Euro area firms embrace AI but deep adoption still limited

Firms boost AI use but skills shortages hold back progressEuro area firms rapidly expanded their use of artificial intelligence in 2025, with around 70 per cent reporting some level of adoption, according to a new European Central Bank (ECB) study based on surveys conducted in June and December 2025. An ECB occasional paper by Annalisa Ferrando, Sara Lamboglia, Judit Rariga and Maurice Schmidt drew on responses from about 6,000 firms across 12 euro area countries, offering one of the most detailed cross-country snapshots yet of how businesses are adopting the technology. The findings showed that while AI uptake is widespread, only 7 per cent of firms describe their use as significant, pointing to limited deep integration despite broad experimentation. The study highlighted sharp differences between countries, with adoption rates exceeding 80 per cent in the Netherlands, Finland and Austria, but falling below 65 per cent in Italy and Ireland. These gaps reflect both how many firms have yet to adopt AI and how intensively it is used among adopters, suggesting differing policy priorities between countries seeking to trigger uptake and those aiming to scale advanced use. The research identified firm size, age and sector as key determinants of adoption, with larger and younger firms leading the way, particularly those established around or after late 2022 when advanced large language models emerged. Firms in technology-heavy industries show the highest levels of significant adoption, while sectors such as construction, accommodation and transport lag behind. Ownership also plays a role, with listed and venture capital-backed firms among the most advanced users, while family-owned and single-owner businesses are more often non-users or only limited adopters. The paper found that improving business processes is the main driver of AI adoption across countries and sectors, while reducing personnel costs and supporting innovation are secondary motivations. Among firms that have not adopted AI or use it only sparingly, the most frequently cited barrier is a lack of relevant skills, followed by concerns over data privacy, ethics and compatibility with existing systems. Smaller firms are also more likely to see AI as irrelevant to their operations, the researchers explained. In terms of financing, the study shows that internal funds remain the dominant source of investment in AI, supplemented by grants and subsidised bank loans, with conventional bank lending playing a smaller role. For small and medium-sized enterprises, all three sources are significantly linked to both current use and planned investment, while for larger firms internal funds and subsidised loans dominate and bank lending plays no meaningful role in future plans. The research found that firms making significant use of AI expect stronger turnover growth and higher fixed investment, indicating a self-reinforcing dynamic in which deeper adoption strengthens competitive positioning. Only these intensive users anticipate higher selling prices, by about 0.85 percentage points compared with non-users, suggesting either improved quality or greater ability to pass on costs. By contrast, wage and input cost expectations show little variation across firms, indicating limited broader economic effects at this stage. On productivity, the study reports no overall statistically significant difference between users and non-users, although firms in technology sectors that adopt AI are more likely to report higher current and expected productivity. Even low-intensity use appears to boost productivity in these sectors, but future gains are expected mainly among firms with significant adoption. The findings also showed no evidence of job losses linked to AI adoption, with firms instead reporting expectations of employment growth, suggesting the technology is currently complementing rather than replacing workers. At the same time, the study finds no clear link between AI use and inflation expectations, except among moderate users who report slightly lower expected inflation across all time horizons. A randomised trial embedded in the December 2025 survey revealed that firms often underestimate how widely AI is used by competitors. Around 80 per cent of firms underestimate adoption rates within their sector, country and size group, with the average gap reaching about 18 percentage points. Larger firms tend to underestimate adoption more than smaller ones, possibly because they view themselves as market leaders, while firms in digital sectors show more accurate perceptions. When given information about competitors’ past AI investments, firms increased their expectations of future adoption by roughly 4 percentage points and raised their own planned investment by nearly 2 percentage points. The authors stated that “disseminating knowledge about aggregate adoption patterns … could help accelerate the spread of new technologies like AI”, highlighting the role of information gaps in slowing diffusion. The study argued that while AI is already widely used across the euro area, deep integration remains limited and uneven, with adoption shaped by firm characteristics, financing constraints and national differences. It warned that the broader economic impact of AI, particularly on productivity and inflation, remains uncertain and will depend on how quickly and extensively firms expand its use. The paper also pointed to financing challenges linked to AI investment, noting that such spending often combines tangible assets like hardware with intangible elements such as software, cloud services and organisational capabilities, which are harder to use as collateral. This may push firms towards internal funding and raise the cost of external finance, the paper explained. The authors suggested that future research should examine in more detail how firms finance AI investments, including the potential role of leasing, corporate bonds and equity financing alongside traditional bank lending. They added that a better understanding of financing patterns could inform policy efforts to deepen European capital markets and reduce barriers to digital transformation.