Arthur Mensch, chief executive officer of Mistral AI, warned in an interview at the Berlin digital summit that Europe could walk into a strategic trap if it focuses on industrial applications while leaving the development of the most advanced base models to the United States and China. Politico reported Mensch’s remarks, in which he argued that the continent needs to build its own compute capacity and adopt a pan‑European strategy to preserve technological sovereignty.
Why dependence is problematic
Mensch stressed that Europe cannot assume uninterrupted access to the latest foreign technologies, because the U.S. and China could impose export restrictions at any time. He criticized a line of thinking voiced by several EU leaders in September that suggested Europe should prioritize applications built on foreign models rather than investing in developing base models itself. According to Mensch, meaningful industrial gains require that the foundational technology powering factories and design tools remain under European control.
German strategy and the need for a Europe‑wide response
The remarks came after the German government on Monday unveiled a new strategy aiming to make Germany a sovereign AI nation by 2030, including support for building three European leading research laboratories. Mensch welcomed Germany’s initiative but argued that the challenge cannot be solved at the level of individual member states: the primary bottlenecks are capital and large‑scale compute capacity, not mainly a shortage of specialists, so a unified European approach is necessary.
EU heads of state and government will hold a first dedicated discussion of the issue at next week’s informal meeting in Dublin.
Industrial partners and the limits of software alone
Mistral’s largest market today is the manufacturing sector, with partners such as Airbus and ASML using AI from aircraft design to reducing machine downtime. Mensch cautioned, however, that industrial software alone does not guarantee independence if the underlying infrastructure is controlled by others.
The spread of European alternatives is further hindered by cheaper—and often higher‑performing—Chinese competitors. Mensch pointed to the example of the German search engine Ecosia, which recently switched from a Mistral system to Chinese models because it achieved better results at roughly half the cost.
Price, quality and Mistral’s positioning
Mensch rejected the notion that companies should accept trade‑offs between price and quality to secure technological sovereignty. While Mistral did not respond directly to Ecosia’s decision, the CEO said the company does not need to top every benchmark; instead it focuses on high‑demand, strategic application areas.
He highlighted the newly introduced Mistral Large 4 model as an example, saying it outperforms open‑weight Chinese rivals in cybersecurity and geospatial imaging, and that there is significant demand for Mistral’s technology in the United States as well.
Conclusion
Mensch’s core message is that without building substantial domestic compute capacity and coordinating investment across Europe, the continent risks losing technological independence. Germany’s recent strategy is an important step, but he insists that a comprehensive solution must be built at the European level.
(This article was prepared with the assistance of an AI tool; its final content was edited and verified by a journalist.)
