Europe's best-funded AI lab has spent two years being described as the region's answer to OpenAI. At its first summit, held at the Carrousel du Louvre in Paris, Mistral tried to become something broader: a company that sells the whole stack, from the model down to the building it runs in.
The most visible change was a rename. Le Chat, the assistant Mistral launched in early 2024, is now called Vibe, and it has been split in two. VentureBeat reports that Vibe for Work connects to tools such as Google Workspace, Outlook, SharePoint, Slack and GitHub, while Vibe for Code is a programming agent with a Visual Studio Code extension.
Into the factory
The more unusual announcement was industrial. Mistral launched a stack for engineering work that pairs its language models with physics simulation, drawing on its acquisition of the Austrian startup Emmi AI. The company named Airbus, BMW Group and the chip-equipment maker ASML as early customers, a roster that points at aerospace, automotive and semiconductors rather than the usual software buyers.
Underpinning all of it is infrastructure. Mistral confirmed a new inference data center near Paris, funded by roughly 830 million dollars in debt raised from a group of banks, according to Data Center Dynamics. The site will run on thousands of Nvidia GPUs and is meant to give European customers a way to use frontier models without their data leaving the continent.
That sovereignty pitch is the thread tying the day together. Mistral is smaller than its American rivals and cannot outspend them, so it is selling something they find harder to offer: models, tools and hosting under European jurisdiction, aimed at industries that care where their data sits. It is a narrower bet than chasing artificial general intelligence, and for a company Mistral's size it may also be the more durable one.
Commentarii · 0