National Science Foundation (SNSF, No. 200429), and by Armasuisse Science and Technology.