Mistral pursues AI superintelligence with €4 billion infrastructure plan

Mistral AI is betting that Europe can close the AI gap with the US through superintelligence research and a €4 billion infrastructure buildout, but its chief executive warns the continent's biggest obstacle is the sheer scale of investment required.

Mistral AI Chief Executive Arthur Mensch said Europe must invest at a far larger scale to counter US dominance in artificial intelligence, as the French startup pursues superintelligence capabilities and expands its data center network.

"As long as we have adversaries that are threatening, and they are threatening, we do need to have our own capabilities," Mensch said at a press conference in Paris on Thursday, pushing back against Pope Leo's criticism of AI in warfare.

The company plans to build a 10-megawatt data center in Les Ulis, France, due to open in the third quarter of 2026, part of a €4 billion investment plan targeting 200 MW of computing power by the end of 2027. Mistral has already announced two other facilities in Sweden and France. The startup provides AI models to the French military, a business line that drew criticism from the Vatican this week.

Europe's ability to compete in frontier AI hinges on closing an investment gap with US rivals. Mistral's €4 billion plan, while significant for a European startup, represents a fraction of what US hyperscalers spend on AI infrastructure annually — the exact gap Mensch identified as Europe's biggest obstacle to tech independence.

The Paris-based company, founded in 2023 by former Google and Meta researchers, has emerged as Europe's leading challenger to US AI dominance. Its open-weight models compete directly with offerings from OpenAI, Google and Anthropic, though at a fraction of the compute budget available to US rivals.

Data Center Push Faces Local Opposition

Mistral's infrastructure expansion comes as data center projects face growing local opposition around the world, including in France. The company's Les Ulis facility will add 10 MW of capacity when it opens next year, a small fraction of the 200 MW target the company aims to reach by the end of 2027 through its three announced sites.

The €4 billion investment plan positions Mistral to train larger models as it pursues superintelligence — AI systems that surpass human capabilities across most tasks. The company has not disclosed specific performance targets for its next-generation models.

Military AI Draws Vatican Criticism

Pope Leo on Monday urged international regulation to curb the development of AI systems, warning they could spread misinformation and risk fuelling perpetual conflict. The document specifically criticized the use of AI in warfare.

Mensch defended the company's military contracts, arguing that Europe cannot afford to cede defense AI capabilities to adversaries. Mistral's models are used by the French military, though the company has not disclosed the specific applications.

The broader public remains uneasy about AI's rapid advancement. Videos showing students booing executives discussing AI at US graduation ceremonies have circulated online. "There is some expected anguish around artificial intelligence, in that it's profoundly changing the way people are working," Mensch said. "It's not the first time that people are a bit anguished at something coming up. But we'll be fine. We'll find a way to use it efficiently."

Conviction founder Sarah Guo, who backed Mistral at the seed stage, represents the early investor conviction that European AI could challenge US dominance. The startup's ability to secure follow-on funding at higher valuations will depend on whether its superintelligence strategy produces measurable performance gains against US frontier models.

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