Europe risks turning artificial intelligence into a source of economic vulnerability unless it builds more of its own technology and computing infrastructure, Christine Lagarde has warned. The president of the European Central Bank has argued that Europe’s reliance on AI technology developed overseas leaves the continent vulnerable if access is restricted, potentially affecting every sector of the economy.  Her intervention places data centers, computing capacity and domestic AI development alongside the more familiar questions of regulation and innovation that have dominated Europe’s technology debate. The issue is no longer simply whether European companies can adopt AI quickly enough to compete with American and Chinese businesses. It is whether Europe can retain meaningful control over technologies that could soon sit beneath critical public services, financial systems and industrial processes. Lagarde’s warning consequently presents AI infrastructure as a question of economic resilience rather than merely technological ambition.

Europe faces an uncomfortable AI dependency

Lagarde said Europe needs AI models that are “good enough” to perform most tasks while running from domestic data centers. She did not argue that Europe must immediately produce the world’s most powerful model or reproduce every capability developed by American technology companies. Instead, her argument centers on maintaining sufficient domestic capability to prevent foreign control of AI from becoming an unavoidable economic dependency. If European institutions and businesses can access capable models through infrastructure within Europe, changes in foreign policy, commercial terms or technology access would carry less force. “The threat of being cut off loses its force,” Lagarde said, describing the strategic value of developing European AI technology. That distinction is important because sovereignty in AI does not necessarily require technological isolation; it requires credible alternatives when external dependencies become politically or economically uncomfortable.

The scale of Europe’s current position makes that objective difficult. Lagarde said the US produced 59 notable AI models last year, while China produced 35, compared with one each from France and the UK. The figures illustrate a concentration of AI development that has left Europe outside the two major centers of frontier model production. The imbalance becomes even more pronounced when computing capacity enters the picture, with the US hosting 75% of the world’s AI computing capacity and Europe accounting for just 5%, according to figures cited by Lagarde. Computing power is not an incidental requirement for AI development because advanced models depend on large-scale infrastructure during training and deployment. Europe’s smaller position in that infrastructure therefore risks reinforcing its weaker position in the development and commercialization of AI itself.

The choice is adoption or dependence

Lagarde described Europe’s position as an “awkward choice” as governments and companies confront the pressure to adopt AI while trying to protect sensitive information and preserve strategic independence. “Either it holds back on adopting, because it cannot protect its data, and forgoes the growth. Or it adopts AI quickly, becomes highly dependent, and risks losing the freedom to organise its economy according to its own values.” That dilemma exposes a tension at the heart of Europe’s AI strategy. Regulation can establish rules around privacy, security and responsible use, but regulation alone cannot create the computing infrastructure needed to enforce technological independence. A continent can write strict rules for AI while still relying on infrastructure, models and platforms controlled elsewhere. Europe therefore faces a more fundamental question about whether its regulatory ambitions have sufficient physical and technological capacity behind them.

The concern becomes more serious as AI moves from experimental applications into systems that underpin essential economic activity. Lagarde argued that AI could soon influence border inspections, tax administration, rail operations, healthcare and banking, making disruption to access far more consequential than the loss of an ordinary software service. “Within a few years it will be screening goods at the border, deciding which tax returns are audited, dispatching trains, watching patients on wards and clearing payments at banks. A withdrawal of access, or a change in its terms, would then reach every sector at once,” she said. Such systems would create connections between AI infrastructure and the functioning of the wider economy that are difficult to reverse once established. If one external provider or technology ecosystem sits beneath multiple critical applications, a change in access could spread across industries rather than remaining confined to one customer.

AI could give foreign suppliers unprecedented leverage

Lagarde warned that this dependency could create a form of leverage that Europe has not previously faced from a trading partner. “That is leverage of a kind no trade partner has ever held over Europe, and it could be used in any negotiation, on tariffs or on digital taxes, for example,” she said. The warning comes at a time when the political relationship between Europe and the US has become less predictable, despite the two remaining major economic and strategic allies. The Trump administration has created tensions through tariffs, demands involving Greenland and disagreements over the future of US military commitments in Europe. Those disputes do not automatically translate into restrictions on technology, but they demonstrate why European policymakers are paying closer attention to dependencies that once appeared commercially secure.

China presents a different set of strategic considerations. Europe cannot assume that replacing one foreign technology dependency with another would resolve the underlying problem. The more significant objective is to develop enough domestic capacity to make external partnerships a choice rather than an absolute requirement. That would allow Europe to continue importing technology where it offers clear advantages while retaining alternatives for sensitive or strategically important applications. Such an approach would also avoid turning AI policy into a simple contest between European and foreign technology. Instead, it would frame technological sovereignty around resilience, optionality and the ability to keep critical systems operating when international conditions change.

The productivity prize is too large to ignore

The argument for greater European AI capability is not limited to geopolitical resilience. Lagarde also sees AI as a potential source of productivity growth at a time when European economies face persistent pressure to improve economic performance. She said rapid AI adoption could raise productivity by up to 4% over a decade, a shift she described as transformative for public finances. The implication is significant because productivity gains can influence the ability of governments, companies and workers to generate greater economic output from existing resources. AI could therefore become part of Europe’s response to broader economic pressures rather than remaining a technology-sector story. But those gains depend on the ability to deploy AI widely, securely and efficiently across the economy.

That creates a difficult policy balance. Europe cannot afford to slow adoption so much that it misses the economic benefits of the technology, but it also cannot pursue adoption without considering the dependencies that accompany it. Domestic infrastructure can provide one answer by giving European organizations greater control over where sensitive workloads run and which systems they depend on. European AI models could provide another layer of choice, particularly where data protection, public-sector requirements or strategic sensitivity make reliance on external systems less attractive. The objective would not be technological self-sufficiency in every category. It would be sufficient capability to prevent AI dependence from becoming an economic constraint.

Europe is financing the technology it may depend on

Lagarde’s concerns extend into financial markets, where the scale of investment required by major US technology companies is already influencing capital flows. US technology companies need enormous amounts of financing to support their AI ambitions, and some have increasingly turned to European debt markets to raise capital. That demand can place pressure on financing conditions as large technology borrowers compete with other companies for available capital. The consequences therefore do not remain confined to Silicon Valley or the US technology sector. European businesses can feel the effects through financial markets even as European consumers and institutions become increasingly dependent on the technologies those companies develop.

European savings also connect the continent financially to the same technology ecosystem. Lagarde noted that European pension funds invest heavily in US technology stocks, meaning a significant market correction could affect European savings. The situation creates a curious combination of dependence and exposure. Europe is financing the expansion of US technology companies while European businesses and investors remain exposed to the technology and financial markets driving the AI buildout. Financial exposure does not by itself give Europe control over the technology and infrastructure on which its businesses increasingly depend.  

Sovereignty means keeping a choice

Europe’s emerging AI challenge is therefore larger than the question of whether it can produce a world-leading chatbot. The more consequential issue is whether European institutions can maintain alternatives as AI expands into critical economic sectors, a development Lagarde expects within the next few years.  Domestic models, data centers and computing capacity would give governments and businesses more options when foreign providers change prices, policies or access conditions. Those alternatives would also strengthen Europe’s negotiating position with technology suppliers and trading partners. A sovereign AI strategy would not require Europe to abandon international cooperation or foreign technology. It would require Europe to ensure that cooperation remains a choice rather than a condition of economic continuity.

The warning from Lagarde ultimately comes down to timing. Europe is being urged to build domestic AI capability, expand computing infrastructure and create alternatives before AI becomes more deeply integrated into critical economic services. Yet waiting for foreign technology to become indispensable before addressing that dependency would make the eventual correction considerably harder. The continent does not need to win every AI race to protect its economic interests, but it does need enough capability to remain an active participant rather than a permanent customer. The strategic value of European AI sovereignty will therefore be measured not only by the models Europe creates but by the infrastructure it controls and the choices that infrastructure preserves. For Europe, building AI capacity is increasingly less about catching up with the US or China and more about ensuring that the future of its economy cannot be switched off elsewhere.