Brief IA

Mistral AI Aims for One Gigawatt of Power by 2030

💼 Business & Startups·Tom Levy·

Mistral AI Aims for One Gigawatt of Power by 2030

Mistral AI Aims for One Gigawatt of Power by 2030
Key Takeaways
1Mistral AI plans to establish an inference site in Les Ulis, operational by 2026, aiming for 200 megawatts by 2027.
2The company is investing €4 billion in Europe to enhance its computing capacity, with a key center in Bruyères-le-Châtel.
3Mistral AI has acquired Koyeb and Emmi AI to develop its AI cloud and advanced engineering models.
💡Why it mattersMistral AI is positioning itself as a major player in Europe against global AI giants, strengthening its technological independence.
Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisis ton rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

📄
Full Analysis

Mistral AI Envisions Colossal Computing Power

At the "AI Now Summit" event held at the Carrousel du Louvre in Paris, Mistral AI unveiled its ambitious vision for computing power. The company, which is celebrating its third anniversary, announced its intention to create a dedicated inference operations site in Ulis, Essonne. This site is expected to be operational by the third quarter of 2026. This initiative is part of a broader plan to achieve a computing capacity of 200 megawatts by 2027, and one gigawatt by 2030. This ramp-up is driven by the need to meet the growing demand for token-intensive AI models.

Mistral AI's Independence Strategy

Mistral AI is now positioning itself as a "full-stack AI company," covering the entire value chain of AI. Timothée Lacroix, co-founder and CTO, emphasized the importance of computing power for clients and the future of the company. "Our roadmap allows us to meet global demand for AI infrastructure while maintaining our independence," he stated. Arthur Mensch, CEO of Mistral AI, added that AI is too strategic to be left in the hands of a few, and that the world needs open platforms to develop AI applications. "We develop cutting-edge models, deploy them within organizations' infrastructure, train them on their proprietary knowledge, and execute them on controllable computing power," he asserted.

Massive Investments in France and Europe

On the sidelines of last year's AI Action Summit, Mistral AI had already announced the establishment of its first data center on French soil, in Bruyères-le-Châtel, Essonne, about thirty kilometers from Paris. Its commissioning is expected in the coming weeks. This center will be used to run a "Mistral Compute" model, a cloud offering dedicated to AI launched with Nvidia in 2025. To support these projects, Mistral AI secured a loan of $830 million in March. The company also plans to establish data centers in Sweden. In total, Mistral is investing no less than €4 billion in data centers in France and Europe. Although this amount is colossal, it remains far from the massive investments made by industry giants like OpenAI and Anthropic. "We don't have Microsoft's financial backing," acknowledged Arthur Mensch at a press conference prior to the "AI Now Summit" event. This year, Mistral AI hopes to surpass the billion-euro revenue mark.

The AION Consortium and Mistral AI's Position

Recently, the issue of computing power in AI has been at the center of discussions in France. Several companies, including Iliad, Ardian, Artefact, Bull, Capgemini, EDF, Orange, and Scaleway, announced the creation of a consortium called "AION" to build an AI gigafactory in France. This project, estimated at €10 billion, aims to achieve a computing power of one gigawatt, with an initial target of 100 megawatts. The founding members hoped that Mistral AI would join this consortium, but Arthur Mensch felt that the project "is not really suited" to their full-stack AI company approach.

Acquisitions and New Initiatives from Mistral AI

To realize its vision, Mistral AI has made two strategic acquisitions in recent months. The first is Koyeb, a serverless cloud provider that allows developers to deploy AI applications without managing infrastructure. This acquisition fits into Mistral AI's strategy to build a dedicated AI cloud. Last week, the company also acquired the Austrian startup Emmi AI, which specializes in Large Engineering Models to create digital twins for industrial simulations that adhere to the laws of physics. These acquisitions aim to optimize design, manufacturing, and production processes.

Additionally, Mistral AI unveiled Vibe, its new agent platform for businesses. "Users can define the specifications and move on, while Vibe thinks, writes, and delivers the final work from a single conversation. Vibe Code writes, tests, and deploys code across different codebases," explained Timothée Lacroix. This initiative allows Mistral AI to strengthen its B2B offering to attract more businesses worldwide, particularly in key markets such as the United States, Japan, and Australia.

Brief IA — L'actualité IA en français

L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.