Mistral, which has been billed as a European rival to OpenAI, is in talks to raise its third round in under a year.
Launched in 2023, the Paris-based startup — which builds AI models — is in talks to raise up to $2 billion in funding, according to three sources familiar with the deal. It is not known which investors are taking part in the deal, but the startup has been in discussions with UK, France, and US-based funds, one source told Business Insider.
The upcoming deal could catapult Mistral’s valuation to around $5 billion to $7 billion, two sources familiar with the discussions told Business Insider. The Information first reported this week that the startup is in talks to raise the round at a $5 billion valuation.
In March, Bloomberg first reported that Mistral was in discussion with SoftBank to raise funding.
The deal is not yet finalized and the round size, valuation figures, and participants could still change.
Mistral did not reply to Business Insider’s request for comment.
The startup last raised $415 million from heavyweights such as Andreessen Horowitz, Salesforce, Elad Gil, and General Catalyst in December 2023, touting a valuation of around $2 billion.
Investor appetite for generative AI skyrocketed in 2023, with enterprises and venture capitalists pouring billions into the likes of Anthropic, OpenAI, and Aleph Alpha at massive valuations. Big Tech companies such as Microsoft, Amazon, and Google have also been frontrunners in backing these AI juggernauts, with Microsoft leading OpenAI’s $10 billion round last February.
Helmed by two former engineers at Google’s DeepMind and Meta, the startup cinched a $113 million seed round just a few weeks after its launch.
Mistral then released its first large language model, Mistral 7B, in September 2023, and its latest version, 8X22B, in April 2024. It also announced a multi-year partnership with Microsoft that will allow Microsoft to tap into the startup’s AI models, and offer them to its enterprise customers.
AI heavyweights such as OpenAI and Stability AI have faced a slew of copyright concerns as they’ve trained their models on unlicensed datasets. Mistral trains its AI models on publicly available data.
