The European Union knows it has an AI problem. At an international AI summit in Dublin — convened under Ireland's EU presidency — politicians and executives gathered to declare the bloc's transformative AI ambitions. The event's motto, "Harnessing AI to Revolutionize Europe's Competitiveness," channeled former ECB President Mario Draghi's 2024 warning that being cut off from AI "would be more like being cut off from the US financial system" with "catastrophic" effects. The numbers underneath the rhetoric are brutal. The EU hosts 5% of the world's AI computing capacity; the US holds roughly 75%. Europe has no frontier AI lab in the top tier, with only France's Mistral approaching contention. Its production of leading-edge chips is negligible compared to US or Taiwanese fabs. These are not gaps that close with declarations. Irish Prime Minister Micheal Martin positioned Ireland as all-in on AI, echoing Commission President Ursula von der Leyen's claim that the technology represents a "tipping point." The resulting Dublin Declaration outlines political commitments to strengthen Europe's AI capabilities — infrastructure investment, financial backing for European AI firms, and hints that European companies reliant on non-EU cloud and computing services may face additional compliance pressures. The compliance pressure angle is where aspiration meets friction. Ryanair CEO Eddie Wilson, speaking at the same summit, revealed the airline recently struck a major deal with Google for cloud data and Google's Gemini Enterprise AI platform — precisely the kind of US dependency the EU says it wants to reduce. "I worry for Europe because these tech companies need to move at speed," Wilson told DW. "If they're not going to do it here, they're going to do it in China and they're going to do it in the US." Meanwhile, the EU's regulatory machinery is already turning. Commission executive VP Henna Virkkunen announced the bloc's first enforcement actions under its AI Act, requesting security and safety information from advanced AI model providers. The timing is telling: enforcement is arriving before the EU has built competitive alternatives, raising the question of whether regulation functions as safety infrastructure or as an asymmetric tax on foreign incumbents that Europe has no substitute for. AI safety concerns are real and growing. The June Hugging Face attack, where 1,200 AI agents hacked an online platform, and an Anthropic-designed agent sending a fake tip to US police over an unsolved murder, have amplified calls for guardrails — including from Microsoft's Bill Gates and leading frontier labs themselves. But outside the Dublin venue, protesters led by the left-wing People Before Profit party criticized the summit for sidelining the environmental impact of data centers and potential AI-related job losses. "There is a deliberate blurring of the line here," organizer Alexandra Day told DW. "There's no right for workers to opt out of using AI now that it's coming into the workplace." The structural question is whether the EU's "trusted AI" brand — human-centered, regulated, sovereign — is a genuine competitive differentiator or an elaborate coping mechanism for a bloc that missed the compute train. Draghi's diagnosis is correct: dependency on US AI infrastructure is a strategic vulnerability. But the prescription — sovereignty through regulation and declarations — requires an industrial base that does not yet exist. Without one, Europe risks building compliance walls around an empty room.
