Explaining Uneven AI Policy Absorption Among Italian Small and Medium-Sized Enterprises

Abstract

The European Union (EU) increasingly treats artificial intelligence (AI) adoption as a condition for productivity, competitiveness, and technological resilience, yet policy provision does not automatically become firm-level capability. This study examines why AI adoption remains uneven among Italian small and medium-sized enterprises (SMEs) despite expanding regulatory, infrastructural, financial, and intermediary support. A qualitative, interpretive, document-based comparative design is used. Italy is the focal case and Denmark, Finland, and Romania provide structured contrasts. The analysis integrates EU and national policy documents, official 2025 enterprise statistics, 2026 Digital Decade country reports, and peer-reviewed research through a Policy Absorption Model that distinguishes policy provision, policy exposure, policy absorption, and adoption outcomes across five capacities: technological, organisational, economic, regulatory, and ecosystem. The evidence shows a pronounced Italian provision-absorption gap. In 2025, 16.4% of Italian enterprises with at least ten employees used AI, including 15.7% of SMEs versus 53.1% of large enterprises. Skills shortages, legal uncertainty, inadequate data, privacy concerns, and cost were the principal reported barriers. Comparative evidence places Denmark and Finland in stronger but not uniform absorption configurations, whereas Romania displays more pervasive weakest-link constraints. The study contributes a multilevel management explanation of why similar policy opportunities generate different organisational outcomes and shows that the weakest critical capacity can constrain the value of strengths elsewhere. For managers and policymakers, the implication is to sequence AI support around readiness and capability conversion rather than programme participation alone.