A group of AI researchers, think-tank figures and investors has published a future scenario titled „Europe 2031″ that plays out Europe’s gradual loss of relevance in global AI development. The initiative explicitly sees itself as a warning — coupled with an appeal to take political countermeasures now.

What it is about

The scenario narrates, over a period of roughly five years, how Europe — according to the authors — slides into economic and political dependence. The starting point, they argue, lies in three misjudgements from 2025: Europe underestimated how quickly AI was advancing, how profoundly it would change everything, and it overestimated its own ability to close the gap again.

The authors trace the trajectory through to 2031. In doing so they point to real milestones such as the Chinese model DeepSeek R1, the Paris AI Action Summit with its announced €200 billion fund, and what they read as the disappointing reception of GPT-5. Their argument: Europe repeatedly drew the wrong conclusions — for instance, that computing power was losing importance, or that this was an „AI bubble”.

For the years from 2026 onward, the authors develop a fictional but, by their own account, plausible scenario: a ransomware wave triggered by a freely available frontier model, a rationing of computing power by the United States along a tiered, country-by-country system, the buy-up of struggling European carmakers and machine-tool manufacturers by US corporations, as well as rising government debt and centrifugal forces within the EU. At the end of the scenario, Europe is left with only three poor options: to become an American protectorate, to turn to China, or to wither away in isolation.

The authors stress that the failure they describe stems from incentives and institutions, not from individuals or ill intent. It is precisely those strengths that made a union of 27 states possible — consensus and careful procedure — that, under time pressure, would lead to uncomfortable decisions being deferred.

What the initiative demands

The authors consider Europe’s response so far to be ten to a hundred times too small and, on top of that, misdirected. „Sovereignty,” they argue, too often means settling for second-rate solutions; what is needed instead is real bargaining power. Concretely, they propose five points:

• Investment in computing power and infrastructure: massive investment in energy, semiconductors and data centres, along with dedicated economic zones, a targeted energy policy and significantly shorter permitting procedures. Cooperation with American providers makes sense, but only on terms that keep the infrastructure in European hands.
• A coalition of like-minded „AI middle powers”: beyond the EU, with countries such as Norway, the United Kingdom, Canada, Japan and South Korea, in order to build collective bargaining power and, where appropriate, to mediate between Washington and Beijing.
• Adapting labour markets: preparing labour markets for AI along the lines of the Danish „flexicurity” model, rather than preserving jobs in their current form at any cost.
• Building on strengths in robotics and industrial AI: scrutinising foreign investment in domestic manufacturers and opening up industrial data for home-grown developers.
• A positive vision for the age of AI: since reforms cannot be pushed through on fear of loss alone. Formulating this vision is the task of civil society and political leadership alike.

Who is behind it

According to its own statements, „Europe 2031″ is the work of a small group of people from AI research, think tanks and investment. The named contributors include Daan Juijn (research director of the Brussels-based Arq Foundation), Stan van Baarsen (Delta Institute, co-author of the Dutch national AI plan), Judith Dada (General Partner at Visionaries Club), Lily Stelling (Oxford Martin AI Governance Initiative), Philip Fox (KIRA Center, co-author of the International AI Safety Report), Alex Petropoulos (Arq Foundation) and Michiel Bakker (MIT, Google DeepMind). All, they say, contributed as private individuals; the views expressed do not necessarily reflect the positions of their organisations. Among those who helped with the narrative form is the author Tom Chivers; as inspiration the authors cite the scenario „AI 2027″, though „Europe 2031″ is not connected to it.

The initiative does not classify itself as a pure growth agenda: it points emphatically to the risks of advanced AI — from cyber and bio misuse to extreme concentration of power and, in the worst case, the permanent disempowerment of humanity. A stronger Europe, the argument goes, could form a responsible counterweight to the major powers.

How it is financed

According to those involved, the bulk of the work was done in their spare time. The Arq Foundation, where two of the co-authors work, contributed limited funds, for instance for the website and translations. They received no money from the AI industry, they say, and pursue no profit motive.

Aus Datenschutz-Gründen ist dieser Inhalt ausgeblendet. Die Einbettung von externen Inhalten kann in den Datenschutz-Einstellungen aktiviert werden: