French AI champion Mistral AI is actively exploring the possibility of designing and eventually manufacturing its own custom chips, CEO Arthur Mensch told CNBC.
The move marks the company’s clearest signal yet of its desire to gain greater control over its infrastructure as it competes with U.S. leaders OpenAI and Anthropic.
“Of course, it is interesting,” Mensch said when asked about developing proprietary semiconductors.
Register for Tekedia Mini-MBA edition 20 (June 8 – Sept 5, 2026).
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab.
He added that the company is not ruling out the move in the future.
Mensch explained that custom chips could significantly lower the cost of inference, the process of running AI models, by optimizing hardware and software integration.
“Owning the chips may come, I think it should come at some point, but for now we are relying on Nvidia, which is a great partner to us, and we’re testing a few things here and there,” he said.
This marks the first time Mensch has publicly discussed Mistral’s semiconductor ambitions, highlighting a strategic evolution for the Paris-based startup. Valued at nearly 12 billion euros, Mistral is often positioned as Europe’s strongest contender in the global AI race. Developing its own chips would allow it to reduce dependency on external suppliers, lower long-term costs, and potentially create differentiated offerings tailored to European regulatory and data sovereignty requirements.
Data Center Expansion and Infrastructure Push
Mistral announced on Thursday a new data center in France dedicated specifically to inference workloads. The company has already invested 4 billion euros in data centers across France and Sweden as it ramps up compute capacity to meet growing demand.
“Europe is lagging behind when it comes to buildout of infrastructure, and so we are investing to close that gap,” Mensch told CNBC.
He framed the issue not just as technological but macroeconomic, noting that Europe cannot afford a massive commercial deficit in AI while aiming to remain competitive globally.
“You can’t afford to have a commercial deficit of a trillion if you actually want to stay competitive in the race, and so that’s something I think that people are realizing that we’re talking about something that should be concerning for any one of us,” he said.
The additional capacity in France will support Mistral’s enterprise customers as well as other AI labs seeking compute resources. Mensch noted strong demand from the broader AI community.
“AI labs are in sore need of compute, and we have some of it, and some of them are actually asking us for a lot of compute today,” he said.
He emphasized that customer needs will take priority, while still allocating resources to collaborative AI development.
New Agentic Platform “Vibe” Targets Enterprise Use Cases
Mistral also unveiled Vibe, a new agentic AI platform designed for enterprises. Agentic AI, systems capable of autonomously planning and executing complex tasks, has become a major focus for the industry, with competitors like OpenAI and Anthropic rolling out similar offerings.
Mistral’s Chief Technology Officer Timothée Lacroix described the platform in a statement saying: “Vibe is the agent platform for the tasks at hand, putting frontier AI to work. Users can set the brief and move on, as Vibe thinks, drafts, and delivers finished work from a single conversation. Vibe Code writes, tests, and deploys code across codebases.”
The platform targets practical enterprise applications such as document drafting, coding, and workflow automation, aiming to deliver measurable productivity gains rather than general chat capabilities.
Mistral is targeting 1 billion euros in revenue for 2026, a significant jump from around 200 million euros the previous year. While ambitious, this figure remains well behind U.S. rivals. OpenAI’s annualized recurring revenue reached $20 billion in 2025, and Anthropic is projected to hit $10.9 billion in revenue for the second quarter of 2026 alone.
The company’s enterprise focus, with customers including chip equipment leader ASML, differentiates it in a market increasingly crowded with general-purpose models. By investing heavily in European data centers and exploring custom silicon, Mistral is betting that sovereignty, compliance, and regional expertise will give it an edge in serving European businesses wary of over-reliance on U.S. cloud providers.
However, Mensch’s comments on custom chips reflect a broader European push for technological sovereignty. With heavy dependence on U.S. and Asian hardware, European AI companies face risks from export controls, data localization rules, and supply chain vulnerabilities. Developing in-house chip capabilities could help Mistral, and by extension Europe, reduce these dependencies while fostering local innovation ecosystems.
The company’s data center investments also address a critical gap. Europe has lagged in AI infrastructure buildout compared to the U.S. and China. By stepping in as both a developer and provider of compute, Mistral is helping close that gap while positioning itself as a key infrastructure player rather than solely a model provider.
However, the company has more challenges to contend with. Custom chip design requires enormous capital, specialized talent, and time. Success is far from guaranteed, and Mistral will need to balance these long-term ambitions with near-term revenue growth and competition from better-funded U.S. rivals.