Europe’s AI Squeeze: Why Brussels Sees a Sovereignty Risk in Falling Behind the US and China

Europe’s race to keep up with artificial intelligence has taken on a new, more existential tone in Brussels. A speculative policy exercise imagining a fractured continent by 2031 has spread quickly through the EU bubble, not because it offers a precise forecast, but because it taps a deep anxiety: what happens if the world’s biggest AI platforms, chips and datacentres are all controlled elsewhere?

The scenario, titled Europe 2031, imagines a future in which the United States and China pull so far ahead in AI infrastructure that Europe becomes economically weaker, politically more brittle and strategically dependent. In the story, American firms pour money into computing infrastructure, Chinese companies dominate robotics and EU businesses fall behind because they fail to adopt AI at scale. The result is a continent struggling with cyberattacks, sluggish growth and rising political tension.

That premise may sound like dystopian fiction, but it has landed in real policy conversations. The paper has circulated among members of the European Parliament and, according to its authors, has even come up in informal discussions involving British and German officials. Its timing helped too: it surfaced just before a move by the Trump administration to restrict access by foreign nationals to a high-profile AI model from Anthropic, a development the authors say underlined their core message about dependency and control.

What makes the debate so potent is that it is not really about one fictional future. It is about a very current fear in Europe: that the continent could become a consumer, not a producer, of the foundational technologies that shape economics, defence, security and public life.

At the heart of the argument is compute — the expensive mix of chips, power, buildings and networks that powers modern AI systems. Europe has long relied on external suppliers for critical digital infrastructure. Now, as AI systems become more resource-intensive and more central to industry and government, that dependence feels more consequential than ever.

Why a fictional scenario hit such a nerve

Europe 2031 is not a formal policy brief. It is a thought experiment designed to provoke. But that may be precisely why it has gained traction. The scenario is written in the style of a political warning, using characters and a near-future narrative to illustrate how Europe might be outpaced if it underinvests in AI while the US and China scale aggressively.

The story follows a fictional Brussels official, Caroline Dubois, who travels to San Francisco and encounters a tech culture defined by long hours, relentless execution and a belief that AI will transform everything quickly. Back in Brussels, she tries to persuade sceptical colleagues that Europe needs to move faster. Her efforts fail. The continent hesitates, regulation remains cautious and investment stays modest.

From there, the scenario worsens. The US, according to the fictional timeline, spends massively on AI infrastructure and captures a dominant share of the world’s compute. European firms lag in adoption, leaving the economy vulnerable to automation abroad and disruption at home. Cyberattacks intensify. Political instability follows. By the end, Europe is weakened both economically and geopolitically.

The appeal of the piece lies less in its plot than in its emotional force. It converts a technical policy debate into a drama about dependence, control and power. That is a powerful framing in a continent where “strategic autonomy” has become one of the most frequently invoked phrases in technology policy.

Who wrote the scenario and why

The project was developed by Brussels-based thinkers associated with Arq Foundation, a group that describes itself as neither an advocacy organisation nor a venture-backed startup. It does not publicly identify its funders, which adds a layer of opacity to a paper explicitly about Europe’s need for more control over essential infrastructure.

One of the contributors, Maximilian Negele, said the work grew out of frustration with the distance between debates in Brussels and the pace of change in Silicon Valley. He previously worked at Rand and left that role this year to focus on the project. In his telling, the issue is not just policy inertia, but a communication gap.

Negele has argued that the lack of shared language between Brussels and San Francisco makes it harder for European institutions to understand how quickly the AI landscape is changing, describing the situation as a slow-moving collision rather than a clean policy failure.

That feeling of disconnect is central to the project’s message. The authors are not simply claiming that Europe is behind. They are saying Europe may be structurally unable to perceive how far behind it is until it is too late.

The other co-author, Alex Petropolous, has made a similar case: Europe should not only regulate AI, but also build the physical infrastructure required to host it. Their argument is that compute is scarce, capital-intensive and geopolitically strategic. In that context, each datacentre built in Europe is one more piece of leverage the continent can keep close to home.

How the scenario maps onto current AI trends

Although the narrative is fictional, it draws heavily on real-world developments. The United States has seen a burst of investment in datacentres and model development. Major AI companies have announced or pursued huge infrastructure deals, and the race to secure electricity, land, chips and cloud capacity has become one of the defining industrial stories of the AI era.

China, meanwhile, has made robotics and industrial automation a central part of its technology strategy, reinforcing fears in Europe that the competitive gap could widen not only in software but across manufacturing and logistics as well.

Europe’s role is more ambiguous. It remains home to strong research institutions, world-class industrial firms and key supply-chain players such as ASML, the Dutch lithography company that is essential to advanced semiconductor production. But its fragmented market, slower permitting process and more cautious capital environment make large-scale AI infrastructure harder to build.

The scenario’s most alarming claim is that Europe could become a passive user of AI systems designed, financed and governed elsewhere. In that world, the continent would still pay for AI, but would have little influence over where the profits go, what the systems optimise for or who gets access to the underlying infrastructure.

The role of compute in the new AI economy

Compute has become the new bottleneck in AI. Training frontier models requires enormous amounts of power, specialised chips and cooling infrastructure. That has turned datacentres into strategic assets rather than just industrial real estate.

The policy implication is straightforward: if Europe does not expand its own compute base, it risks becoming dependent on foreign providers for access to the most advanced systems. That dependence could affect everything from government services and defence applications to the competitiveness of European start-ups and industrial firms.

In the authors’ view, the market for datacentres is not elastic enough to assume Europe can simply buy capacity later if it needs it. Once global supply is allocated, the best-located, best-connected and best-funded projects are likely to be built first. Their conclusion is that Europe should move early, not just because it wants innovation, but because it wants leverage.

What the fictional timeline says would happen

The scenario lays out a chain reaction. The US invests heavily, Europe hesitates, and over time the infrastructure gap becomes a productivity gap. Companies in America use AI to redesign workflows and reduce labour costs. European firms, slower to adopt, lose competitiveness. That drag eventually shows up in growth, employment and public finances.

In the fictional version of events, Europe’s weaker economy then compounds other political pressures. Populism rises. The euro comes under strain. Cyberattacks target businesses and public institutions. As AI-powered tools become more capable, they are also used for espionage and manipulation.

One of the more dramatic elements of the story is that EU officials attempt to negotiate around their weakness by using ASML as a bargaining chip. Since the company’s equipment is vital to advanced chipmaking, it becomes one of the few hard assets Europe can still wield in diplomatic negotiations. But by then, in the scenario, the moment has passed.

The result is a continent that has not just fallen behind technologically, but has also lost bargaining power in the geopolitical competition around AI.

Key milestones in the scenario

Why policymakers are paying attention

One reason the paper has resonated is that it reflects a broader shift in Brussels. European policymakers increasingly see AI not just as an innovation policy issue but as a strategic infrastructure question. That includes chips, power grids, cloud capacity, datacentre permits, talent pipelines and cross-border investment rules.

The European Parliament has already been debating how to encourage domestic capability without completely abandoning the bloc’s caution around safety, competition and privacy. The rise of American and Chinese AI giants makes that balancing act more difficult. If Europe wants to remain technologically relevant, the argument goes, it may need to accept faster development and more permissive infrastructure planning than it is used to.

That is where the debate becomes politically sensitive. Critics of the “build, build, build” approach worry that Europe could spend billions on infrastructure that mostly benefits foreign companies. Supporters argue that refusing to build would be worse, because it would lock the continent into dependence while allowing others to capture the economic gains.

A Spanish member of the European Parliament, Nicolás Casares, said the scenario may exaggerate the urgency, but it has helped spotlight real questions about who owns Europe’s AI infrastructure and who ultimately profits from it.

Casares also questioned whether the continent should automatically assume that hosting foreign AI datacentres is the right strategy if the resulting infrastructure remains controlled by non-European companies.

The tension between alarm and reality

Not everyone is convinced by the most dramatic claims in the scenario. Some of the headline numbers it references have already shifted or become uncertain. Several of the major deals used to illustrate American AI momentum have either changed shape or lost credibility. The industry is still in flux, and not every ambitious infrastructure announcement survives contact with financial reality.

That does not necessarily weaken the broader warning. If anything, it underlines a more complicated point: AI investment is speculative, but the strategic race around it is real. Even if some companies overextend or projects are delayed, the countries and regions that accumulate the most chips, talent and power capacity are likely to hold the strongest position when the market settles.

The authors acknowledge that not every AI company will succeed and that some of today’s exuberance may prove excessive. But they say Europe should not use the possibility of a bubble as an excuse for inaction. Their view is that the continent is at risk of confusing volatility with safety.

That distinction matters. A market correction may hurt investors. A structural lack of AI infrastructure could reshape an entire continent’s industrial future.

What could go wrong with the warning itself

• It may overstate how quickly frontier AI becomes economically transformative.
• It may underestimate political resistance in the US to more datacentres and power use.
• It may assume European adoption lags permanently rather than catching up.
• It may treat infrastructure as destiny, when regulation, talent and market design also matter.

Even so, the scenario’s authors believe the bigger risk is complacency. They argue that Europe is already in danger of treating AI as a niche technology policy topic rather than a central issue of industrial strategy and sovereignty.

The case for datacentres, and the political backlash against them

For Negele and Petropolous, the answer is not abstract rhetoric about competitiveness. It is physical infrastructure. Europe should build more datacentres, faster, and in places where planning and energy rules can be streamlined.

They even favour special “AI zones” where regulatory and permitting hurdles would be reduced to speed up construction. The logic is that if global demand for compute is finite in the short term, Europe must compete aggressively for a share of the next wave of build-outs.

That idea is controversial for obvious reasons. Datacentres are energy-hungry, visually intrusive and politically unpopular in many places. Communities often oppose them because they are seen as using scarce land and electricity while offering relatively few local jobs.

The authors of the scenario are aware of this backlash. They argue that the dislike of datacentres is itself a reason for governments to take the issue seriously rather than assume the market will solve it on its own. In their view, the political unpopularity of the infrastructure is one more reason it may not be built quickly enough without state intervention.

Petropolous has framed the issue as a contest over a limited pool of global datacentre capacity, suggesting that the crucial question is not whether these facilities are built at all, but where they are built and who benefits from them.

How the Trump administration episode sharpened the debate

The scenario gained an extra layer of relevance when the Trump administration moved to block foreign nationals from using a highly regarded AI model developed by Anthropic. Even if temporary or narrow, the step reinforced a central fear in Brussels: access to advanced AI may not remain open or universal.

If countries can be cut off from frontier systems during a geopolitical dispute, then access itself becomes a weapon. That is a profoundly different model from the open internet era, when Europe could buy most digital services from abroad without worrying that the service might one day be withdrawn for strategic reasons.

For EU policymakers, that reality raises uncomfortable questions. Should Europe rely on US firms for core digital functions if those firms may be constrained by US politics? Should public administrations and critical industries be dependent on foreign models that can be restricted, priced up or switched off? And if Europe wants independence, how much public money should be used to build alternatives?

These questions sit at the intersection of industrial policy, national security and economic planning. They also explain why a fictional scenario has triggered such a serious response. It is not just about AI as technology. It is about AI as infrastructure of power.

Europe’s strategic dilemma

Europe faces a choice that is becoming clearer with each new AI milestone. It can continue to prioritise caution, market fragmentation and regulatory oversight, hoping that innovation will emerge elsewhere and remain accessible. Or it can try to build the compute, energy, chip and cloud stack needed to host frontier AI on European soil.

The first approach is less risky in the short term and more consistent with European political instincts. The second is more expensive, more contentious and more ambitious. It may also be the only path that preserves real autonomy if AI becomes as economically decisive as its advocates predict.

That is why the debate is likely to intensify. As companies automate more tasks and governments become more reliant on digital systems, the gap between hosting AI and simply using AI will matter more. Europe may be able to regulate imported technology, but if it does not control enough of the underlying infrastructure, its influence could remain limited.

In that sense, the scenario’s real value is not as prophecy. It is as a stress test. It forces European leaders to imagine what dependency might look like when AI is embedded in finance, logistics, defence, administration and manufacturing.

Whether that future arrives exactly as the paper imagines is almost beside the point. The more important question is whether Europe wants to find out the hard way what happens when the centre of technological gravity moves elsewhere.

What happens next

For now, the “Europe 2031” warning is doing what its authors hoped: sharpening the conversation. It has given politicians, officials and analysts a vivid story to debate, even if they disagree on the specifics.

Some will dismiss it as alarmism. Others will see a useful corrective to Brussels’ tendency toward delay. But the scenario’s rapid spread suggests that Europe’s AI anxiety is no longer a fringe concern. It is moving closer to the mainstream of policy debate.

The central issue remains unchanged: if the US and China are building the next generation of AI at scale, and Europe is not, then Europe must decide whether it is comfortable being a dependent market or whether it wants to invest in the infrastructure that gives it leverage, resilience and choice.

That is the real message behind the fictional disaster. The danger is not just that Europe could lose an AI race. It is that it could wake up to discover the race was about sovereignty all along.

Whether Europe chooses to respond with more funding, faster permits or a more assertive industrial strategy will determine whether this warning remains speculative fiction or becomes a retrospective description of missed opportunity.

superintelligencenews@gmail.com