Cocorico: three billion euros in a single raise. Mistral AI has just completed the largest financing round ever achieved by a private European technology company. The operation, led in particular by Samsung Electronics, PSG Equity and the Scaleup Europe fund, brings the valuation of the French company to around 21 billion euros. A little over three years after its creation in Paris by former researchers from Google DeepMind and Meta, Mistral AI is changing dimension. This new capital must finance its future models, its products and above all the computing infrastructures necessary to compete with the American and Chinese giants.
Key Points
Mistral AI completes record fundraising of 3 billion euros led by Samsung, key supplier of HBM memories
After the 1.3 billion ASML in September 2025, the French company collected almost 5 billion euros in less than a year
The funds mainly finance computing: 18,000 Grace Blackwell chips deployed with Nvidia in a data center in Essonne
The strategy of open weight models appeals to European armies, banks and industrialists seeking control of their data
Mistral raises 3 billion euros, Samsung in the lead
Samsung Electronics already occupies a strategic position in the artificial intelligence industry. The South Korean group notably manufactures HBM (High Bandwidth Memory), ultra-fast components used with AI accelerators. Samsung Electronics also operates its own foundries and integrates artificial intelligence functions into its Galaxy smartphones.
By investing in Mistral AI, Samsung Electronics complements this hardware presence with exposure to models and software that exploit this computing power. This merger could also promote future industrial collaborations between the two groups.
Mistral AI's previous funding round dates back to September 2025. ASML then contributed 1.3 billion euros out of a total fundraising of 1.7 billion, obtaining around 11% of the capital. The operation valued the French company at 11.7 billion euros.
Nvidia, General Catalyst, Lightspeed, Andreessen Horowitz and Bpifrance were already among its investors. By adding the current fundraising to the 2025 funding round and the $830 million in debt obtained in March 2026, Mistral AI has raised more than five billion euros in financing in less than a year.
The cost of computing at the heart of Mistral AI’s strategy
A significant part of these resources must be used to strengthen the company's computing capabilities. Mistral AI is notably preparing the opening of Mistral Compute, its own data center located in Bruyères-le-Châtel, in Essonne.
According to the French company, the site must accommodate 13,800 Nvidia GB300 accelerators, based on the Grace Blackwell architecture, for an initial electrical power of 44 megawatts. Its financing is based in particular on a debt of 830 million dollars contracted from a banking consortium.
The objective is twofold: to have an infrastructure capable of training future Mistral AI models and to directly offer computing power to its customers. The company is thus seeking to better control its technical chain, from data centers to professional applications.
The amounts nevertheless remain very far from those committed in the United States. Sam Altman indicated in 2025 that OpenAI was preparing around 1.4 trillion investments in computing infrastructures over eight years! The comparison shows the gap in resources, but also the strategic importance of this lifting for Mistral AI.
Open models as an argument for sovereignty
Mistral AI also relies on its open weight models. Some, like Mistral Small 4 and Magistral Small, are offered under the Apache 2.0 license. They can be downloaded, adapted and deployed on a company or administration's own servers.
This possibility is of particular interest to sectors handling sensitive information. Customers can keep their data within their own infrastructures, customize models and limit their dependence on an external provider.
The French Ministry of the Armed Forces has signed a framework agreement giving access to Mistral AI models and services to several public organizations. CMA CGM has also entered into a five-year partnership supported by an investment of up to 100 million euros. Stellantis, ASML, HSBC and TotalEnergies are also among the companies using its solutions.
This search for autonomy echoes certain concerns of the cryptocurrency sector, where code transparency and reducing dependence on centralized operators occupy an important place. Decentralized computing networks are also trying to offer an alternative to large infrastructure providers, but with still much lower resources.
At the same time, the European Union is seeking to catch up with InvestAI, a program announced to mobilize 200 billion euros in public and private funding, including 20 billion devoted to future artificial intelligence “gigafactories”.
