Mistral AI just raised 3 billion euros. Its best open model still loses to Chinese ones you can download for free. Both of those things are true, and the second one is the more interesting story.
The Series D closes three years after the French company launched, and it pushes Mistral’s valuation past 21 billion euros. Mistral said it’s the largest equity funding round any European tech company has ever raised.
Who’s writing the checks
Samsung Electronics is leading. The Scaleup Europe Fund, managed by EQT, and PSG Equity are co-leads.
New money is coming from Advent, funds managed by BlackRock and the Grand Duchy of Luxembourg. Existing backers a16z, Nvidia, ASML and General Catalyst are all in again.
That’s a sovereign wealth participant, a chip toolmaker, a GPU vendor and the world’s largest asset manager on the same cap table. Read into that what you will about who wants a European model provider to exist.
The valuation roughly doubled in a year
When ASML put in 1.3 billion euros in September 2025, Mistral was worth about 12 billion euros. Now it’s north of 21 billion.
Nothing about the model lineup explains that jump. Mistral Medium 3.5 trails Chinese open competitors like Qwen and Kimi, and it doesn’t compete with closed US models at all.
So the multiple isn’t buying benchmark wins. It’s buying a customer list and a jurisdiction.
Enterprise is the actual product
Mistral operates in 20 countries and serves more than 125 companies. Airbus, ASML and HSBC are among them.
Since early 2026, the company has benefited heavily from European customers looking to cut their reliance on US providers. That’s the pitch, and unlike the benchmark story, it’s working.
Being second-tier on capability matters less when the buyer’s requirement starts with where the weights run and who’s subject to which subpoena. A European bank choosing Mistral over a US lab isn’t making a performance call.
The debt nobody talks about
In March, Mistral took out an $830 million loan to fund its own data centers.
That’s the part worth watching. Renting compute is an expense you can cut. Owning it is a commitment, and it means the 3 billion euros isn’t only going toward research salaries.
Vertical integration is expensive and it’s how you stop paying a US hyperscaler for the privilege of being a European alternative to US hyperscalers.
What to actually judge this on
If you’re evaluating Mistral for a deployment, ignore the funding number. It tells you the company will still exist in two years, which is genuinely useful information and nothing more.
Test Medium 3.5 against Qwen and Kimi on your own workload first. If Mistral wins, great. If it loses and you pick it anyway because of where the data sits, that’s a defensible reason, and it’s the same reason 125 other companies got there ahead of you.
