Mistral AI to Build 1 Gigawatt of European Compute Capacity by 2030

French artificial intelligence firm Mistral AI has announced ambitious plans to construct one gigawatt (GW) of AI compute capacity in Europe by 2030. This expansion includes regional inference endpoints, allowing customers to choose whether their AI workloads run in Europe or the United States.

The company is also introducing a "Priority Tier" service backed by an uptime guarantee for mission-critical deployments. Furthermore, Mistral has formed a coalition of European enterprises that have made multi-year compute commitments. These commitments are set to underpin 200 megawatts (MW) of infrastructure across Europe by the end of 2027, and a full gigawatt by 2030.

Mistral will also begin hosting third-party open models on its platform, starting with GLM-5.2 from China-based Zhipu.ai. This strategic move positions Mistral as a provider of European AI infrastructure, combining regional control with contractual reliability.

The expansion of this capacity requires substantial investment. Estimates suggest that a one-gigawatt AI data center could incur up to $38 billion in upfront capital expenditure. Mistral's co-founder and CTO, Timothée Lacroix, highlighted the necessity of securing capital through long-term customer commitments to finance the infrastructure build-out.

The company has also introduced "European Compute Units" (ECUs). These are multi-year commitments that lock customers into using capacity without an option for early termination, designed to secure financing for large infrastructure projects. Several major European industrial companies, including Amadeus and ASML, are among the first to sign these agreements.