Microsoft and Mistral AI expanded their artificial intelligence partnership on Tuesday, with a multibillion-dollar commitment from Microsoft to help the French AI company build data centers in Europe.
The deal is the latest attempt by both companies to reach a greater number of European and enterprise AI clients, as the two go up against rival AI labs and other U.S. tech giants.
"Distribution is king," said Mistral Chief Executive and Co-founder Arthur Mensch, adding that Microsoft's heft would facilitate the AI company's expansion with businesses in Europe and across the world.
Mistral has touted the importance of Europe's independence from U.S. technology companies. At the same time, it has been ramping up its push for enterprise clients -- an area where U.S. behemoth Microsoft offers immense reach.
For Microsoft, working with Mistral is part of the European digital commitments it made last year, in which it pledged to increase data-center capacity in the European Union. Its latest move comes as the European Commission, the EU's executive arm, is considering classifying Microsoft's Azure cloud platform as a gatekeeper under the bloc's Digital Markets Act -- a designation that requires companies to take extra steps to ensure they aren't stifling competition.
"Our partnership with Mistral is becoming a core pillar of our broader European efforts," Brad Smith, Microsoft's vice chair and president, said in an interview. "This is a moment when European customers are especially focused on resilient supply of AI."
The partnership includes a "multibillion-dollar" agreement from Microsoft -- which it declined to specify -- to support Mistral's data center build-out in Europe. As the company builds a "GPU fleet" in Europe, some of those AI chips will be used by Microsoft to serve its customers, said Mensch, referring to the data centers that house graphics processing units.
Microsoft expects to "get a return" when customers use its cloud platform Azure, said Smith, while Mistral will "get a return" when customers use its AI models.
Also as part of the deal, two of Mistral's top AI models -- Medium 3.5 and OCR 4 -- will be available in Microsoft's products, including Microsoft Foundry and Microsoft Copilot Studio. Mistral's models will additionally be available on Azure Local, a Microsoft infrastructure offering that lets customers run their information-technology systems in the cloud or their own data centers. Smith said that option would allow customers to be "disconnected" from the cloud.
"This responds to what we're hearing from customers and the market," Smith said. "There is a heightened need for customers to have more choice and more control, and that's what this gives them."
This isn't Microsoft's first investment in Mistral. The two companies, which began working together in 2023, are targeting a similar segment of European and global business customers that are in regulated sectors such as financial services, manufacturing and healthcare.
There's already some overlap: Mistral's Mensch said two-thirds of the French AI company's customer base works with Microsoft. In February, Mistral struck an agreement with Accenture that made the Dublin-headquartered technology consulting group its latest high-profile client.
According to Mensch and Smith, those customers are seeking a combination of greater AI model choice -- in other words, from outside the U.S. -- and the ability to dictate where and how their infrastructure is run.
"This means that customers can rely on Mistral models. They don't need a model from Anthropic or OpenAI or another American supplier," Smith said. "They've got the benefit of our technology, which obviously comes from the United States, but by running it in a disconnected way."
Arun Chandrasekaran, an analyst focused on AI at market research and IT consulting firm Gartner, said the partnership helps Microsoft better serve European customers with a buzzy concept called "sovereign AI." The goal of sovereign AI -- which is especially important to Europe -- is to build out a localized computing infrastructure, control the data used and generated by AI, while also raising domestic capabilities in areas such as AI model development and chipmaking.
Microsoft, which helped usher in the AI boom, is now part of a growing cohort of voices criticizing the concentration of power among a handful of AI companies like OpenAI and Anthropic. CEO Satya Nadella recently made clear that the tech giant is seeking to steer the AI race away from a future dictated and controlled by frontier model-builders.
At the same time, however, Microsoft is one of OpenAI's oldest partners and has invested billions in building the company. Microsoft also reached a multibillion-dollar agreement with Anthropic last year.
Microsoft is under pressure -- it has trailed its peers in developing its own AI systems. In the latter half of 2025, Copilot subscribers increasingly preferred other options, such as Google's Gemini, according to market-research firm Recon Analytics.
Mistral has similarly warned of Europe's reliance on U.S. tech companies. It has become increasingly vocal about the need to develop homegrown, European superintelligence, and was valued last year at around $14 billion. It has promised to reach more than $1 billion in revenue this year -- a fraction of the size of its Silicon Valley rivals.
Write to Belle Lin at belle.lin@wsj.com
(END) Dow Jones Newswires
July 21, 2026 08:30 ET (12:30 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments