Microsoft (MS) teamed up with French artificial intelligence (AI) company Mistral AI (hereafter Mistral) to expand cooperation on AI infrastructure in Europe. The strategy is to target European corporate customers by applying the data centers and AI models Mistral is building in France to MS's cloud platform and enterprise AI services.
MS signed a multibillion-dollar cooperation deal with Mistral on the 21st (local time), according to Reuters. Under the agreement, MS will use the AI compute infrastructure Mistral is building in Europe, and Mistral will broaden the sales channels for its infrastructure and AI models. However, new equity investment in Mistral was not included in the deal.
With this deal, Azure customers can develop AI software using compute resources at Mistral's AI data centers built in France. Because they can use local European infrastructure instead of relying only on infrastructure controlled by a U.S. company, it is seen as targeting demand in industries where the location of data storage and processing and regulatory compliance are critical. Reuters said the agreement allows MS to expand compute resources in Europe while giving customers in tightly regulated sectors such as finance, the public sector, and healthcare the option to process data within Europe.
Mistral will supply the Generative AI model "Medium 3.5" and the document and image recognition model "OCR 4" to MS's AI development platform, Azure AI Foundry. Medium 3.5 can also be used in Copilot Studio, where corporations build AI agents. In particular, Azure Local customers, who consolidation their own data centers with MS services, will also be given the option to run Mistral's open models directly.
The two companies pursued the partnership as demand for "sovereign AI" grows in Europe. Sovereign AI is the concept of controlling and operating data, AI models, and compute infrastructure within national or regional regulatory frameworks. After the United States temporarily paused overseas access to two advanced models from U.S. AI company Anthropic last month, calls grew in Europe to build independent AI infrastructure that is not swayed by external policy shifts.
Mistral is regarded as Europe's leading AI company competing with OpenAI and Anthropic. It is building AI data centers centered on France and expanding into manufacturing, financial services, and defense. It also set a plan to increase AI compute capacity in Europe to 1 gigawatt (GW) by 2030.
MS President Brad Smith said the partnership will help European customers use AI services on local infrastructure. Mistral CEO Arthur Mensch also said the deal would serve as a catalyst supporting the company's AI infrastructure expansion strategy. The industry says the partnership is meaningful in that it gives Mistral a base to secure stable corporate customers and expand data center investment, while providing MS with local compute infrastructure to meet Europe's sovereign AI demand.
However, even if Europe boosts the self-reliance of its AI infrastructure, it will likely be difficult to sever ties completely with U.S. corporations for core technologies. Nvidia's AI Semiconductor designed in the United States is used in the data centers Mistral is building. As a result, there is an outlook that Europe will, for the time being, expand regional control over data, AI models, and data centers while maintaining a structure of cooperating with U.S. corporations in core AI Semiconductor fields.