Europe's AI gap
Skills shortages, data privacy fears and legal uncertainty are keeping European businesses away from AI — and Brussels' own rulebook may be part of the problem.
At a glance
Skills shortages are the single biggest barrier to AI adoption in European businesses, cited by more than 10% of both mid-sized and large companies.
Data privacy concerns and legal uncertainty rank second — more acutely felt by large companies than by small and medium-sized enterprises.
Fewer than 2% of European businesses consider AI tools irrelevant to their operations — meaning this is an adoption problem, not a relevance problem.
This image is used for illustrative purposes only.
A reluctance that has nothing to do with skepticism
More than a third of European businesses believe AI could transform their operations. They’re not using it anyway. This paradox — measured and documented in Eurostat’s 2025 survey — sits at the heart of a debate that has consumed Brussels since the AI Act entered into force. The regulation, the world’s first comprehensive legal framework for artificial intelligence, was built on a promise: regulate early, build trust, accelerate AI adoption. The data now raises an uncomfortable question about whether that promise has been kept.
The primary barrier is structural: a lack of technical skills. More than 10% of mid-sized companies (those with 50 to 249 employees) and 10.32% of large companies (250 or more) cite this as their main obstacle. Among the most affected: Denmark at 15.44%, Germany at 14.63%, and Finland at 13.99% — three economies routinely cited as Europe’s digital leaders. That the most advanced countries are also the most candid about their skills deficit suggests self-awareness rather than denial.
Only 2.09% of mid-sized companies and 1.55% of large ones say AI tools would be of no use to them. This is not a market rejecting a technology. It is a market that wants it — but cannot, or will not, take it on.
The law as a deterrent
Concerns about data protection and legal uncertainty rank second among stated barriers: 7.95% of mid-sized companies and 9.31% of large companies cite the risk of data privacy violations; 7.51% and 8.12% respectively flag a lack of clarity around legal consequences.
These figures must be read in context. The EU’s AI Act — passed in 2024 and currently being phased in — is the world’s first comprehensive AI law. It coexists with the General Data Protection Regulation (GDPR), the EU’s landmark data privacy law in force since 2018 and one of the most far-reaching privacy frameworks in the world. The interaction between the two texts is not always straightforward, and many companies may find themselves navigating what amounts to two overlapping sets of rules. It is plausible that this layering of regulatory obligations is itself a factor in non-adoption — though the available data does not allow for a formal causal link to be established.
Cost, by contrast, barely registers: only 5.67% of mid-sized companies and 5.51% of large ones cite it as a barrier. For EU policymakers, that is not where the battle is being fought.
Analysis: what Brussels would rather not hear
European businesses are not primarily looking for subsidies — they are looking for clarity.
The EU has built its AI policy around a core premise: regulate to inspire trust, and trust will drive adoption. The Eurostat data suggests this logic may be operating in reverse. The AI Omnibus, currently under negotiation in Brussels, is a package designed to simplify the AI governance framework by streamlining overlapping rules between the AI Act, the GDPR, and related legislation. It responds to a documented need. But its scope and timeline remain uncertain.
It is also worth noting that ethical concerns rank last among stated barriers (3.45% for mid-sized firms, 3.36% for large ones). Europe has invested enormous legislative energy in framing AI through the lens of fundamental rights. That effort is not illegitimate. But the data suggests it is not where the operational friction lies for most businesses.
Skills as a sovereignty issue
Integrating AI into a German mid-sized manufacturer or a Scandinavian large enterprise means retraining entire teams, restructuring processes, and making complex build-or-buy decisions. This is diffuse, long-horizon investment that is difficult to budget and harder to mandate. It requires targeted public incentives — not just a regulatory framework, however well-intentioned.
The EU channels billions of euros through its structural and cohesion funds to support digital transformation. Whether those mechanisms are effectively accelerating real AI adoption — rather than funding certified training programs that fail to change operational practice — would be a legitimate question to raise as the bloc negotiates its 2028–2032 Multiannual Financial Framework (MFF), the EU’s long-term spending plan.
The international context
In the United States, AI adoption by businesses has accelerated in a regulatory environment that looks almost nothing like Europe’s: no federal equivalent to the AI Act, a fragmented sector-by-sector approach, and a tech labor market fed by large-scale skilled immigration. These structural differences could partly explain the adoption gap, even if no direct causal link between EU regulation and lagging adoption can be formally established on the basis of this data alone.
The bottom line
Europe chose to regulate AI before it had adopted it at scale. That was a deliberate political choice. The Eurostat data now forces a question that Brussels can no longer defer: did that choice generate the confidence it was meant to build — or did it, by adding legal uncertainty on top of an already documented skills deficit, contribute to the very lag it was designed to prevent?
Sources: Euronews · Eurostat