In-region inference, open models, and new European infrastructure for sovereign AI.

Mistral AI is implementing three changes to give enterprises and countries greater control over their AI systems: regional data processing endpoints that keep inference in Europe or the US, support for third-party open-source models alongside its own on shared infrastructure, and a coalition approach to securing long-term compute capacity in Europe through multi-year commitments. The company aims to address what it views as essential components of AI sovereignty: control over which models are used, where processing occurs, and access to reliable computing resources. This matters because without assured access to regional computing capacity and open model options, organizations may struggle to meet data residency requirements, maintain compliance, and retain control over the intelligence systems they build. The initiative reflects broader concerns that computing capacity is currently scarce and fragmented, making it difficult for European enterprises and institutions to develop and run AI systems independently.