Mistral’s chief executive, Arthur Mensch, has warned that Europe will get “leftovers” if it stops building frontier AI models. Apps built on American or Chinese models would also give Europe a “big trade balance problem”, he said on 20 October.
Mensch spoke on the European Technology Network talk show, Pieter Haeck reported for Politico. He and DeepL’s chief executive, Jarek Kutylowski, are pushing back against European leaders. Some of them say Europe should drop out of the race for the most advanced models.
Build or surrender
Mensch first set out his case on 15 October, in a LinkedIn essay titled ‘Build or surrender’. He called the idea of leaving frontier models to the US and China “a fundamental mistake”.
“It will end up costing Europe what remains of its place in the world and lock it into a regime of economic dependence unprecedented in the history of civilizations.” Arthur Mensch, Mistral
Europe followed the same playbook with the internet, he wrote, and now pays more than €260bn a year to American cloud and software providers. Whoever owns a closed model sets the terms, he argued: the price, the uses allowed and which versions stay available. Mistral plans to increase its computing capacity sixfold over the next year, and to raise more capital.
A Mistral spokesperson told Politico that walking away would be a “strategic mistake”. The Paris company was valued at €21bn in September.
Kutylowski made the same case in an interview with Politico.
“We need our own models, and we need the whole value chain of AI. We also need to capture the value that lies in the technology and in AI itself… Otherwise we’re just going to be leaving a large part of the value creation [to others].” Jarek Kutylowski, DeepL, to Politico
He pointed to software engineering. Companies there now spend part of their budgets on AI models rather than engineers’ salaries, he said. Without European models, that money flows to the US.
The case for applications
The other camp includes Ursula von der Leyen. The Commission president made her case in her State of the Union address in September, Politico reported.
“[Europeans don’t need to] be the ones who develop the frontier technology to be the ones who draw the greatest value from it.” Ursula von der Leyen, State of the Union, September 2026
Max Junestrand, chief executive of the Swedish legal AI firm Legora, helped start the argument. He has already said Europe should not build a frontier lab. He told Politico that Europe should work out how to “capitalize on the technology” instead.
“[Efforts to build a] competitive frontier lab right now feel like very wishful thinking.” Max Junestrand, Legora, to Politico
Politico reported on 13 October that von der Leyen and several prominent founders favour that shift. Europe should get value from AI through applications rather than chase the US labs, in their view.
Europe trails the US and China on the most advanced models. Training and running its own would take vast amounts of computing power. The Commission started work at the end of July on seven AI compute hubs, a €30bn project, Politico reported. Italy’s Domyn leads a consortium that is training a frontier model for all of Europe, in every EU language.