Digital infrastructure, innovation capacity, and AI technology adoption in EU manufacturing

This paper examines the determinants of artificial intelligence adoption among manufacturing and service enterprises across the EU-27, focusing on digital infrastructure and firm-size differences. Using a balanced panel covering 27 member states and two survey waves, 2023 and 2025, we combine data from Eurostat's ICT Usage in Enterprises survey, the Digital Intensity Index, and the Community Innovation Survey. Pooled OLS, random-effects, and fixed-effects models estimate the relationship between AI adoption and digital intensity, innovation capacity, firm size, and Central and Eastern European location. Digital infrastructure emerges as the strongest predictor of AI adoption, with the Digital Intensity Index showing a significant positive effect across model specifications. Large firms consistently report higher adoption rates than small and medium-sized enterprises. Although Central and Eastern European countries continue to lag behind Western European states, they record the fastest growth, with Romania showing a 223% increase. Tests of sigma and beta convergence indicate narrowing cross-country disparities and support a catching-up process in AI diffusion. The findings inform EU industrial policy, quality-management upgrading, and targeted AI-adoption programmes in lagging regions.