The Mistral AI logo Credit: Thrive Studios ID / Shutterstock Mistral has raised €3bn in a Series D at a post-money valuation of more than €21bn, which the company says is the largest equity fundraising ever completed by a European technology firm. Samsung Electronics led the round, three years after the Paris company was founded. The co-leads are the part worth pausing on. One is PSG Equity, an existing backer since the Series C.
The other is the Scaleup Europe Fund, the roughly €5bn vehicle the European Commission set up to keep the continent’s largest technology companies from having to raise abroad, and which EQT was selected to manage in May. Europe’s biggest private technology round now has publicly backed European capital sitting at the front of it. The trajectory has been steep. Mistral closed a €1.7bn Series C at an €11.7bn valuation in September 2025, led by ASML.
Twelve months and one round later, the valuation has nearly doubled. TNW reported in the summer that Samsung was in talks at around €20bn; the final figure came in above that. Both of the last two rounds were led by manufacturers rather than by software investors. ASML makes the lithography machines that make advanced chips possible; Samsung makes the chips, the memory and a great deal else.
For a company whose argument is that AI has to be run on infrastructure its customers control, backing from the industrial base rather than from the American venture stack is not incidental. Mistral’s framing of why now is worth quoting back. In the first wave of generative AI, it argues, the question was who could build the most powerful model. The question organisations and governments are asking instead is how to harness one for work that matters to them without handing over control of the infrastructure and the intelligence loop.
Whether that reframing is analysis or marketing depends on how the next few years go, but it is the thesis the round is priced on. That argument is the pitch. Mistral describes itself as building the whole stack, open-weight models, the compute they run on, and the products on top, so that customers are not locked into one vendor’s roadmap, pricing, or availability. It frames sovereignty as control across four dimensions: data that stays inside the organisation, models that can be customised, compute that is private and predictable, and production systems that are auditable. “Organisations want to industrialise AI inside systems that are already complex: their own data, their own tools, their own compliance constraints,” said Arthur Mensch, co-founder and chief executive.“That’s the real problem.” He said the round lets the company move faster on research, products and infrastructure. “Most companies at the frontier choose a layer,” said Kirk Lepke, a partner at EQT and the Scaleup Europe Fund.
Mistral is building across models, infrastructure and applications from Europe, he said, calling it the harder path and one designed to give customers real control. The customer base is the evidence for it. Mistral says it operates across 20 countries and supports more than 125 global enterprises, naming Airbus, ASML and HSBC. Those are precisely the organisations for which the question of whose cloud the model runs in is a board-level one rather than a procurement detail.
The new money arrives alongside a list of investors that reads like a map of the argument. Advent, funds managed by BlackRock and the Grand Duchy of Luxembourg, came in new. Existing backers taking part include NVIDIA, a16z, Bpifrance, BNP Paribas CIB, Salesforce Ventures, General Catalyst, Index Ventures, Lightspeed, Eurazeo and ASML. What the capital funds are is frontier research and, underneath it, compute.
Mistral has been assembling that separately, raising $830m in debt from seven banks for a data centre outside Paris and selling capacity to European industrial groups before it was built. Mistral has spent three years arguing that depending on American AI infrastructure is a strategic exposure rather than a procurement preference, a case that got its clearest illustration this year. The capital structure now matches the argument. Whether the models do is the question the next twelve months will answer.
