Interactive forecast 'Europe 2031' projects a 12.3x US AI compute advantage over Europe by 2031

Everyone smart eventually arrives at this idea. There are many arguments as to why it wouldn't work. The fundamental blocker is 1) brainwormed Atlanticist ideology and 2) reasonable fear of American retaliation. The technical blocker is ASML's US dependencies. Were the EU serious, diversifying away from the latter would be a subject of analysis. Surreptitious stockpiling, timelines, opportunity cost, possible Chinese suppliers, what IP they'd need for the last mile that you can afford to sell, how would you negotiate this, how would you support ASML and the Dutch in the transition period, crucially: what do you ask from China in return for this sacrifice – and what do they give you upfront.

But the mere process of deliberating on this plan (and Americans and Brits constantly produce vastly more unhinged plans they do not act upon) would probably do more for the EU's strategic position than any non-credible noises about "Trade bazooka". Alas.