Skip to content
StartupXO Startup ideas, news, talent
Menu

Quick links

Language

Investment & M&A

The Money That Priced Mistral at €21B Came From Companies That Run Fabs

Published: 2026-09-09

MistralSamsungSeries DOn-Premises AIEuropean AI

In short

Mistral raised €3 billion at a post-money valuation above €21 billion, in a round led by Samsung Electronics. Twelve months ago the Series C lead was ASML.

Mr. Latte's take

Samsung led the round and, in the same week, put Mistral Large on-premises inside its chip division. Those are not two events but one procurement decision priced twice. Set next to ASML leading the Series C at an €11.7 billion valuation a year earlier, the pattern is that Mistral's price is being set by the companies that intend to run the model inside their own buildings rather than by clouds or crossover funds. Sovereign AI is the label on the round; what actually cleared a price is a deployment model.

Who Led, and Who Followed

Mistral raised €3 billion in a Series D. Post-money valuation sits above €21 billion. The company calls it the largest equity fundraising round ever completed by a European technology company, and notes it comes three years after launch.

Samsung Electronics led. EQT’s Scaleup Europe Fund and existing investor PSG Equity co-led. Advent, funds and accounts managed by BlackRock, and the Grand Duchy of Luxembourg came in new. Existing backers including a16z, ASML, Bpifrance, DST Global, General Catalyst, Index Ventures, Korelya Capital, Lightspeed, Nvidia and Salesforce Ventures followed on.

The company also stated where the money goes: expanding frontier research, scaling the compute used to train models, expanding infrastructure, and accelerating commercial growth and international footprint. It says it operates across 20 countries and supports more than 125 global enterprises, naming Airbus, ASML and HSBC among them.

A Year Ago the Same Seat Held ASML

The Series C was announced on September 9, 2025. Mistral took €1.7 billion at a post-money valuation of €11.7 billion, in a round led by ASML, which builds the lithography systems used to make chips.

The doubling of the valuation over twelve months is the easy number. The interesting one is where both leads came from. One makes the machines that make chips, the other makes chips. Neither sells cloud capacity, and neither is a financial investor participating only through a balance sheet. ASML stayed in this round as an existing investor and also appears on the customer list Mistral published with the announcement.

Samsung Took More Than Equity

Samsung’s Device Solutions division will work with Mistral to build models tuned for semiconductor design and manufacturing, based on Mistral Large. The deployment is on-premises. The stated use is defect detection and equipment optimization in advanced memory and logic chip production.

Young Hyun Jun, who runs the DS division, framed it as continuous innovation being required as AI chip design and manufacturing grow more complex. Arthur Mensch, Mistral’s co-founder and CEO, said AI is reshaping how complex technologies get built, from silicon to software. How much Samsung itself put into the round was not disclosed.

The announcement landed alongside a Korea-France summit, and Korean outlets covered it next to the bilateral technology cooperation documents signed that week. The contract itself is narrower than the framing: it decides where one division’s process data gets processed.

The Number to Recount in Twelve Months

For founders, the useful part of this is not that there is one more frontier model to call. It is which deployment shape just cleared a price. Customers who cannot move data out of their own facilities, workloads narrow enough that a wrong answer costs real money, and the overlap between the two is where the lead investor of a €3 billion round decided to sit.

The caution sits in the same place. The first three uses of proceeds the company lists are research, compute and infrastructure. A large round does not convert into margin, and on-premises contracts carry per-customer implementation work, so headcount accumulates against revenue differently than it does in SaaS. Teams chasing this shape should decide at the first contract how far they will staff deployment.

Two figures are worth writing down: 20 countries, 125 enterprises. Both come straight from the company’s own announcement. What those two numbers read as twelve months from now is the scorecard for this round.