Mistral AI is considering developing its own semiconductors to gain greater control over its operating framework. Mistral CEO Arthur Mensch told CNBC that while the company currently relies on Nvidia’s technology, in-house chips are a viable long-term strategy to significantly reduce the cost of processing data tokens. Such a move would bring Mistral into line with US giants like Google and Amazon, which use application-specific integrated circuits to optimise the synergy between hardware and software.
To boost its computing power, the Paris-based start-up is investing heavily in infrastructure. Mistral has earmarked €4 billion for data centres in Sweden and France, including a new facility dedicated specifically to AI inference. According to Mensch, these investments are crucial if Europe is to overcome its infrastructure shortfall and treat artificial intelligence as a key strategic asset, comparable to energy resources. In doing so, Europe can avoid a massive commercial imbalance and preserve its global competitiveness. These facilities will provide essential computing power for both Mistral’s corporate clients and other AI research labs.
In a bid to challenge the market dominance of Anthropic and OpenAI, Mistral has introduced ‘Vibe’, an agent-based platform tailored to businesses. The system allows users to delegate complex tasks such as programming and drafting documents. The AI carries out these tasks autonomously on the basis of a single prompt. CTO Timothée Lacroix said that Vibe can manage the full lifecycle of coding, from the initial writing through to final deployment.
These strategic expansions are designed to fuel rapid financial growth. Mistral aims to generate €1 billion in revenue in 2026, a sharp increase from €200 million previously. Despite this ambitious target, the company remains significantly smaller than its American counterparts, whose expected revenues run to tens of billions of dollars.
Business AM
Ltd
