CAPE - Superpowering the edge!
Bielefeld, June 2025
The CAPE “European Open Compute Architecture for Powerful Edge” project, funded by the European Commission, is an ambitious initiative to strengthen Europe’s sovereignty over challenging edge computing and AI applications. CAPE aims to harmonize the edge-cloud continuum by defining standards for European data centers. Based on these standards and open blueprints, fully composable server platforms are developed, accompanied by an open-source software stack. The integration of hardware and software innovations will boost productivity in developing edge applications, which are crucial to the backbone of many European SMEs.
CAPE’s expertise and pillars
CAPE was kicked off in December 2024 for a three-year duration. Jointly coordinated by Bielefeld University (administrative) and Hiro-MicroDataCenters (technical), the project unites expertise from seven small and medium-sized companies (SMEs), four academic partners and one standardization consortium. The project’s name – “European Open Compute Architecture for Powerful Edge” – reflects the project’s multifaceted innovations, which are built on four key pillars:
- Scalable edge computing hardware platforms
- Hardware platforms eHPS “embedded High performance Server” and EMDC “Embedded Micro Data Center” (Bielefeld University (AG Kognitronik und Sensorik), Germany, and PCB Design Ltd., Hungary)
- Modular FPGA and GPUSoC hardware accelerators for COM-HPC (paraXent GmbH, Germany)
- Building integration and cooling (CAD-Terv Kft., Hungary, and associated partner EPFL, Switzerland)
- Low-level server management (Christmann Informationstechnik + Medien GmbH & Co. KG, Germany)
- Unified software stack leveraging the edge-cloud continuum
- Cognitive Framework (Hiro-MicroDataCenters B.V., Netherlands)
- Multi-Cluster Orchestration (Ryax Technologies, France)
- Infrastructure From Code (associated partner, University of St. Gallen, Switzerland)
- Standardization & Open Source
- COM-HPC is the base of CAPE’s edge computing hardware platforms (associated partner, PICMG, USA)
- Open blueprint for CAPE’s hardware and software components
- All partners follow an “open source first” mindset
- Three challenging use cases
- Smart Grid Anomaly Detection (Independent Power Transmission Operator S.A., Greece)
- Deep Learning at the edge (Fraunhofer Gesellschaft ITWM, Germany)
- Satellite Payload Data Processing (PCB Design Ltd., Hungary)
Meeting Europe’s Computing Demands with a Sovereign and Scalable Cloud-Edge Infrastructure
The ever-increasing demand for computing power in all areas of life poses new challenges for operators of both SMEs and large data centers. In the EU, with its strict demands for data security and privacy, on-premises edge computing is increasingly essential. Ideally, applications should run cost-effectively on local edge servers, while peak loads can be handled by scalable cloud solutions. CAPE responds to this need by developing two dynamically composable, high-performance edge server platforms built from modular COM-HPC components in line with open standards.
The modular computing nodes support multiple heterogeneous target architectures like x86, RISC-V, GPUSoC, and FPGAs, delivering performance acceleration and boosting energy efficiency. A standout innovation is CAPE’s efficient cooling technology, which not only delivers superior Power Usage Effectiveness (PUE) and passive cooling capabilities but also opens the door to waste heat reuse—enabling these systems to run effectively outside traditional data centers in living or production environments.
Shaping the Cloud-Edge Continuum with Sustainable Software Stacks
CAPE introduces a comprehensive software stack to simplify managing and deploying various applications. A unified management interface is developed for commodity servers as well as CAPE hardware platforms. A multi-cluster orchestrator dynamically allocates, and provisions resources based on the application requirements and available resources. In combination with an AI-assisted cognitive framework for automated deployment, applications can be dynamically transferred between local edge clusters, or parts can be offloaded to the cloud. This is particularly advantageous for mitigating peak loads for typical SME workloads using cloud on-demand resources.
More Information
Our progress will be regularly published on www.cape-project.eu and LinkedIn.
