A viral thought experiment circulating among European policymakers paints a grim picture: by 2031, a tech-dominant United States and a robot-building China carve up a sidelined Europe, its economy in ruins and its AI ambitions stillborn.Titled Europe 2031 and published by the Brussels-based Arq Foundation, the scenario landed one day before the Trump administration reportedly blocked “foreign nationals” from using Anthropic’s Claude Fable model—a coincidence its authors now treat as eerie validation. In the week of G7 talks that followed, the piece went viral, read by members of the European Parliament and raised in British-German official discussions.
How the scenario sees Europe falling behind
The story follows Caroline Dubois, a fictional Brussels policy worker, and her German founder friend Christian Vogt, who runs a startup in San Francisco. The setup is simple. The US pours hundreds of billions into datacentres while the EU offers a tepid investment package. China floods the market with cheap humanoid robots. American firms restructure around AI and cut staff; European workers take long lunches and hand grunt work to AI agents they’re often barred from using properly.From there it spirals. America ends up controlling 70% of the world’s compute. AI-powered cyberattacks shred European firms. Unemployment climbs, the euro wobbles, and populism surges. By the final act, EU officials are scrambling to leverage their one remaining card—Dutch lithography giant ASML, indispensable to chipmaking—into concessions from Washington or Beijing. They fail. American spyware has already learned which officials are having affairs and what they fear most.Maximilian Negele, a former Rand researcher who left his job to work on the project, said the gap between Brussels and Silicon Valley felt like watching “a slow-moving car crash.” The authors’ prescription is blunt: Europe needs to build datacentres faster, ideally in deregulated AI zones with streamlined power and planning.
Why critics say the numbers don’t quite hold up
There’s a catch. Several of the eye-popping deals the scenario name-checks have already collapsed. The $100 billion OpenAI-Nvidia agreement, last year’s biggest AI deal, evaporated in February. The $300 billion OpenAI-Oracle pact looks shaky, with reports suggesting OpenAI is still billions underwater. Even the Texas bulldozers the piece references may have gone quiet, since OpenAI pulled out of that flagship project.The authors aren’t fazed. Negele concedes some “exuberance” and admits one or two AI companies might go bankrupt, but insists the broad direction holds. The piece also pre-empts its own critics—its fictional officials worry AI is a bubble too, and end up tragically wrong.Not everyone is fully sold. Spanish MEP Nicolás Casares thinks the authors are deliberately ringing the alarm louder to grab attention. Still, he sees the point. Buying the narrative that Europe needs endless datacentres it may never get to use, he warns, is “crazy.”
