It is 2031 and the United States and China are about to tear Europe apart. The US ploughed vast sums into datacentres and the EU did not. China built robots and Europe did not. American companies restructured their workflows around AI and fired people, while EU workers handed administrative tasks to Claude. Europe’s economy is a shambles. Populism is surging. The euro is wobbling. It looks like the end of the European Union.
That is the premise of “Europe 2031,” a speculative thought experiment penned by Brussels-based thinktankers Maximilian Negele and Alex Petropolous that has gone viral among EU policymakers. Published one day before the Trump administration blocked foreign access to Anthropic’s Fable model, the scenario now reads less like fiction and more like a forecast.
🔍 THE BOTTOM LINE
As The Guardian reports, the scenario has been read by members of the European Parliament and was raised in track 1.5 discussions between British and German officials. The authors say they feel “vindicated” — one of their core predictions, that the US would restrict global access to advanced AI models, briefly came true when the Commerce Department issued its export control directive against Anthropic. For New Zealand, the scenario is not about Europe at all. It is about what happens to a small, technologically dependent country when the great powers decide to wall off their AI.
What Changed: Fiction Became Policy Reference
The “Europe 2031” document is part of a burgeoning genre of AI doomsday scenarios that have gained surprising traction among policymakers. In 2025, “AI 2027” — a thought experiment culminating in a superintelligent AI killing humanity to make way for more datacentres — was reportedly read by US Vice-President JD Vance. In February, another speculative scenario about AI upending the US economy contributed to a stock market wobble.
What makes “Europe 2031” different is timing. The scenario was published just before the Trump administration’s export control directive against Anthropic, which classified Fable and Mythos as munitions and cut off foreign access. The authors had predicted exactly this kind of restriction. When it materialised within 24 hours of publication, the scenario went from thought experiment to policy reference document.
The G7 AI summit that followed made the scenario’s central tension explicit: Europe wants access to American frontier models; America has decided to restrict that access. Macron’s meeting with Anthropic and OpenAI at the G7 was an attempt to negotiate a way around the controls. It did not produce a breakthrough.
Context: The Scenario’s Specific Claims
The scenario unfolds from the perspective of a fictional Brussels staffer who visits San Francisco and is struck by the 70-80 hour working weeks and the conviction among tech workers that everything is about to change. Back in Europe, she fails to convince her bosses that AI is not a bubble.
The specific claims are grounded in real deals — though several have since collapsed. The scenario references the $100 billion OpenAI-Nvidia deal (the biggest AI deal of 2025, which evaporated in February), the $300 billion OpenAI-Oracle agreement (now reportedly doubtful), and bulldozers breaking earth in Texas for an AI datacentre (OpenAI has since pulled out of the flagship project). The authors are sanguine about these failures. “I wouldn’t rule out that there’s some exuberance and that one or two AI companies might go bankrupt,” Negele told The Guardian. “But what we wanted to get across is a general feel for a version of what we think will happen.”
The core argument is structural: AI development requires an exponential feedback loop between models, compute power, and data. If Europe does not secure its own compute capacity, it becomes a premium consumer of foreign AI services rather than a creator. The scenario imagines the US monopolising 70% of the world’s compute, with Europe’s economy gasping because its companies have not adopted AI fast enough.
NZ Angle: The Sovereignty Question We Have Not Answered
New Zealand faces the same structural question as Europe, with less leverage and fewer options. We have no domestic frontier model capability. We depend on US cloud infrastructure (AWS, Azure, Google Cloud) and US-developed AI models (Claude, GPT, Gemini) for everything from agriculture tech to government services. Five Eyes membership provides intelligence sharing — it does not provide immunity from export controls, as the Anthropic ban demonstrated.
The “Europe 2031” scenario’s relevance to NZ is not in the specifics (we are not the EU, we do not have the euro, we are not a regulatory superpower) but in the pattern: a small, technologically dependent economy discovers that its AI supply chain is subject to the policy decisions of a larger ally, and the discovery comes too late to do anything about it. The scenario’s proposed solution — massive coordinated investment in local datacentres and compute capacity — is harder for NZ than for the EU, because our market is smaller and our power grid is already constrained. But the principle is the same: if you do not control the compute, you do not control the AI.
The Other Side: Thought Experiments Are Not Data
The scenario’s critics — and the authors themselves — acknowledge its limitations. The thought experiment is built on assumptions about AI progress whose outcomes are uncertain. The $100 billion OpenAI-Nvidia deal collapsed. The $300 billion OpenAI-Oracle deal is doubtful. OpenAI’s own financials, as the leaked audited losses showed, are billions of dollars underwater. The “bulldozers in Texas” moment was followed by OpenAI pulling out. The mega-deals the scenario uses as evidence of American AI dominance are themselves fragile.
There is also a question about whether the scenario’s policy prescription — build more datacentres, deregulate faster — is actually the right response, or whether it is a panic-driven overcorrection. Negele himself admits: “People hate AI in general. People hate datacentres. They destroy the landscape. They support big tech. It’s a very unpopular policy.” Building datacentres to avoid being a technological colony is a harder political sell when the datacentres themselves are politically toxic.
The Bigger Picture: The Sovereignty Scramble
The “Europe 2031” scenario is one data point in a larger pattern: the AI sovereignty debate has moved from thinktank papers to active policy in less than 18 months. The EU’s Virkkunen interview with Bloomberg confirmed that the EU is actively discussing reducing reliance on non-European technology providers in critical sectors. The FT analysis of Anthropic’s export control situation noted that the company warned about AI dangers more than rivals — and was then punished for it.
For NZ, the sovereignty question is not abstract. If the US can cut off Anthropic’s models from the world by executive order, it can cut off any US-hosted AI service. The policy response is not autarky — we cannot build a domestic Anthropic — but resilience: multi-jurisdictional suppliers, open-source alternatives, and a regulatory framework that treats AI model access as critical infrastructure subject to supply chain risk.
❓ FAQ
Q: What is “Europe 2031”? A: A speculative fiction scenario written by Brussels thinktankers Maximilian Negele and Alex Petropolous, imagining Europe’s technological decline by 2031 due to insufficient AI investment. It has been read by EU parliament members and discussed in UK-German diplomatic channels.
Q: Why did it go viral? A: It was published one day before the Trump administration blocked foreign access to Anthropic’s Fable model — one of the scenario’s core predictions. The timing made it feel prophetic rather than speculative.
Q: Does this matter for New Zealand? A: Yes. NZ depends on US-hosted AI models and US cloud infrastructure. If the US can cut off AI model access by executive order, NZ’s AI supply chain is subject to US policy decisions made without our input. Five Eyes membership does not protect against export controls.
Q: Are the scenario’s predictions accurate? A: Mixed. Several mega-deals it cites as evidence of US AI dominance (OpenAI-Nvidia $100bn, OpenAI-Oracle $300bn) have since collapsed or are doubtful. OpenAI’s own finances are billions underwater. The scenario’s direction may be right even if its specifics are shaky.
🔍 THE BOTTOM LINE
“Europe 2031” is fiction that became a policy reference because reality caught up with it within 24 hours. The Trump administration’s export control ban on Anthropic’s models was the scenario’s first prediction made real. For New Zealand, the lesson is not in the European specifics but in the structural pattern: a small, technologically dependent country that does not control its AI compute supply is subject to the policy whims of the countries that do. The sovereignty question is no longer theoretical. It is operational.
