Opinion: Europe Has Finally Stopped Talking About AI as a Regulation Problem — But the Industrial Strategy Test Begins Now
For the past three years, European political conversations about artificial intelligence have been dominated by a single frame: how to regulate it. The AI Act, adopted in 2024, is the most ambitious AI regulatory architecture in any major jurisdiction. The accompanying debate over the Omnibus regulation through 2025-2026 has reinforced the regulatory frame. Brussels has talked about AI as something that needs guardrails — and largely avoided the conversation that the United States and China have been having for the same period: how to build the AI industrial base that allows you to set the global pace.
The Brussels Economic Forum’s pivot
The 26th Brussels Economic Forum, opening this week, marks a clear inflection. The 2026 BEF theme — the EU’s strategic role and ambitions in the global AI race — is the most explicit reframing of European AI policy from a regulatory file to an industrial-strategy file that we have seen at this level of institutional weight. It matters that the European Commission itself, through its flagship economic event, is now leading with industrial strategy. It matters that Mario Draghi’s competitiveness diagnosis has been visibly internalised. It matters, most of all, that the question is finally being asked openly: can Europe still build the AI industrial base it needs?
The honest answer
The honest answer is: not without a sustained shift that the institutions have not yet committed to. The capital gap is real — US AI labs have raised cumulative investment of over $300 billion in three years, against European AI investment that runs at perhaps a tenth of that scale. The talent gap is real — graduate-level AI researchers in Europe systematically receive offers from US firms at multiples of European salaries, with knock-on effects for the domestic research base. The compute gap is real — the largest European AI clusters are smaller than the largest individual training runs at OpenAI, Anthropic and Google DeepMind.
What can change
None of these gaps is permanent. Capital can be marshalled — through the European Investment Bank, through the savings and investment union architecture moving through the Cypriot Council Presidency, through joint procurement instruments emerging out of AccelerateEU. Talent can be retained and attracted — through targeted research funding, through the European Research Council, through the talent visas and tax structures that several member states are starting to deploy. Compute can be scaled — through the EU’s flagship supercomputing programme EuroHPC, through the dedicated AI gigafactories announced in 2025-2026, through partnerships between national champions and European public infrastructure.
What the BEF must achieve
The BEF will only matter if it produces commitments that outlast the conjunctural moment. Three measurables will determine whether 2026 is the year the European AI industrial strategy becomes real, or another moment of rhetoric that fades. First, capital commitments visible in the room — not announcements of future commitments, but actual term sheets and binding allocations. Second, political signals on energy and infrastructure that survive the post-Iran energy political consensus and continue to support hyperscale clean energy build-out for AI compute. Third, connectivity between European AI start-ups and the deep-pocketed industrial users they need as customers — the Siemens, ASML, SAP, and Schneider Electric of the European industrial base.
The opposition’s case
It is worth taking seriously the case against an industrial-strategy turn in European AI policy. Critics argue, with some force, that European industrial policy has historically picked losers more often than winners, that the comparative advantage of an integrated single market lies in market processes rather than state-directed champion-building, that AI’s general-purpose nature means dedicated industrial policy may be capturing rents from sectors that would have absorbed AI productivity gains anyway. These arguments have weight. But they sit awkwardly against the empirical reality of the US and Chinese trajectories — neither of which is purely market-driven, and both of which are now several rotations ahead of the European pace.
The decade-defining call
What Europe decides on AI in 2026 — through the BEF, through the One Europe One Market roadmap, through the MFF 2028-2034 negotiations — will shape the continent’s productive capacity for at least the next decade. Get it right, and Europe builds an AI industrial base genuinely competitive with American and Chinese alternatives, embedded in the European industrial system that already exists. Get it wrong, and the continent slides further into being a regulator of AI rather than a producer of it. The Brussels Economic Forum is not the final answer. But the conversations it enables are the most consequential European industrial-policy conversations in three decades — and how seriously the institutions and the capital owners take those conversations will determine which path Europe ends up on.
