Mistral raises €3 billion as Samsung backs open-weight AI push

A Korean hardware giant has placed the biggest single bet yet on Europe's answer to American AI dominance. Samsung Electronics led a €3 billion Series D into Mistral AI that values the three-year-old French startup at more than €21 billion — the largest equity raise by a European technology company — and hands open-weight models a credible challenger to the closed systems of OpenAI and Anthropic. The round, announced Sept. 8, nearly doubles Mistral's €11.7 billion valuation from a Series C led by ASML a year earlier.

"Long term, the plan is to fully rely on capacity that we are building ourselves," Arthur Mensch, Mistral's co-founder and chief executive, said in an interview. He argued the company remains undervalued and that macroeconomic conditions are starting to favor open-source models, which let enterprises keep data and training inside their own walls rather than renting intelligence from a single vendor.

The round was co-led by the EU-backed Scaleup Europe Fund, managed by EQT, and existing investor PSG Equity. New investors include Advent, funds and accounts managed by BlackRock, and the Grand Duchy of Luxembourg, while a16z, ASML, BNP Paribas CIB, Bpifrance, General Catalyst, Index Ventures, Lightspeed, NVIDIA and Salesforce Ventures added to their stakes. The Scaleup Europe Fund, anchored by a €1 billion European Commission commitment, counts Novo Holdings and Santander among its backers.

Mistral plans to spend the capital on owned data-center infrastructure, with compute capacity expected to roughly double over five years, and on training larger models. The company, which operates in 20 countries and supports more than 125 enterprises including Airbus, ASML and HSBC, expects annual recurring revenue to exceed $1 billion this year and to accelerate growth into 2027. It will triple headcount at its Singapore office as it expands across Southeast Asia.

Why Samsung, and why now

Samsung's lead role extends a pattern of strategic investors from the semiconductor supply chain backing Mistral, following ASML's Series C lead. The Korean hardware giant, dominant in memory chips that feed AI training, expects Mistral to integrate its models into manufacturing processes in South Korea, mirroring work already underway at ASML. Kirk Lepke, a partner at EQT and the Scaleup Europe Fund, framed the investment as a bet on execution, saying the combination of ambition and execution puts Mistral in a position to shape the next era of AI.

The capital structure matters as much as the amount. Mistral is the only AI company building the full stack of open-weight models, the infrastructure they run on and the products that bring them into production, according to the company. That positioning — which Mistral calls the sovereign AI layer, spanning data, models, compute and production systems — is the pitch that separates it from OpenAI and Anthropic, whose proprietary models lock customers into a single vendor's roadmap and pricing.

The open-versus-closed math

Mensch's claim that macroeconomics favor open models rests on cost and control. Enterprises deploying Mistral's open-weight systems can run them on their own or rented compute, avoid per-token inference fees, and customize weights without exposing proprietary data. Chinese open-source developers, including DeepSeek, offer similar flexibility at lower cost, and Mensch acknowledged Chinese models can run on Mistral's infrastructure. But he argued European enterprises cannot rely on long-term support for Chinese models given uncertain upgrade roadmaps and potential export restrictions, calling volatility in that space "quite extreme."

The competitive stakes are measurable. OpenAI and Anthropic have raised tens of billions from Microsoft and Amazon respectively, funding proprietary frontier models and massive cloud commitments. Mistral's €3 billion round, by contrast, buys owned compute and a sovereign alternative that governments across the EU are actively subsidizing through vehicles like the Scaleup Europe Fund. For investors, the question is whether open-weight economics — lower inference margins but broader enterprise adoption — can sustain a company valued at $24.4 billion before it proves the $1 billion revenue run-rate it projects for this year.

This article is for informational purposes only and does not constitute investment advice.