欧元区企业AI的采用与投资,基于统一企业层面数据的洞见

Adoption and investment in AIacross the euro area Annalisa Ferrando, Sara Lamboglia, Judit Rariga,Maurice Schmidt Insights from harmonised firm-level data Contents 1Introduction62AI adoption and investment in the SAFE132.1June 2025 survey round132.2December 2025 survey round143AI adoption: a tale of heterogeneity163.1Across countries163.2Across firm demographics (size, sector, age, ownership)173.3AI adoption and structural characteristics234Factors shaping firms’ use of AI254.1Drivers of AI adoption among AI using firms254.2Barriers to adoption and greater AI utilisation315Financing AI adoption and investment396AI adoption and firms’ business outlook456.1Impact on investment, productivity and employment45Box 1AI adoption and labour productivity: evidence from a novelsurvey-based measure476.2Impact on pricing behaviour506.3Impact on inflation expectations517Firms’ expected investment in AI537.1Which SMEs are planning to invest in AI?568Firms’ response to competition in AI investment588.1How much did competitors invest in AI by June 2025?588.2Firms’ assessment of shares of competitors investing in AI598.3The impact of information about domestic competitors on firms’investment plans in AI61 9Conclusion63 Abstract This paper explores the adoption of artificial intelligence (AI) technologies amongeuro area firms, using harmonised firm-level data from twodedicated modulesoftheSurvey on the Access to Finance of Enterprises (SAFE) conducted in June andDecember 2025. Based on responses fromaround 6,000 firms across 12 euro areacountries, the study examines AI adoption rates, drivers, barriers and economicimplications.The findings suggest that AIdiffusionamong euro area firmsisprogressing rapidly but unevenly, with significant variation across countries and firmcharacteristics.Approximately 70% of firms report some level of AI use, but only 7%classify their adoption as significant. Adoption is highest in the Netherlands, Finlandand Austria, and lowest in Italy and Ireland. Larger and younger firms, particularly intechnology-intensivesectors, are leading adopters. Firmsidentifyexpectedimprovements in business processes as the main driver of adoption,while keybarriersincludeskill shortages, data privacy concerns and system incompatibilities.CurrentAI useand investmentareprimarily financed through internal funds,complemented bygrants andsubsidisedbankloans. AI adoptionispositivelyassociatedwith firm productivity, turnover growth, fixed investment andown sellingpriceexpectations, particularly among intensive users.Survey data show noevidenceyetof aggregate labour shedding; instead, AI adoption is positivelyassociatedwithemployment growth.However, firms’ inflation expectations appearlargelyunaffected by current AI use. JELcodes:C93, D22,E31,L25, O33 Keywords:artificial intelligence, firm-level survey data, productivity, inflationexpectations Non-technical summary Artificial intelligence(AI)is increasingly viewed as a transformative technology withthe potential to reshape production processes, business models and economicstructures. Whether AI will raise productivity, lower costs, affect prices and ultimatelyalter themacroeconomy depends critically on how broadly and deeply firms adopt it.For monetary policymakers, understanding the pace and distribution of AI adoption isessential to assess implications for inflation, investment and potential output.1 This paper provides the first comprehensive cross-country picture of AI adoptionamong euro area firms, drawing on two dedicated modules of theEuropean CentralBank (ECB) and European CouncilSurvey on the Access to Finance of Enterprises(SAFE), a harmonised firm-level survey covering over6,000 firms across12euroarea countries. The June 2025 modulefocused onpast AI investment and its shareof total investment, whilethe December 2025 module examined the depth of AI useacross technologies, motives and barriers, financing patterns,businessexpectations(such asturnover and selling prices)and planned investment. Around 70% of euro area firms report some AI use, but only 7% describe their useas significant–a share that is similar across size classes, pointing to a commonceiling on intensive adoption. About 30% remain non-users, indicating that AI has notyet diffused uniformly. Adoption rates differ markedly across countries: theNetherlands, Finland and Austria report AI use above 80%, while Italy and Irelandfall below 65%. These gaps reflect both the proportion of non-adopters and the depthof use among adopters, implying that policy priorities differ between countriesfocused on triggering initial adoption and those focused on scaling up intensive use. Firm characteristics systematically shape adoption. Large firms are more likely touse AI, but the share of significant users is similar across sizes. Firms foundedaround or after the emergence of advanced large language models in late 2022show substantially higher rates of intensive AI use, suggesting that they embeddedAI int