No Leverage, No Sovereignty

Europe still believes regulation is power.

It is not.

Regulation is power only when the regulated party needs you more than you need them. A law without leverage is not law in the geopolitical sense. It is a request with better formatting.

This is the uncomfortable point Europe keeps avoiding in the AI debate. We can write rules. We can announce frameworks. We can fine American technology companies. We can say, with great moral confidence, that we do not want monopolies deciding the shape of our digital lives.

Fine.

But what happens when the technology becomes something Europe needs every day? What happens when frontier AI becomes essential infrastructure for software, education, security, medicine, science, business and defense? What happens when access to the model matters more than access to the European consumer?

Then the balance flips.

Europe is no longer the gatekeeper. Europe becomes the customer. And customers do not enforce sovereignty. They negotiate access.

The Apple illusion

The Digital Markets Act gave Europe a taste of regulatory power. In April 2025, the European Commission fined Apple €500 million for breaching its anti-steering obligations and fined Meta €200 million over its consent model. The Commission presented this as proof that Europe could discipline Big Tech.

In one sense, it can.

Apple wants access to the European market. Meta wants access to European users. The market is still large enough, wealthy enough and legally coherent enough to matter. Europe can impose costs. It can slow companies down. It can force interface changes. It can make executives take Brussels seriously.

But this creates a dangerous illusion.

Because Apple is not the same as frontier AI.

With Apple, Europe is regulating consumer access to an already mature platform. With AI, Europe may soon be asking for access to the platform itself: the models, the chips, the compute, the cloud, the cybersecurity systems, the research tools, the agentic workflows, the defense applications.

That is a very different negotiation.

If you depend on someone else's system to keep your economy productive, your institutions secure and your companies competitive, then your power to regulate that system becomes fragile. You can still complain. You can still legislate. You can still hold hearings and publish statements. But underneath the noise sits a brutal dependency.

You need them.

And they know it.

Europe 2031

The scenario essay Europe 2031: What getting AI wrong means for us captures this danger with unusual clarity. It imagines a Europe that keeps talking about sovereignty while losing the material basis for it: compute, chips, frontier models, energy, capital, talent and strategic leverage.

Its central warning is simple: Europe may lose the ability to shape its own future.

That sounds dramatic until you look at the actual substrate of power. AI is not just another software sector. It is becoming the layer through which work is coordinated, code is written, science is accelerated, cyber defense is conducted, companies are reorganized and military decisions are supported.

If that layer is owned elsewhere, sovereignty becomes theatrical.

Europe 2031 points to the compute gap because compute is not merely technical infrastructure. It is political gravity. The continent that controls the data centers, chips, power supplies and frontier models controls the tempo of the next economy. The continent that does not control them negotiates from need.

That is where Europe is drifting.

Not because Europeans are stupid. Not because Brussels is evil. The problem is worse. Europe is mistaking procedure for agency.

It believes that if something is morally correct and legally articulated, it has already become real.

But reality does not care about articulation. Reality cares about power, execution and dependency.

The InvestAI gap

Europe knows the problem exists. That is why Ursula von der Leyen announced InvestAI, a plan to mobilize €200 billion for artificial intelligence, including €20 billion for AI gigafactories.

Good.

But here again Europe risks confusing announcement with leverage.

Money that is partly repackaged, partly aspirational and partly dependent on private capital is not the same as bulldozers, power contracts, chips, permits, talent, energy and deployed systems. A press release is not a data center. A strategy is not a model. A vision is not a bargaining chip.

The question is not whether Europe can produce ambitious language. It can. The question is whether Europe can build disruptive capacity fast enough to make others need Europe.

That is the only sovereignty that matters.

Leverage comes from being indispensable. ASML has leverage because the world needs its machines. Apple has leverage because it owns a platform hundreds of millions of people use every day. Nvidia has leverage because its chips sit inside the AI boom. OpenAI, Anthropic, Google and xAI have leverage because their models increasingly shape what individuals, companies and governments can do.

Europe, too often, has position papers.

There is nothing wrong with rules. A civilization without rules becomes primitive. But rules without power become moral decoration.

Consumers are not bargaining chips

Europe often tells itself that 450 million consumers are enough.

That argument worked better in the internet era. If you wanted to sell phones, software, ads, media or subscriptions, Europe was too large to ignore. Brussels could set conditions because companies wanted distribution into the European market.

AI may not work like that.

If frontier AI becomes compute-constrained, security-sensitive and strategically controlled, access may become selective. The most capable models may not be released everywhere at the same time. Governments may influence who gets access. National security arguments may override market logic. The best systems may be reserved for trusted partners, defense ecosystems, domestic companies or regions with reciprocal value.

At that point, Europe cannot simply say: but we have consumers.

Consumers are not leverage if they are dependent consumers.

A hungry customer does not dictate the menu.

If European banks, hospitals, universities, defense systems, startups and public institutions all need foreign AI systems to remain functional, then the foreign provider has the stronger hand. Europe may still have legal tools, but every hard enforcement action will carry a hidden threat: what if access becomes worse, slower, more expensive or more restricted?

This is how dependency disciplines the regulator.

Not through one dramatic surrender. Through small adjustments. Exceptions. Delays. Interpretations. Safe harbors. Strategic compromises. A little more flexibility here. A little less enforcement there. A quiet understanding that the rules can be firm only as long as access remains available.

This is what toothless sovereignty looks like. It still speaks in a deep voice. It still holds press conferences. It still says values. But everyone in the room knows the teeth are borrowed.

Build disruption first

The missing European instinct is disruption.

Europe wants to regulate monopolies without building anything monopolies need. It wants to discipline platforms without producing platforms. It wants strategic autonomy without strategic assets. It wants the moral authority of power without the discomfort of building power.

That does not work.

If you want to negotiate with the world, you need bargaining chips. Not slogans. Chips.

Compute is a bargaining chip. Energy is a bargaining chip. Talent density is a bargaining chip. Capital that can move at speed is a bargaining chip. Frontier research is a bargaining chip. Defense capability is a bargaining chip. Industrial scale is a bargaining chip. Companies that create global dependency are bargaining chips.

Disruption is a bargaining chip.

The United States can enforce its will because it owns platforms, capital markets, clouds, models, chips and military protection. China can enforce its will because it owns industrial scale, supply chains, manufacturing depth and strategic patience. Apple can enforce its will because it owns a closed ecosystem that people and developers cannot easily abandon.

Europe has opinions.

Opinions matter. Values matter. Law matters. But none of them float above material reality. If you cannot enforce your preference, your preference becomes commentary.

And this is where Europe has become dangerously comfortable: as the most sophisticated commentator in the room.

The Draghi diagnosis

The Draghi report on European competitiveness made the same structural point in institutional language. Europe can no longer rely on the conditions that supported its old prosperity. Productivity is weak. Energy is expensive. Demographics are unfavorable. The digital gap matters. The green and digital transitions require levels of investment and coordination Europe has not yet shown it can deliver.

But even the Draghi frame can become another European ritual if it remains a report rather than a rupture.

Europe is brilliant at diagnosis. It is less brilliant at metabolizing diagnosis into force.

It can name the gap. It can convene panels about the gap. It can fund studies about the gap. It can produce a careful roadmap for narrowing the gap.

Meanwhile, someone else builds the thing that makes the gap irrelevant because the game has already moved.

That is the tempo problem.

AI does not wait for institutional digestion. It compounds. Models improve. Agents spread. Workflows change. Security assumptions shift. Capital concentrates. Talent follows momentum. Infrastructure takes years to build, but dependency can form much faster than that.

By the time Europe reaches consensus, the negotiation may already be over.

This is not anti-regulation

The easy misunderstanding is to read this as an argument against regulation.

It is not.

Civilization requires regulation. Markets need boundaries. Technology needs law. Human beings should not be reduced to behavioral data, productivity units or passive consumers inside machine economies. Europe is right to care about dignity, privacy, competition and democratic oversight.

The problem is not that Europe wants to regulate.

The problem is that Europe wants to regulate from a position of insufficient power.

Good regulation channels strength. Bad regulation compensates for weakness. One shapes reality. The other creates the feeling of action while dependency grows underneath.

If Europe wants humane AI, it must build the power to insist on humane AI. If Europe wants open markets, it must build companies capable of competing in those markets. If Europe wants privacy, it must build trustworthy infrastructure people actually use. If Europe wants strategic autonomy, it must control enough of the strategic substrate to make autonomy more than a word.

You cannot outsource the future and then regulate it back into alignment.

The sovereign human and the sovereign continent

This is not only a geopolitical problem. It is the same pattern that plays out in human life.

A person who does not act loses agency. A company that does not build loses relevance. A continent that does not create leverage loses sovereignty.

The language changes, but the structure is identical.

You can have beautiful values and still be dependent. You can have correct opinions and still be powerless. You can know what should happen and still be unable to make it happen.

Truth requires contact with reality. Curiosity requires direct engagement with the tools that are changing the world. Beauty requires the courage to shape, not merely critique.

Europe's deeper failure is not technical. It is anthropological. It has trained itself to believe that if a process is legitimate, the outcome will eventually become real. But reality is not obligated to pass through European process.

Reality moves through builders.

This is why Europe2031 feels uncomfortable. It does not describe a sudden collapse. It describes a slow loss of agency by reasonable people making reasonable decisions inside institutions that cannot move at the speed of the thing they are trying to govern.

That is the real warning.

Not that Europe will be conquered.

That Europe will negotiate itself into irrelevance while still sounding civilized.

No leverage, no sovereignty

Europe should regulate AI.

But first, or at least at the same time, Europe must build enough disruptive capacity to make regulation enforceable. It must build compute, companies, energy, talent, capital, deployment speed and strategic assets. It must become useful to the future, not only concerned about the future.

Because the future will not respect Europe for having values it cannot defend.

If Europe does not build bargaining chips, others will bargain with Europe. An American president. A Chinese supply chain. A frontier AI lab. A cloud provider. A chip company. A defense partner. A platform owner.

And Europe will adjust.

It will call the adjustment pragmatism. It will call it partnership. It will call it strategic alignment. It will call it a temporary exception in extraordinary circumstances.

But the name will not matter.

The structure will be dependency.

No leverage, no sovereignty.

That is the rule. For people. For companies. For continents.

And if Europe wants to say something meaningful about the future, it has to build something the future cannot ignore.