The European cities beating major hubs at AI innovation aren't the ones you'd expect

Europe's AI powerhouses aren't in Paris, Warsaw, or Madrid. Smaller cities are producing more specialized and influential AI research than major metropolitan hubs, according to a new study that suggests the EU is pouring investment into all the wrong places.

When most people picture the motor of Europe's AI ecosystem, their minds drift to the universities of Paris, the tech corridors of Warsaw, or the labs of Madrid. Those cities pump out the biggest absolute volumes of AI research on the continent, and have done for years.

But a new study covering 781 European regions suggests that while they draw the eyeballs, they’re not the places where AI is pursued with the greatest intensity.

Using bibliometric data from Clarivate InCites between 2015 and 2024, researchers mapped which European regions devote a disproportionate share of their scientific output to AI and machine learning, and how often that work is cited in the research community.

And while the biggest metropolises dominate the absolute numbers, the highest levels of relative specialisation in AI are actually on the continent's edges – particularly across Eastern Europe and Spain.

Local pride

Two supposedly fringe regions are leading the way in this kind of AI development. Granada, in southern Spain, and Vilniaus apskritis, the Lithuanian region that includes the country’s capital (and CyberNews’s home), Vilnius, combine high specialisation with strong volumes of research citations.

There’s often a perception that Europe's biggest cities are also its most dynamic AI drivers, with Paris, Warsaw, and Madrid leading the league tables. But when researchers investigate how much of a region's output is dedicated to AI relative to everything else they do, smaller cities tend to appear in a more positive light.

That matters for policymakers in Brussels, who have spent the last two years trying to position Europe as a serious counterweight to American and Chinese AI dominance.

The EU's AI Act, watered down in late 2025 to keep American investment flowing into European data centres, was meant to give the bloc's firms an edge through trust and transparency. And there are indications that it is being delayed even more in an attempt to appease Donald Trump’s opposition to it.

However, this new research suggests that some of that investment may be flowing to the wrong places if officials assume the established hubs are also the best places to invest.

Impact and influence

What’s most interesting about the paper is that the authors found a “weak relationship” between regional specialisation and citation impact. In plain English, being focused on AI doesn’t mean you produce influential research.

Some regions in Europe are highly specialised in AI but find their work is rarely cited, just being pumped out like an AI slot. What the researchers call “diversified scientific systems”, which combine moderate specialisation with strong citation impact, are what Europe should be aiming for, the academics argue.

Fyn, a Danish region that includes the city of Odense, is highlighted as an extreme case of very high citation impact despite comparatively low specialisation. It’s spotlighted as evidence that regions without a strong AI focus can still produce influential work when the underlying scientific base is strong enough.

The research is good evidence for Europe's smaller cities to make their case. While Brussels worries about keeping up with Silicon Valley and Hangzhou, university departments in Granada, Vilnius, and Odense are getting on with the work – and in some cases, getting cited more widely than their bigger city counterparts.

It poses a challenge for European policymakers to channel investment, infrastructure, and talent towards the regions that are doing that heavy lifting, rather than continuing to pour money into the metropolitan hubs that already dominate the headlines.

Unlock more exclusive Cybernews content on YouTube.