Less than three months after raising a $415 million Series A led by Andreessen Horowitz, Mistral AI, a 10-month-old developer of large language models, has partnered with Microsoft.
The Paris-based startup is looking to challenge peers such as Anthropic and OpenAI—a startup Microsoft backed a year ago with a $10 billion investment—underscoring the tech giant’s far-reaching AI ambitions.
Microsoft made a €15 million (about $16.3 million) investment in the company, which will convert to equity in Mistral’s next funding round, according to a Microsoft spokesperson. Mistral didn’t respond to a request for comment.
The partnership further complicates the web of relationships between Big Tech and generative AI startups. In addition to OpenAI and Mistral, Microsoft has also backed Anthropic. And Mistral’s backers include cloud software giant Salesforce and chipmaker Nvidia.
Mistral will get access to Microsoft’s Azure cloud platform and provide its model on the service. The only other company providing large language models to Azure clients is OpenAI.
Under the deal, the two companies will collaborate on research focusing on customer-specific models to court European clients.
Mistral also unveiled its latest model, Mistral Large, as well as a chatbot competitor to OpenAI’s ChatGPT called Le Chat. The startup was founded by former Meta and Google DeepMind employees and raised a $113 million seed round led by Lightspeed in June. It was valued at around $2 billion after its Series A funding in December.
Led by CEO Arthur Mensch, Mistral has sought to challenge OpenAI with a different approach, initially releasing its Mistral Small model as open-source. These types of models have been gaining traction among startups and their VC backers. The startup has since released other models with a closed licensing system similar to OpenAI, and tweaking its lineup, though it still offers and updates its open-source large language models.
Big Tech investments have recently drew antitrust scrutiny from the Federal Trade Commission, which on Jan. 25 opened an inquiry into deals made between OpenAI and Microsoft, as well as Anthropic, Alphabet and Amazon. Days after the FTC opened its inquiry, the European Union’s member states endorsed the proposed AI Act, a landmark framework of AI rules that would ban some applications and impose strict limits on others.
Mistral co-founder Cedric O has lobbied against restricting foundational models, saying in an October interview with Sifted that the AI Act “could kill Mistral.”
Update: This article was updated with financial details on Microsoft’s investment in Mistral. (Feb. 27, 2024)
Arthur Mensch, co-founder and CEO of Mistral AI
WPA Pool/Getty Images
Learn more about our editorial standards.
