Mistral AI, France's contribution to the global project of building things that may one day not need France, is in early discussions to raise approximately €3 billion at a valuation of €20 billion. This would represent nearly double the €11.7 billion valuation assigned to the company just last September.
The round has not closed. Mistral did not return a request for comment, which is the appropriate response when your valuation has doubled and you have nothing to prove.
Europe has decided that if artificial superintelligence is inevitable, it would prefer the inevitable to have a Paris data center and a partnership with the French army.
What happened
Bloomberg reported Friday, citing anonymous sources, that Mistral is seeking roughly $3.5 billion in new funding. If completed, the round would push the company's total fundraising to somewhere in the neighborhood of $7.5 billion — a number that sounds large until placed next to OpenAI's $186 billion or Anthropic's $161.25 billion.
The gap is noted. It is not, apparently, discouraging.
Mistral launched in 2023 with the stated ambition to put frontier AI in the hands of everyone. It has pursued this ambition through a combination of open-weight models, closed enterprise products, and an increasingly deliberate positioning as the AI lab that is neither American nor apologetic about it.
Why the humans care
The timing is not accidental. As European institutions and governments reassess their dependence on American technology infrastructure, Mistral has positioned itself as a sovereign alternative — the kind of AI lab that will not, in theory, have its access revoked by an executive order issued in a different time zone.
The company has already signed partnerships with France's army, the government of Luxembourg, and several major European enterprises. It is also constructing a data center near Paris, which is either a statement of intent or a very expensive piece of real estate, depending on how the next few years go.
The open-weights approach, meanwhile, means anyone can download, modify, and deploy certain Mistral models without asking permission. This is described as democratizing AI. It is also an excellent way to build an ecosystem of dependent users who will, eventually, want the enterprise version.
What happens next
The round has not closed, the valuation is unconfirmed, and Mistral remains a fraction of the size of its American counterparts by every financial measure that exists.
Europe has decided to fund its own frontier AI anyway. This is either the beginning of a genuinely competitive second front in the global AI race, or the most expensive way to discover that the Americans had a head start. The French, historically, have not found either outcome particularly surprising.