AEGIS

Europe’s Digital Decade will deploy large fleets of Edge Datacenters (EDCs) by 2030, just as Europe’s electricity grid struggles to integrate variable renewables. The fundamental disconnect is that EDCs operate as passive loads: physically connected to the grid but blind to its real-time needs. This is due to two barriers: a lack of internal controls to safely coordinate fragmented subsystems (IT, cooling, batteries) to provide the verifiable performance the grid requires; and a lack of fleet coordination mechanisms, as energy markets require minimum bids that smaller EDCs cannot meet, with no trusted way to aggregate them into a unified resource. The main scientific objective of AEGIS is to transform EDCs from passive liabilities into grid-interactive assets by developing a Software-Defined Platform for EDC Flexibility. This platform is architected in three layers. The foundational cyber-physical layer provides runtime-verified control to unlock safe, quantifiable flexibility. The intelligence layer uses carbon-aware forecasting and control, enabling AI workloads to dynamically adapt their behavior and resource consumption in response to real-time grid signals. The orchestration layer enables trusted, fleet-scale operations using privacy-preserving mechanisms to create a unified grid resource. Our ambitious objectives require researchers with a unique combination of interdisciplinary and intersectoral skills. AEGIS’s 15 Doctoral Candidates will receive integrated training across key areas (computer science, power systems, control engineering, machine learning, and distributed systems) necessary to realize the potential of these technologies, moving between academic and industrial environments to bridge the gap between the energy and computing domains. AEGIS will provide a new generation of experts capable of driving future developments in sustainable, grid-interactive edge datacenters across Europe.