Microsoft said it will rent AI infrastructure from Mistral in a multi-billion-dollar deal and add the French company's models to products including Copilot for enterprise customers.
The tie-up is aimed at surging European demand for cloud and AI services while easing concerns about reliance on U.S. tech, with Microsoft stressing customers can keep data and operations in Europe.
Mistral said the agreement should speed financing for Mistral Compute, its infrastructure arm, as it invests 4 billion euros in data centers in France and Europe.
The three-year-old company, with 1,000 staff, is targeting 200 megawatts of capacity by 2027 and 1 gigawatt by 2030, though that remains far smaller than U.S. hyperscalers' buildouts.
The deal extends a broader industry shift in which AI developers also sell compute capacity, as scarce high-end chips and booming model demand reshape the cloud market.
With a global chip shortage, how will Europe's ambitious AI infrastructure plans avoid becoming empty data centers?
Can a US tech giant truly deliver European AI sovereignty, or is this just a strategic illusion?
Microsoft and Mistral’s $Multi-Billion AI Partnership: Building 200MW of European Compute for Digital Sovereignty by 2027
Overview
Microsoft and French AI startup Mistral have formed a multi-billion-dollar commercial partnership to boost AI capabilities across Europe. Instead of taking an equity stake, Microsoft is funding Mistral’s European GPU infrastructure, which will use NVIDIA’s latest Vera Rubin chips. The goal is to build 200 megawatts of AI compute capacity by 2027, with Mistral scaling its advanced AI models for Microsoft Azure customers, especially those needing European data residency. This collaboration aims to address strict data control and regulatory needs, supporting industries like healthcare and finance, and strengthening Europe’s digital resilience and economic competitiveness.