Mistral AI, the Paris-based developer of large language models, has acquired SkipLabs, a startup that builds tools meant to make code written by AI faster to produce, more reliable and more secure. The French startup news site Maddyness and Dealroom News reported the deal on November 5, 2026.
The price was not disclosed, and neither company published details of how the deal was structured.
SkipLabs' engineers and scientists will join the team of Guillaume Lample, Mistral's co-founder and chief scientist. Founder Julien Verlaguet moves to Mistral with them. The acquisition is Mistral's fifth of 2026.
What SkipLabs built
SkipLabs was founded about six years ago and has offices in Paris and London, according to Maddyness. Its best-known product is Skip, an open-source framework for building the back end of applications that react to changing data. It is written in SkipLang, the company's own programming language, and developers use it through a TypeScript interface.
With Skip, a developer describes how results are computed from input data, and the framework updates those results by itself whenever the inputs change. That removes the need to track dependencies or clear caches by hand, a frequent source of bugs in real-time features. Amplify Partners, the venture firm that led SkipLabs' seed round, compared the approach to what the React library did for user interfaces. Version 1 of Skip came out in early 2025 and is distributed as npm packages under the @skipruntime name.
More recently the company turned its attention to AI. Maddyness and Dealroom describe its current work as technology to make AI code generation faster, more reliable and more secure.
The founder
Verlaguet spent nearly nine years as an engineer at Meta, Dealroom reported, including time at FAIR, the AI research lab Yann LeCun set up at the company. At Facebook he designed the typechecker for Hack, one of the company's programming languages, according to Amplify Partners, which called him one of the leading programming language designers in the world. Amplify wrote that the idea for Skip came to him while he was working on Hack.
In comments in French quoted by Maddyness, Verlaguet said that with AI the author of code is no longer a person. According to Dealroom, he argued that this undermines the assumptions behind today's programming tools and presented it as a chance to rebuild them from scratch. He also said that joining Mistral would let his team carry these ideas into the way the company's models are trained.
Funding and investors
Amplify Partners led SkipLabs' seed round and announced its investment in March 2025, without giving an amount. The startup database VCBacked lists the round at $8 million in March 2025, with Tapestry VC, Olivier Pomel of Datadog, Yann LeCun and Spencer Kimball also taking part. These figures were not confirmed by Mistral or SkipLabs.
VCBacked lists SkipLabs' headquarters as New York, and the Skip framework's website carries a copyright notice for SkipLabs, Inc., which suggests the company also has a US entity. French reports describe it as a French startup.
Mistral's year of acquisitions
Mistral made its first acquisition in February 2026, when it bought the cloud company Koyeb. It went on to buy Emmi AI, an Austrian industrial AI company, in May; Pimento, an advertising specialist, in September; and New Material, which makes user-experience tools for AI, in October.
Dealroom wrote that the purchases are building out Mistral's technology, infrastructure, specialist skills and solution design so that large companies can put its models to work more easily. Mistral has partnerships with BMW, Airbus, the shipping group CMA CGM and TotalEnergies.
The company has the money for more deals. In September 2026 it raised €3 billion (about $3.5 billion) at a €21 billion (about $24 billion) valuation in a round led by Samsung, with ASML and Nvidia taking part. In early October it released Mistral Large 4, its newest large model. Dealroom framed the run of deals as a sign that Mistral is competing with OpenAI and Anthropic on infrastructure, talent and tools as well as on models.
