Mistral AI buys Austrian physics AI startup in industrial push
Europe's leading artificial intelligence firm, France's Mistral AI, said on Tuesday it has acquired Linz-based Emmi AI for an undisclosed sum, aiming to enhance its offering for industrial clients across Europe.
Emmi AI, which raised 15 million euros in Austria's largest funding round in 2025, specialises in models capable of handling complex physics such as airflow, heat transfer and material stress.
[...]
Mistral designs solutions around each client's needs, assembling multiple AI tools where one might monitor production for defects, another control a robotic arm, a third process logistics data, while all operate in coordination.
Adding Emmi's capabilities will allow these systems to simulate and interact with the physical world more precisely, it said.
The company cited its work with ASML where Mistral-equipped EUV lithography machines now use vision models to detect engraving defects, cutting diagnostic times from hours to just eight minutes and minimising waste of costly silicon wafers.
"You just save 10 hours of downtime on very expensive equipment," ASML CFO Roger Dassen told shareholders at the company's April AGM.
The company, whose clients include Stellantis, Veolia, and drone manufacturer Helsing, told Reuters that purpose-built models trained on company-provided data will outperform off-the-shelf alternatives trained on general datasets, and emphasised Europe's century of manufacturing expertise as an advantage.
[...]
Emmi AI, which raised 15 million euros in Austria's largest funding round in 2025, specialises in models capable of handling complex physics such as airflow, heat transfer and material stress.
[...]
Mistral designs solutions around each client's needs, assembling multiple AI tools where one might monitor production for defects, another control a robotic arm, a third process logistics data, while all operate in coordination.
Adding Emmi's capabilities will allow these systems to simulate and interact with the physical world more precisely, it said.
The company cited its work with ASML where Mistral-equipped EUV lithography machines now use vision models to detect engraving defects, cutting diagnostic times from hours to just eight minutes and minimising waste of costly silicon wafers.
"You just save 10 hours of downtime on very expensive equipment," ASML CFO Roger Dassen told shareholders at the company's April AGM.
The company, whose clients include Stellantis, Veolia, and drone manufacturer Helsing, told Reuters that purpose-built models trained on company-provided data will outperform off-the-shelf alternatives trained on general datasets, and emphasised Europe's century of manufacturing expertise as an advantage.
[...]