Europe 2031, AI sovereignty, compute dependency, and European agency

A widely discussed scenario text sketches Europe in 2031: dependent on American compute, without a leading AI model of its own.

“Europe 2031” is not a conventional study, but a fictional short story. It follows an EU official in Brussels who watches Europe underestimate the speed of AI development, defer warnings, and lose more and more options in the technological contest between the US and China.

That format is what makes the text powerful. It turns an often abstract AI debate into a plausible story about power, dependency, and lost agency.

The visible part of the problem is compute. “Europe 2031” puts Europe’s current share of global AI compute at around five percent. That order of magnitude shows the problem: Europe currently plays hardly any independent role in frontier AI and compute.

There is another issue that is not at the center of the story: the value base of the models.

AI systems are shaped by training data, safety rules, product decisions, and cultural assumptions. If the most important systems come from Silicon Valley or China, companies and public administrations are not only dependent on the technology. As long as AI mainly writes text, the risk remains manageable. Once it controls processes and increasingly prepares or makes decisions autonomously, these learned evaluation standards become operationally relevant: what gets optimized, slowed down, or treated as a risk.

Still, the scenario does not have to materialize. The next major stage of AI development does not necessarily have to come from today’s language models. World models, robotics, and industrial AI could shift the playing field.

That is where factories, machines, process knowledge, industrial data, energy supply, and physical execution matter. These are not European side issues. They are European strengths.

The question is whether Europe can turn them into agency in time: with compute, capital, faster permitting, and the ambition not to be merely a customer in frontier AI.

Anyone who wants to understand why AI sovereignty is more than a political slogan should read this short story: https://europe2031.ai/