European LLM player Mistral has acquired Emmi AI, the Linz, Austria AI startup, in a deal that will be announced today and described by sources as “definitely not” an ‘acqui-hire’. Emmi builds AI models that generate industrial engineering simulations, completing them in seconds rather than days.
The acquisition is only Mistral’s second ever, after it acquired Koyeb, a high-performance serverless infrastructure for AI, earlier this year.
It is a fast exit for a company that only raised its €15 million seed round in April 2025, described at the time as the largest seed round ever raised by an Austrian startup. That round was led by 3VC, Speedinvest, Serena Data Ventures and PUSH VC.
The price is not being disclosed. But Pathfounders understands the transaction was “significant” and came after the company had already begun validating both its science and its commercial demand.
The deal gives Mistral a vertical AI asset in an the unglamorous but strategically important corner of industrial engineering. Emmi AI uses AI to model and simulate physical systems (fluid dynamics, heat, stress, aerodynamics, crash behaviour, and semiconductors), the sort of work that underpins the design and manufacturing of things such as aircraft, cars, chips, and even energy infrastructure.
This market is dominated by players such as Schrodinger and Dassault Systèmes, both of which provide software tools to perform computational chemistry and material simulations.
Pathfounders understand the original plan for the startup had been to raise a Series A, not sell. Emmi had “runway for another more than two years” and was not under pressure, said a source. But the company had moved faster than expected, converting a proof of concept into a seven-figure contract and signing further industrial customers.
That traction brought unsolicited acquisition interest. The logic of the Mistral deal was not just money, but acceleration for the goals of the founders.
Industrial companies still spend vast amounts of time waiting for simulations to complete. An engineer designs an aircraft wing, a car body, a chip or a transformer component, sends it to the simulation team, waits 48 to 72 hours, and then iterates. Emmi’s offering is an AI that can collapse that loop from days into hours or less.
That is why the acquisition is significant. Mistral is best known as Europe’s leading large language model company, but to survive, it must move beyond chat interfaces and office productivity into other areas such as industrial systems, infrastructure and applied science.
Indeed, Mistral’s own public positioning has increasingly emphasised Europe’s need to build self-reliant AI infrastructure and capability.
And it’s clear that Emmi fits that direction, given that it is about teaching AI the language of physics.
The acquisition is Mistral’s second known deal after its agreement to acquire French cloud startup Koyeb, a transaction reported in February this year as its first acquisition. It’s worth noting that Serena Data Ventures was also an investor in Koyeb, giving the firm two exits into Mistral in quick succession.
For Emmi’s investors, the deal is a fast outcome.
But, say sources, to survive, Emmi needed access to proprietary engineering data from its customer to train its models, and that was coming at a slower pace than was required.
Mistral, in turn, needed to shore up its penetration into Europe’s industrial sectors, where the continent still has some deep technical advantage.
The marriage was thus consummated.
