AI cloud platform Fluidstack and Eclairion, a French maker of modular, high-density data centers, have partnered to build what the companies said is Europe’s largest GPU supercomputer that they will deliver in 2025 for Mistral AI, the French AI startup.
The 40 MW facility, located at Eclairion’s Bruyères-le-Châtel, Essonne (40km south from Paris), facility will provide compute infrastructure, “reinforcing France’s leadership in sovereign AI capabilities and energy-efficient AI infrastructure,” the companies said.
This multi-year partnership brings together Fluidstack’s focus on large-scale AI clusters, Eclairion’s data center design, and Mistral’s AI research. The cluster will be deployed in phases, leveraging advanced GPUs to support demands of foundation model training and AI innovation.
“This partnership strengthens France’s position as a leading AI compute hub. With Mistral AI’s innovation, Fluidstack’s rapid deployment, and Eclairion’s infrastructure, we are building Europe’s most advanced AI cluster,” said César Maklary, Co-founder and President of Fluidstack.
The companies they have plans for expansion of the cluster to 100MW+. They said their is available connection to the decarbonized French energy grid, making the Bruyères-le-Châtel facility a good location for hosting large-scale AI infrastructure.
This announcement follows Fluidstack’s recent commitment to building a standalone 1-gigawatt AI supercomputer in France with an initial investment of €10 billion. This giga-cluster was announced earlier this month at the AI Action Summit in Paris under the leadership of President Emmanuel Macron.
Fluidstack said it is cementing its role as a key player in France and Europe’s AI expansion, “delivering next-generation compute infrastructure for AI research and development. The Mistral AI cluster is part of this broader vision, ensuring France and Europe remain competitive on the global AI stage alongside the United States and China.”
Project highlights:
- Multi-Year Deployment – The first tranche of the cluster will be operational by Summer 2025, with phased rollouts continuing throughout the year.
- Massive Scale – Designed to scale beyond 18,000 GPUs, with potential for further expansion.
- Next-Gen AI Hardware – Powered by the best-performing GPUs, optimized for AI model training and inference.
- Decarbonized Energy – The facility will run on 100% decarbonized electricity provided by nuclear and renewable energy sources, ensuring sustainability.
- Sovereign AI & Security – Aligns with France’s goal of developing local AI infrastructure and becoming a leader in AI alongside the US and China.
