Mistral Raises €3 Billion Led by Samsung in Largest Tech Round in European History
French company Mistral closes Series D round of €3 billion at a valuation of €21 billion, with Samsung Electronics as the lead investor and ARR on track to reach $1 billion before the end of 2026.
Mistral announced on September 8 a Series D funding round of €3 billion at a post-money valuation exceeding €21 billion, the largest equity investment ever made in a European technology company. Samsung Electronics entered as the lead investor, with Scaleup Europe Fund, managed by EQT, and PSG Equity, already partners, as co-leads. Advent, funds managed by BlackRock and the Grand Duchy of Luxembourg joined for the first time. a16z, ASML, Bpifrance, NVIDIA, and Korelya Capital returned as investors.
CNBC confirmed the valuation in dollars at $24 billion. According to Arthur Mensch, co-founder and CEO, the company is expected to surpass $1 billion in annual recurring revenue before the end of 2026, and its own computational capacity will grow by about 100% over the next five years. "This round means we can move faster on R&D, products, infrastructure and keep giving organizations the ability to run AI at scale, on their own terms," Mensch stated in a release accompanying the announcement.
Neocloud is the label, data control is the product
Mistral positions itself as a "neocloud": open-weight models, dedicated compute infrastructure, and deployment layers that keep data within the client's perimeter. It's not rhetoric. The company claims to operate in 20 countries and already has over 125 corporate clients, including Airbus, ASML, and HSBC. Airbus and ASML are anchor clients of what the European Commission refers to as industrial data sovereignty. HSBC enters from the other side, that of the financial sector subject to the PRA in the UK and the ECB's oversight of European subsidiaries, both of which have data residency rules that the traditional OpenAI or Anthropic client still resolves with legal workarounds, not architecture.
The strategy of owning data centers, rather than just renting capacity, positions Mistral in the same structural axis as Nscale, a British provider that signed a $45 billion contract for six years with Anthropic in August. The difference is that Mistral controls both model and infrastructure in the same hand, something neither OpenAI nor Anthropic can do alone.
Why Samsung, and why now
Samsung Electronics' presence as a lead investor changes the geographic reading of the deal. The Korean giant plays on three fronts: HBM (high-bandwidth memory for accelerators), foundry (competing with TSMC for fabless clients), and devices (Galaxy, TVs, Bixby). An anchor stake in Mistral closes the triangle with an AI application layer that runs on any stack, including silicon designed by Samsung itself. For Korea, it's the institutional response to the perception that it was left out of the first cycle of commercial LLMs.
The fact that Nvidia returned as an investor but not as a leader is a data point to observe. Nvidia participated in previous funding rounds for Mistral and maintains a position but has ceded the stage to Samsung. At the same time, it has appeared in discussions about investing $2 billion in Nscale. This is not scatter; it’s deliberate diversification of the compute client portfolio outside the block of American hyperscalers.
What the success thesis doesn't tell on its own
Some analysts in Paris and London are taking a more nuanced view. Benedict Evans noted in a recent letter that the unit cost of training a company with €3 billion in cash is still a small fraction of what Anthropic and OpenAI burn per quarter, and questions whether an open-weight model with consistent margin is economically sustainable when the capacity frontier lies in closed models. Mistral's practical response has been to combine open weight with deployment products and services, but the LTV math of European corporate contracts needs to prove it supports the announced capex.
For the global reader, the design matters. Europe has put heavy institutional money into a company that aims to compete across the full stack with American players, and it did so with Asian leadership, not European. Asia has bought a seat at the de facto governance table of the AI model that Airbus, ASML, and HSBC will operate. And Brazil, whose largest corporate AI buyers today anchor in Azure OpenAI or AWS Bedrock, gains, for the first time, a platform option where data sovereignty is not a contractual add-on but rather a product attribute.