Country
France
AI data center dossier
Country
France
Operator
MGX, Bpifrance, Mistral AI, NVIDIA
Energy
Nuclear
Known capacity
1.4 GW
Evidence profile
Readiness reflects whether the record has citations, narrative context, structured power data, coordinates, and at least one dated milestone.
Readiness
100%
2 citations linked
48.8566, 2.3522
1 dated field available
HTML, JSON, and GeoJSON all available
Machine-readable outputs
Canonical record surfaces for audit, programmatic use, and direct citation.
MGX, Bpifrance, Mistral AI, and NVIDIA announced a landmark joint venture at France's Choose France investment summit in 2025 to develop Europe's largest AI campus near Paris, with a planned capacity of 1.4 gigawatts (1,400 MW) — a facility that would dwarf every existing European AI data center by a factor of three or more when complete.
The project's structure reflects a new model for sovereign AI infrastructure: MGX (the Abu Dhabi-based technology investment firm) provides Gulf capital, Bpifrance (France's state investment bank) anchors EU sovereign interest, Mistral AI (France's leading open-weight model company) drives AI model development, and NVIDIA supplies next-generation GPU infrastructure. The campus will deliver exascale-class training capability for European-origin AI models, explicitly designed to reduce European dependence on US AI infrastructure.
France is uniquely positioned to host Europe's largest AI campus: the country's nuclear-dominated electricity grid (approximately 70% nuclear) provides among the lowest-carbon, most reliable, and cheapest industrial power in Europe — a critical advantage for 24/7 AI training workloads. EDF (Électricité de France) and RTE (the French transmission system operator) have secured grid access agreements for the campus.
The initiative aligns with France's Stratégie Nationale pour l'Intelligence Artificielle and the European AI Act's provisions for sovereign AI capacity. When operational, the 1.4 GW campus would give France — and by extension Europe — the compute sovereign capacity to train frontier AI models without reliance on US cloud hyperscalers, a strategic priority emphasized by both the French government and European Commission. First phases are expected to open in 2027, with full capacity by 2030.
No obvious coverage gaps detected in the current structured record.
Other tracked AI data centers within 300 km of this location.
Structured analysis covering this facility's operator and market context.