Summary of
Instead of radical measures that deepen US dependence, Europe should consider radical measures that reduce it (state investment in open-weight models, procurement policy, interoperability rules) — without sacrificing democracy, labor rights, or climate targets to a hypothetical.
What Europe 2031 says: A scenario written by investors, AI researchers, and think tankers predicting that if Europe doesn't secure "frontier" AI model access, it collapses economically and politically by 2031, forced to become a US protectorate, turn to China, or become irrelevant.
Its main "fix": Build massive AI computing power in partnership with US tech giants (Amazon, Microsoft, Google), form a coalition of countries to negotiate access, loosen labor laws, and pivot to robotics — all urgently, bypassing normal democratic caution.
Kaltheuner's core objection: This is one possible future dressed up as inevitable fact. It rests on a chain of unproven assumptions — that AI will transform society within years, that the "frontier" gap is real and permanent, that using top AI models makes countries richer, and that Europe can somehow win in robotics while still depending on the same US firms. None of these are settled.
Key practical problems she flags:
The compute build-out demanded (55 GW by 2031) would use more electricity than all of Germany — wrecking climate goals
It would deepen, not reduce, Europe's dependence on US hyperscalers
The "security guarantees" the scenario proposes are worthless — the scenario itself shows the US cutting off access when convenient
Cheaper "good enough" open-weight models (like China's GLM 5.2) already challenge the idea that only the top models matter