France Advances Into Exascale Era With New AMD Powered Alice Recoque Supercomputer

The project is funded by EuroHPC JU and the Jules Verne Consortium

The system targets major scientific, industrial and AI workloads across Europe

France Launches Its First Exascale System

France is moving into Exascale computing with Alice Recoque, a next generation system built by AMD and Eviden. The supercomputer is designed to support high performance computing and artificial intelligence and will serve as an AI Factory. It is overseen by Grand equipement national de calcul intensif and operated by Commissariat a l energie atomique et aux energies alternatives.

Alice Recoque is the first Exascale system located in France and the second in Europe. The project carries a total cost of $641 million funded by EuroHPC JU through the Digital Europe Programme and the Jules Verne Consortium. France, the Netherlands and Greece are formal contributors through GENCI, CEA, SURF and GRNET.

“We are committed to enabling the next generation of innovation across AI and HPC,” said Dan McNamara, senior vice president and general manager, Compute & Enterprise AI, AMD. “The Alice Recoque supercomputer represents a major step forward for European sovereign AI, uniting national ambition, regional collaboration, and AMD’s high-performance and AI compute technologies. Through our continued collaboration with EuroHPC JU, the Jules Verne Consortium, and Eviden, we are proud to support Europe’s scientific and industrial leadership with a platform purpose-built for scale, efficiency and discovery.”

The build reflects a collaboration between AMD and Eviden, the Atos Group product brand responsible for advanced computing. The companies aim to support research scale AI and HPC workloads with a complete hardware and software stack.

A System Built To Address European Scientific And Industrial Challenges

The Alice Recoque platform is designed to take on major European challenges through its mix of simulations, data analysis and AI acceleration. The system integrates hardware, system software and established use cases to form a full computing lifecycle.

AMD contributes next generation EPYC CPUs, codenamed Venice, along with AMD Instinct MI430X GPUs. The GPUs are part of the MI400 Series and are engineered for sovereign AI and scientific computing. AMD FPGAs and Eviden network technology complete the configuration inside the BullSequana XH3500 platform. DDN storage supports the data layer.

The system aims to advance climate modeling, next generation materials research and energy innovation. Digital twins for personalized medicine and new European AI model development are also core targets. The hardware has been configured for both throughput and adaptability to address diverse workloads.

Efficiency As A Core Design Requirement

Alice Recoque consists of 94 racks and is expected to rank among Europe’s top systems for double precision HPC work. Its memory capabilities are designed to support deeper insights and faster simulations with room for future scientific growth.

Eviden states that the architecture uses about 25% fewer racks and components than other Exascale class systems. It also targets up to 50% improved energy efficiency per GPU. This aligns the project with European goals for greener large scale computing.

Emmanuel Le Roux, Group SVP, Global Head of Advanced Computing and AI, Eviden at Atos said, “Alice Recoque represents another critical step toward Europe’s digital future, defined by a sovereignty, sustainability and scientific excellence. As a catalyst for scientific and industrial breakthroughs, from climate modeling and healthcare to advanced materials and AI innovation, it will empower researchers and industries across Europe. Born from a shared European vision, this AI Factory reflects on what we can achieve collectively toward a common goal. Eviden is fully dedicated to its success, bringing deep expertise, a collaborative spirit, and a long-term dedication to responsible technological leadership.”

Each AMD Instinct MI430X GPU supports advanced AI data types including FP4 and FP8. The GPUs integrate 432 gigabytes of HBM4 memory and 19.6 Tbps of bandwidth. These specifications position the system to deliver significant performance per unit of energy.

Eviden pairs these components with integrated hardware and smart software to improve application efficiency. Real time monitoring and energy optimization stem from the company’s Argos intelligent software. Direct Liquid Cooling using warm water is used across all components within each rack to maintain sustainability and performance.