In recent months, Mistral AI has made a series of announcements about ambitious projects. The French startup, which specializes in generative artificial intelligence, is positioning itself to bring these projects to fruition. In early September, it officially announced a new round of funding totaling 1.7 billion euros. The majority of this amount is being provided by ASML, the Dutch giant in lithography machines, which are essential for etching chips.

This deal sets a record for a French startup, as does the company's valuation of 11.7 billion euros. It also allows Mistral to triple the amount of capital raised from investors. Nevertheless, the company remains far behind the financial clout of American giants such as OpenAI and Anthropic, with which it still intends to compete.

In this major international competition, Mistral cannot afford to slow down. It must continue to develop new models capable of generating text, computer code, or images. It must also accelerate the rollout of its tools, particularly its consumer-facing chatbot, Le Chat, which competes with ChatGPT and Gemini. These requirements entail high costs to secure the computing power needed for AI training and inference.

AI Campus

With this in mind, Mistral is placing a strong emphasis on creating an AI campus in the Paris region. Billed as the largest in Europe, it will house a massive data center, high-performance computing infrastructure, and research centers. The groundbreaking ceremony is scheduled for the second half of 2026. The facility is expected to become operational in 2028.

This project will be carried out by a joint venture involving Mistral, Nvidia, the state-owned bank Bpifrance, and the Emirati fund MGX. MGX is expected to cover the bulk of the investment, initially estimated at approximately 8.5 billion euros. Ultimately, the total cost could range from 30 to 50 billion.

At the same time, Mistral aims to roll out its own cloud platform. Called Compute, it will be paired with a software layer designed to enable companies to quickly develop and deploy AI services. The startup plans to build a data center comprising 18,000 GB200 superchips, Nvidia’s most powerful graphics cards. Such computing power will be unmatched among European companies.

Revenue increased tenfold

Mistral hopes to attract European government agencies and businesses seeking an alternative to the American giants in the sector. The company aims to capitalize on the issue of technological sovereignty, particularly by capitalizing on fears sparked by Donald Trump’s return to the White House. Orange, BNP Paribas, and the Kyutai research lab will be among the first customers of this cloud offering, which is expected to launch next year.

By moving into the cloud, Mistral is aiming not only to drive revenue growth but also to build a high-margin business. The startup now reports annualized revenue of 300 million euros. That’s about ten times more than at the end of 2024… though it remains a far cry from OpenAI’s performance, which has surpassed the 12-billion-dollar mark.

In reality, Mistral can't truly compete with ChatGPT's developer. Its goal isn't to offer models as powerful as those of the major U.S. companies, but to provide models that are competitive enough to serve as a credible alternative. In recent months, it has therefore signed a number of commercial agreements with major French companies, for reasons of sovereignty as well as image.