When people talk about the artificial intelligence race, they tend to picture a two-horse contest: the US with its tech giants and venture capital firepower, and China with its state-backed industrial ambition. Europe, in this framing, barely registers. It is a large, wealthy and increasingly anxious spectator watching a race it did not enter.
But is Europe truly condemned to the sidelines or does it still have a distinct role to play in shaping how AI is developed and deployed? This was the subject of the European Commission’s Brussels Economic Forum earlier this month.
It is true that Europe does not have a frontier model that rivals ChatGPT or Gemini. It lacks the infrastructure that makes those models possible. And its AI unicorns, when they do emerge, tend to scale up overseas. The gap is real, and pretending otherwise would be a mistake.
But the framing of the race itself is worth questioning. If the contest is solely about who builds the largest model first, Europe is behind. If it is about who shapes how AI is used, who sets the standards, who ensures the technology serves people and who leads the industries that AI will transform, then Europe’s position is considerably more interesting.
Europe’s quiet strength: research, industry and trust
The continent’s research and academic base is world-class. European universities produce a disproportionate share of the foundational AI research that eventually powers the models everyone else is racing to build. The problem, as economists have noted for years, is translation. Research is done in Europe, but implementation happens elsewhere. The innovators of 20 years ago remain the innovators of today, while American tech has cycled through waves of disruption that have kept the frontier moving.
Alongside the research base, Europe has something more valuable in the long run: industries. Advanced manufacturing, pharmaceuticals, logistics, financial services and healthcare are all sectors where AI can generate enormous productivity gains and where Europe remains a serious player. The value of AI does not reside only in the models themselves; it lies in what those models enable. A Europe that deploys AI effectively across its industrial base could capture significant economic gains without necessarily winning the model-building contest.
There is also the question of trust. Democratic institutions, independent regulators, the rule of law and strong data protection regimes are all present in Europe. These are not obstacles to AI adoption, despite what more impatient voices in the industry argue. Over time, they are a competitive advantage. Businesses and governments around the world that want AI they can rely on, which does not expose them to unpredictable legal or reputational risk, will increasingly look for partners who can offer that.
Fragmentation, heavy rules and other challenges
The most visible challenge that Europe faces is regulation. The EU’s instinct to regulate first and ask questions later has real costs. For startups and scale-ups trying to move quickly, heavy compliance requirements are a brake on ambition. The AI Act, regardless of its merits, arrived in a climate of regulatory uncertainty that discouraged exactly the kind of bold investment the moment demands. Smart regulation and clunky regulation are not the same thing, and Europe has to start leaning towards the former.
Infrastructure is a deeper problem. Europe lacks the data centre capacity, chip manufacturing base and energy infrastructure to support AI at scale. Data centres that do exist face an energy market that is expensive by global standards and still too dependent on fossil fuels, creating both a cost disadvantage and a strategic vulnerability. The jump from research to deployment – from knowing how to build something to actually building it at scale – requires physical infrastructure that Europe has not yet prioritised.
The retention of human capital also creates a disadvantage. Talent is produced in European universities and promptly recruited by American firms. The missing link is not the talent; it is the ecosystem that would allow them to apply their knowledge within Europe: companies willing to hire aggressively, investors willing to back ambitious bets and labour markets flexible enough to absorb the disruption that innovation always brings.
Mobilising the capital required to scale AI investments is another major constraint. European financial markets remain fragmented, bank-dependent and structurally misaligned with the kind of long-horizon risk-taking that AI requires. The US venture capital ecosystem can absorb the losses that come with funding 10 companies to find a transformative one. European markets, constrained by fragmentation and cultural risk-aversion, struggle to do the same. The result is a financing gap that public money alone cannot close. A single, deep, integrated capital market is a precondition for Europe being able to fund its own AI future rather than selling promising companies to investors elsewhere.
Jumping from spectator to builder
The first thing that Europe can do is to lead the global conversation on how AI should be governed. This is not the consolation prize it is sometimes made to sound. Standards, norms and governance frameworks have enormous economic and geopolitical weight, as Europe demonstrated with data protection, where the General Data Protection Regulation shaped global corporate behaviour well beyond its formal jurisdiction. A Europe that leads credibly on AI ethics, transparency and accountability creates leverage that extends far beyond its own borders.
The second is the twin transition. Sustainable AI, with models that are efficient, powered by renewable energy and honest about their resource consumption, is increasingly a market requirement, as the energy costs and environmental footprint of large models become impossible to ignore. Europe’s combination of renewable energy ambition and regulatory sophistication puts it in a plausible position to set the global standard for green AI infrastructure. Renewable-powered data centres, environmental labelling on model resource use and incentives for task-specific rather than resource-intensive models are practical differentiators that Europe can adopt.
The third is capital markets reform. The debate about the capital markets union and the savings and investment union predates this technological moment by a decade. The arguments for deeper, more integrated European capital markets have always been compelling; they have simply never been compelling enough to overcome the political inertia that keeps national markets fragmented and European savings flowing into bonds rather than into growth. AI does not change those arguments, but it does raise the cost of ignoring them. The investment needed to compete in this technology cycle cannot come from public budgets alone. Private capital is not optional, and attracting it requires purpose-built markets.
Finally, regulation itself needs to become more agile. The argument here is not to abandon oversight; it is to design oversight that works with innovation rather than around it. China, at the forefront of the AI race, is in fact heavily regulated, yet its AI sector continues to advance. Regulation and innovation are not inherently enemies. The question is whether the regulation is intelligent, proportionate and fast enough to keep pace with the technology it governs. Sandbox regimes for start-ups, clearer, more predictable deployment rules and regulators with the technical capacity to understand what they are regulating are needed to advance in this front.
Europe is not out of the race. But it is running on hard ground, the field is highly contested and time is running out.
Andrea Correa is Head of Research at OMFIF.
Join OMFIF on 1 July to explore Global trade in transition: fragmentation, disruption and emerging opportunities.’
Interested in this topic? Subscribe to OMFIF’s newsletter for more.