Roland Busch, the chief executive of Siemens, told the Financial Times on Monday that Europe's approach to artificial intelligence regulation risks becoming a "disaster." His argument is not the standard corporate complaint about red tape. It is a specific objection to the order of operations: Brussels, he said, is prioritizing the construction of sovereign AI infrastructure — European-built models, European data centers, European chips — ahead of the more immediate task of deploying existing AI tools to boost economic growth. [1]

"If we first say we must build everything ourselves before we use it, that is a disaster for European competitiveness," Busch said. "The Americans are deploying. The Chinese are deploying. We are regulating." [1]

The remarks were made at Siemens' annual capital markets day, where the company simultaneously announced a one-billion-euro investment in industrial AI — its largest technology commitment to date. The juxtaposition was deliberate. Busch is not arguing against AI investment. He is arguing that Europe's political class has confused sovereignty with strategy, and that the pursuit of the former is delaying the latter. [1][2]

The EU AI Act, which entered force in 2024, is the world's most comprehensive regulatory framework for artificial intelligence. It classifies AI systems by risk level and imposes requirements ranging from transparency disclosures to outright bans on certain applications. European Commission President Ursula von der Leyen has added an AI sovereignty agenda — calling for European foundation models, European cloud infrastructure, and European semiconductor capacity — that would reduce dependence on American and Chinese technology. [2]

Busch's critique is that the sovereignty agenda has become a prerequisite for deployment rather than a parallel track. European companies, he argued, are waiting for regulatory clarity on which AI systems they can use, which data they can train on, and which infrastructure meets sovereignty requirements — while their American and Chinese competitors simply build and ship. [1]

The numbers support his frustration, at least in part. McKinsey estimated in January that European companies have adopted generative AI at roughly half the rate of their American counterparts. Venture capital investment in European AI startups was $8.4 billion in 2025, compared to $97 billion in the United States. The gap is not closing. [2]

Bloomberg reported Monday that Siemens' own industrial AI strategy — focused on factory automation, building management, and supply chain optimization — relies heavily on models developed by American companies, including OpenAI and Google. Busch acknowledged the dependency and called it rational: "We do not need to build a European language model to optimize a German factory. We need to optimize the German factory." [2]

The counterargument, articulated by European Commission officials, is that dependency on foreign AI infrastructure creates strategic vulnerability. If American AI providers are subject to export controls, political pressure, or corporate decisions that conflict with European interests, companies like Siemens would have no alternative. The sovereignty agenda, in this framing, is insurance. [1]

Busch's response: "Insurance you cannot use is not insurance. It is a cost."

-- DAVID CHEN, Beijing