Europe’s dilemma over artificial intelligence was highlighted this week when Christine Lagarde, president of the European Central Bank, warned that the continent faces a stark choice: reject AI and forfeit growth potential, or adopt it and risk dependence on systems developed by the United States and China. The growing discussion about AI safety, involving experts who disagree on the severity of potential risks, could render this dilemma irrelevant if worst-case outcomes materialize.

The continent’s limited role in shaping the safety discourse stems from its inability to produce AI leaders comparable to Apple, Google, or OpenAI. While Europe excels in regulation — a stance that has drawn criticism from Washington — it lacks a dominant AI company of the scale of Anthropic or Nvidia, which has weakened its influence in high-level safety conversations.

European Commission President Ursula von der Leyen proposed convening major AI developers to discuss ways to support industry efforts to pace AI development. However, her initiative failed to gain traction without backing from Washington or Beijing, making coordinated pacing efforts unlikely regardless of European input.

Lagarde emphasized on Monday that Europe must build its own AI capabilities and expand data center infrastructure to avoid being cut off from critical technology. She argued that self-sufficiency would reduce the threat of external disruption and create space for Europe to develop its own approaches to AI safety.

Margrethe Vestager, the European Commissioner for competition and digital policy, co-authored a report titled A Transformative AI Strategy for Europe, which reinforced Lagarde’s concerns. The report noted that AI is accelerating changes already affecting European citizens, who are “already bearing the economic and social consequences of decisions taken elsewhere.”

To address these challenges, the report recommends expanding data center capacity — described as the central nervous systems of AI — and strengthening resilience to AI-related crises, such as models slipping out of control. Vestager pointed to the EU AI Act as evidence that Europe does have a voice in safety discussions, requiring high-risk AI developers to report safety incidents and manage risks like uncontrolled model behavior.

She added that Europe must continue to build its own AI ecosystem to retain talent, foster innovation, and reduce dependency. While acknowledging that existential threats require global solutions, Vestager stressed the need for a supranational body that transcends the U.S.-China rivalry, suggesting the UN General Assembly as a forum for broader dialogue.

Frederike Kaltheuner, an adviser to the AI Now Institute, noted Europe’s deep reliance on U.S. technology — from cloud infrastructure to semiconductor components — making full independence difficult. She argued that narratives emphasizing European lag are shaped by a narrow definition of AI advancement promoted by American tech firms, focusing disproportionately on access to so-called frontier models like those behind ChatGPT.

In practice, many companies are not adopting these frontier models due to high costs and concerns about data control. Kaltheuner warned that if Europe responds to these fears by adopting U.S. AI systems across public and private sectors, it could deepen sovereignty risks, with critical infrastructure governed by a small number of foreign-owned companies.

Despite these challenges, the EU AI Act is not seen as a global regulatory benchmark. Legal experts note that its compliance requirements have caused delays and that its geographic focus limits broader adoption. Some policymakers in other regions are actively seeking to differentiate their approaches from the EU’s model.

Critics caution that alarmist narratives about AI risks could push European institutions toward overreliance on U.S. technology out of fear of falling behind, even when adoption is not strategically advantageous. This dynamic, they argue, threatens long-term technological sovereignty and innovation capacity.