How Europe Plans to Avoid Dependence on Foreign AI Powers

Europe’s Transformative AI Strategy argues that the continent should treat AI infrastructure, strategic data and supply chains as foundations of economic resilience—not simply as a race to create a European equivalent of a leading foreign model provider. It is a proposal from its contributors, not an adopted EU programme. Its core objective is to give Europe durable access to advanced AI while ensuring that the value created by AI can support European productivity, institutions and businesses. 1214

The strategy’s central answer: build leverage as well as capacity

The proposal does not frame success solely as winning the frontier-model race with a single European champion. In the near term, it emphasizes securing dependable access to advanced systems through bargaining power: semiconductor supply, European data-centre locations, strategically valuable data, procurement and other positions in the AI value chain. A domestic frontier-AI effort remains an option, but the strategy cautions against a poorly funded project that would consume resources without ending dependence. 12

That distinction matters. Europe could expand access to AI without controlling every layer of the stack, but it would remain exposed if compute, data processing and chip supply were controlled elsewhere. The strategy therefore treats infrastructure and data governance as economic assets as well as sovereignty safeguards. 1214

Use European supply-chain chokepoints collectively

One of the recommendations is a Member State Alliance for Supply Chain Security, coordinated with the European Commission. The strategy argues that individual European countries hold important positions in semiconductor and AI supply chains but do not fully convert them into shared strategic leverage when they act separately. 14

The intended role of the alliance is to coordinate Europe’s approach to chip supply, investment and potential export restrictions, including around European semiconductor-industry strengths such as ASML-related lithography capabilities. The goal is not autarky; it is stronger resilience and negotiating power over long-term access to critical AI inputs and services. 14

Raise Europe’s compute share to 15% by 2030

The most concrete infrastructure objective is to host at least 15% of global AI compute on EU soil by 2030, compared with an estimated current share of around 5%. The strategy’s own projection puts that goal at roughly 45 gigawatts of total facility power in the EU, against a projected 300 GW global total. 1415

That is a major physical-economy project, not just a technology policy. Achieving it would require large-scale financing, quicker permitting, grid connections, dependable energy supply and data-centre development that earns local legitimacy. The proposal explicitly argues that the build-out should benefit local communities rather than impose costs while value accrues elsewhere. 14

The European Commission is already pursuing related infrastructure measures through its AI Continent agenda, including AI Factories and planned AI Gigafactories. The Commission says its programme is designed to support startups, industry and researchers with large-scale computing capacity. 15

Keep data and infrastructure contestable

The strategy’s idea of sovereignty is not simply to replace foreign dependence with a closed European market. It calls for arrangements that preserve portability, multi-vendor access and contestability in compute and data services. 12

This is where the proposal aligns with Mario Draghi’s warning that Europe should not trade Silicon Valley concentration for a European oligopoly. The shared logic is that pooled European demand can help finance needed infrastructure, but the resulting capacity should remain broadly accessible and competitive rather than locked under one protected provider. 1315

In practical terms, that makes procurement and governance central: Europe would need enough shared scale to fund infrastructure and negotiate with suppliers, while preventing that scale from becoming permanent market foreclosure.

Turn AI infrastructure into productivity growth

The strategy presents compute and data control as means to a wider economic end: diffusion of AI through Europe’s manufacturing and service economy. It estimates that broad AI adoption could add roughly 0.3 to 0.4 percentage points to annual total-factor-productivity growth. 12

That is consistent with the EU’s wider AI policy direction. The Commission’s Apply AI Strategy focuses on adoption in key industrial sectors and the public sector, alongside support for small and medium-sized businesses, workforce readiness and market trust. 6

The implication is that data centres alone are not the finish line. The economic return depends on whether European firms, public bodies and workers can actually use AI capabilities across real workflows.

Why execution is the real test

The strategy’s contributors portray the opportunity as time-sensitive and call for decisive implementation beginning in 2026. 12 But its numerical goals also reveal the scale of the challenge. Building European-controlled capacity at the proposed level would require aligned investment decisions, electricity generation and networks, faster approvals, workable state-aid rules, procurement coordination and credible safeguards for communities.

The proposal’s strongest insight is therefore broader than “build more data centres.” Europe’s route away from permanent AI dependence is to combine physical capacity with collective negotiating leverage, strategic data control and broad-based adoption—while keeping the resulting market open enough to avoid creating a new dependency at home. 121415