AI Now Institute: Europe's AI market is 'structurally captured'

The essay opens flat. "Across this four-part series, we have argued that Europe's AI ecosystem is subject to the gravitational pull of dominant US players," write Frederike Kaltheuner and Leevi Saari of the AI Now Institute in TechPolicy.Press. The market, in their reading, is "structurally captured," one where "start-ups like open-source model platform Hugging Face get bought by Nvidia" and "Germany's AI DeepL partners with AWS."

They put a price tag on the dependency. "Bain & Company estimates that making the spending on AI infrastructure profitable will require $2 trillion in annual revenue by 2030," they write. Meanwhile, "value flows to whoever controls the hardware that inference runs on (ASML, TSMC, Nvidia, the memory makers)."

Their central move is to argue Europe treats AI's trajectory as weather rather than a policy variable. "The AI market is already being shaped," they write, "by a small number of companies with unprecedented power who are shaping the market in their own interest. The question is not whether the market gets shaped, but by whom."

Four recommendations follow: take the market's uncertainty seriously, actively shape the market to be more open and interoperable, treat cloud dependence and AI sovereignty as the same problem, and decide what AI is actually for. On that last point, the authors contrast a "replacement logic," in which "LLMs replace workers," with augmentation, which "means adopting AI services only when and if they truly improve."

The load-bearing claim is structural. "This bundling between closed models and hyperscalers is what makes AI dependence different from the cloud dependence Europe has already known," they write. Two researchers we track posted the essay the day it ran.