The talent is here. The capital is arriving. So why doesn’t it always feel like it?

The answer isn’t regulation, though the reflex is to blame Brussels. The answer is that European founders are building on infrastructure they don’t own, distributing through platforms they don’t control, and scaling with capital that comes with strings attached — strings that run directly back to seven American corporations. Apple. Microsoft. Alphabet. Amazon. Meta. Tesla. Nvidia. The Magnificent Seven don’t just dominate stock indices. They own the real estate every European startup builds on. And until Europe confronts that structural reality, record funding figures will keep flattering a dependency problem.

The Infrastructure Trap

Start with the basics. A European AI startup in 2026 will likely rely on Microsoft Azure or AWS. It reaches customers through Apple’s App Store or Google Play. It finds those customers via Meta’s advertising algorithms or Google’s search dominance. Before a single line of proprietary code creates value, the company is already a tenant in someone else’s building — paying rent to landlords who are also, increasingly, its competitors.

This is what “walled gardens” means in practice. It is more than an abstract competition policy concern. It is the operating reality for thousands of European founders who have no structurally independent alternative. The Meta-Google duopoly alone commands over 50% of global digital ad spend. Distribution, discovery, and data monetization flow through their platforms by default. European founders aren’t building for Europe. They are building within constraints set in Menlo Park and Cupertino.

The capital picture makes this dependency structural rather than incidental. At the early stage, European and American AI startups attract roughly equivalent funding — a genuine sign of Europe’s competitive talent base. But by the later growth stage, 73% of European AI companies’ lead investors are American. The ratio of early-stage funding between Europe and the US is 1:1. By the later stages, it becomes 1:6. What begins as a level playing field ends as a funnel — one that routes Europe’s most promising companies toward American capital, American strategic interests, and ultimately, American exits.

Former Meta President of Global Affairs Nick Clegg has described the dynamic bluntly: Europe risks becoming a vassal state, trading long-term digital sovereignty for short-term capital access. That framing is uncomfortable, but it is directionally correct. When the infrastructure you depend on, the platforms you distribute through, and the investors who back your growth stage are all American, sovereignty is a polite fiction.

The Talent Paradox

Europe has approximately 325,000 AI professionals — a deep, technically sophisticated workforce produced by some of the world’s finest research universities. That is the supply side of a genuine competitive advantage. The demand side is where the problem compounds.

The Magnificent Seven are not passive observers of Europe’s AI talent pool. They are its most aggressive recruiters. Google’s London office, Meta’s Paris AI lab, and Microsoft’s expanding European engineering centers are talent absorption mechanisms. Many of Europe’s most capable engineers who choose to stay on the continent end up working for American corporations anyway — choosing Big Tech salaries over the risk and reward of building something of their own.

This is the talent paradox at the center of Europe’s AI moment. The continent is producing exactly the people it needs to win but failing to benefit from the economic value they create. European founders are competing with organizations that can pay two to three times the market rate, offer liquidity through globally traded stock, and absorb a promising hire’s entire career trajectory. More funding helps, but it does not solve a structural compensation gap against companies whose market capitalizations dwarf the GDP of most European nations.

What Breaking the Shackles Actually Requires

Europe’s regulatory instinct is correct but has been incompletely executed. GDPR, the Digital Markets Act, and the Digital Services Act represent genuine regulatory courage. But the first iteration of GDPR became an obstacle that hyperscalers cleared with compliance teams, while European publishers struggled — it entrenched incumbents rather than challenging them. Structural outcomes require structural interventions.

The AT&T divestiture of 1984 is the relevant historical precedent. Breaking up the Bell System did not weaken American telecommunications — it unleashed the competitive infrastructure that produced mobile telephony, fiber optics, and the foundations of the modern internet. Concentrated power, when broken up purposefully, generates more total innovation than it suppresses. Europe has both the regulatory mandate and the strategic incentive to apply that logic to digital infrastructure in a way Washington currently will not.

Three interventions would materially shift the structural balance. First, data and metadata ownership must not automatically transfer to infrastructure owners when users accept end-user agreements. The consent box cannot be the mechanism by which European consumer data becomes permanently American corporate property. Second, interoperability mandates under the Digital Markets Act must have teeth — genuine technical requirements that allow European alternatives to access distribution channels on fair terms, not voluntary commitments that are renegotiated at will. Third, the structural underinvestment in European late-stage growth capital must be addressed directly, whether through sovereign wealth mechanisms, pension fund reallocation, or co-investment structures that keep strategic ownership in European hands.

None of this is protectionism. Protectionism shields incumbents from competition. What Europe needs is precisely the opposite — the conditions under which its startups can compete rather than operate as permanent tenants of American infrastructure.

The Window Is Open, But Not for Long

Europe missed the social media wave. It missed the mobile internet platform era. Generative AI infrastructure is already American-dominated — the foundation models, the compute clusters, the hyperscaler cloud capacity that underpins them. That race is largely run.

But the race that matters most economically is not foundation models. It is the multi-trillion-dollar market for vertical AI applications — the companies applying intelligence to healthcare, industrial automation, financial services, climate technology, and the dozens of other sectors where Europe has deep domain expertise and genuine competitive advantage. That market is not yet determined. European companies are well-positioned to lead it.

They will not lead it if the infrastructure stack they build on, the platforms they distribute through, and the capital that funds their growth all remain structurally controlled by seven American corporations. The Magnificent Seven will always outspend European startups on the assets they control. Europe cannot win that game. It can only change the rules of it.

Record funding and world-class talent have given Europe its best window in a generation to build independent technological leadership. Whether Europe’s most talented founders spend the next decade building their own future — or executing someone else’s vision of it — depends entirely on the structural decisions made right now.

The shackles are visible. The tools to break them exist. What remains is the will to use them.

Disclosure: Northzone has financial interests in European technology companies that may benefit from the regulatory and policy changes described in this article. The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.