Using textual data from annual reports of A-share-listed companies, this study constructs measurement indicators for corporate AI application levels in China. It systematically examines the impact of AI applications on the total factor productivity of Chinese manufacturing enterprises through empirical analysis. The analysis of Chinese listed manufacturing firms’ panel data and fixed effects regression, moderation effect models, and threshold regression procedure shows applied institutional moderation mechanisms through both an intermediary path of policy moderated by government AI policy and a significant non-linear threshold influence on AI adoption’s impact on firm productivity. It also discusses an alternate way in which supply chain resilience (SCR) and financing constraints display threshold effects for what AI adoption means toward productive outcomes. Our findings confirm that AI applications can significantly enhance corporate production efficiency in the presence of regional and industrial heterogeneity. Additionally, the government’s AI policies, SCR, and financing constraints result in positive, moderating, and significant threshold effects, respectively. We propose strengthening incentive measures, implementing differentiated policies, and stabilizing expectations as countermeasures.

Keywords: Artificial intelligence, total factor productivity, government policy, supply chain resilience, financing constraints