BRUSSELS — Top tech entrepreneurs are lining up to denounce calls by European leaders to pull out of the race to build AI frontier models.

POLITICO last week reported that key leaders, including European Commission President Ursula von der Leyen and prominent tech founders, believe Europe should shift away from trying to catch U.S. frontrunners in developing the most advanced AI models, and focus instead on deriving value from the technology through applications.

“We need our own models, and we need the whole value chain of AI,” Jarek Kutylowski, CEO of German AI translation-focused research lab DeepL, told POLITICO in an exclusive interview. “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].”

The rift exposes how Europe is struggling to find its feet in the global AI race.

The continent trails the U.S. and China in developing the most complex and advanced models, which have become a security concern because they can exploit software vulnerabilities.

Yet pursuing its own capable models would require the EU to find the means to fund the enormous compute power required to train and run them. In recent weeks, Europe’s top tech executives and officials have been in a standoff, with some saying trying to match the U.S. and China no longer makes sense — while others argue that giving up on owning AI models again leaves Europe dependent on foreign technology. In the wake of POLITICO’s reporting, Arthur Mensch, the CEO of French AI giant Mistral, penned a widely shared post titled “Build or surrender,” in which he warned that giving up the frontier would “end up costing Europe what remains of its place in the world.”

Both Mensch and Kutylowski are seeking to counter abandoning the frontier with language Brussels understands: saying it risks a new area of tech dependency and a potential new kind of trade deficit with the U.S.

Europe will be saddled with the “leftovers” and a “big trade balance problem” if it only builds applications on top of foreign models, Mensch said on the livestreamed European Technology Network talk show on Tuesday. Loss of value Max Junestrand, CEO of Swedish AI legal service platform Legora, helped ignite the debate in an interview with POLITICO, declaring that efforts to build a “competitive frontier lab right now feel like very wishful thinking.”

Junestrand said the European public and private sectors should instead put their resources toward figuring out how to “capitalize on the technology rather than how do we try to catch up.”His remarks echoed von der Leyen, who said in mid-September during her State of the European Union address that Europeans don’t need to “be the ones who develop the frontier technology to be the ones who draw the greatest value from it.”

But those working on frontier AI pushed back against that narrative. The rebellion was led by Mistral, recently valued at over €21 billion, with a spokesperson telling POLITICO that abandoning the frontier would be a “strategic mistake.”Kutylowski singled out software engineering as one domain where a part of companies’ expenditures are now going toward AI models instead of engineers’ salaries. He argued that not having European models would mean part of the value flows to the U.S. Europe has several initiatives underway to ensure it still can compete at the frontier.

At the end of July, the Commission fired the starting gun on building 7 AI compute hubs, a €30 billion investment project. Italian Domyn leads a European consortium that is currently training a frontier AI model that will be available to all of Europe, in all EU languages.