Artificial Intelligence: what drives adoption in EU countries?

This paper uses results from the 2025 Eurostat's firm-level survey to study the determinants of AI adoption in a cross section of 17 countries and 24 industries of the European Union. We relate adoption to several potential drivers: the perceived obstacles to adoption (as resulting from the survey), the business environment (such as market conditions), the policy context (such as policies directly or indirectly affecting the capabilities and incentives to adopt and use AI technologies) and demography (shares of different age groups within employees). We summarize some of the potential drivers using Principal Components Analysis and take a diffin-diff approach to regression analysis to highlight how the influence of nation-wide drivers of adoption differs depending on industry exposure to them. We find robust statistical associations of first time AI adoption (use of a single AI technology) or extensive AI use (at least three AI technologies) with concerns related to legal uncertainty and ethics or anticipated costs, respectively. Human capabilities (such as ICT skills) and innovation policies are positively related to AI adoption, while policies curbing workforce flexibility or competition in industries that are key providers of intermediate inputs are inversely related to adoption. Finally, we find some evidence of an inverted U-curve between age and AI, with adoption being weaker where the age distribution of the workforce is tilted towards very young or elderly workers. However, AI exposure lessens the negative association between young age and adoption perhaps due to skill-leveling effect of the new technology.