Adoption doubles, but small firms remain behind
According to CNEL’s 2026 Annual Report on Productivity, presented in Rome on November 4, in 2025 16.4% of Italian firms with at least ten employees used at least one artificial intelligence technology, compared with 8.2% in 2024 and 5.0% in 2023. The European average stands at 20.0%. The figure, cited in the report on the basis of Istat data, signals rapid growth but also a lag that mainly concerns smaller firms: large companies reach 53.1%, in line with the European figure of about 55%, while SMEs stop at 15.7%. In two years the distance between the two groups rose from about 20 to 37 points.
Uncertain effects and a gap at risk of widening
Italian estimates of the effects are not consistent. According to Ropele and Tagliabracci (Bank of Italy, Questioni di economia e finanza No. 1005, 2026), among firms with at least fifty employees, adoption is associated with value added per employee 5.2% higher. Bellomarini et al. (No. 1009, 2026), on about 900 firms with at least twenty employees, instead find no systematic effects in the short term; their simulations estimate for the next ten years annual growth in labour productivity 0.2, 0.7 or 1.1 percentage points higher depending on the speed of adoption, but these are scenarios and not observed effects.
The report warns that the return on AI increases with firm size and worker skills: without corrections, the technology tends to widen differences between firms and territories, overlapping with a digital divide that has not yet been closed. Between 2024 and 2025 SMEs doubled their share, but large firms, starting from higher levels, added more points: the gap grows even when smaller firms accelerate.
Italian productivity and obstacles to adoption
The report places the issue in a structural framework. Between 1995 and 2025, labour productivity in Italy grew by 0.2% a year, against 1.1% in the European Union; between 2019 and 2025, GDP increased by 1.0% a year, more than in Germany and France, but hours worked grew by 1.3% and productivity fell by 0.3%. The authors speak of extensive growth: more is produced because more work is done, not because each hour yields more.
Among firms that evaluated AI without adopting it, the main obstacles cited were lack of skills (58.6%), unclear rules (47.3%) and unavailability or poor quality of data (45.2%). One third of adopting firms, largely small ones, do not associate AI with a specific business function. Bank of Italy surveys report that among firms with at least twenty employees, 32% were users in 2026, but intensive use concerns barely 5% and 70% report no effects on productivity, although about half expect them within three years.
