Early Adoption of Generative AI among SMEs: Industry-level analysis using cloud accounting data

This study examines how small and medium-sized enterprises (SMEs) in Japan adopted generative artificial intelligence (AI) during the early phase of its diffusion. Generative AI spread rapidly after late 2022, yet little is known about how firms actually initiated adoption. Using monthly industry-level data constructed from cloud accounting service logs, we analyze actual payment records for generative AI services from 2022 to 2025, covering approximately 87,000 SMEs. We find that initial adoption was highly synchronized across industries, with a sharp increase in early 2023 following major technological releases. However, subsequent differences in adoption levels are primarily associated with industry characteristics and pre-existing digital technology usage structures. Sectors with more advanced digital infrastructures exhibit higher sustained adoption rates. By documenting real adoption behavior at its formative stage, this study provides baseline evidence for future evaluations of economic impacts and the design of SME technology policies.