The impact of AI adoption on technological innovation performance in manufacturing: mediating by ambidextrous innovation and moderating by strategic flexibility

Corporate innovation activities involve a dynamic balance between exploration and exploitation, yet the theoretical understanding of how AI drives innovation performance by influencing both exploration and exploitation remains underexplored. Based on ambidextrous innovation theory, this study constructs a conceptual model of how AI adoption affects technological innovation performance in enterprises, with exploration and exploitation as mediators and strategic flexibility as a moderator. Using panel data from Chinese A-share manufacturing companies (2011–2023) and a two-way fixed effects model, empirical tests were conducted. The study finds that AI adoption significantly enhances technological innovation performance in manufacturing enterprises. Exploration and exploitation partially mediate this relationship. Strategic flexibility positively moderates the relationship between AI adoption and technological innovation performance. Heterogeneity analysis further reveals that the innovation-promoting effect of AI is more prominent in state-owned enterprises, large enterprises, and non-high-tech industries. The findings provide valuable insights into the relationship between AI adoption and innovation and offer practical guidance for enterprises leveraging AI to drive innovation.