Abstract
This study investigates the impact of network structural characteristics on sustainable innovation performance within regional collaborative networks in China’s artificial intelligence (AI) industry. Provincial-level innovation networks were constructed and analyzed using social network analysis to trace their evolutionary pathways using patent application data from 2010 to 2024. The findings reveal that China’s AI innovation network has developed into a multi-tiered, polycentric structure with Beijing as the primary hub. An inverted U-shaped relationship was identified between network centrality, structural holes, and regional collaborative innovation performance at various developmental stages. The external institutional environment, particularly through government R&D subsidies and intellectual property protection, plays a significant moderating role, generally diminishing the effect of centrality while enhancing that of structural holes during the rapid expansion phase. Regional heterogeneity analyses confirmed these patterns in eastern, central, and western China, whereas in the northeast, only centrality showed a significant association with performance. By integrating network location theory with an institutional perspective, this study offers a dual-perspective framework for understanding how sustainable innovation ecosystems can be fostered through network governance and policy interventions. The results provide evidence-based policy implications aimed at enhancing collaborative innovation capacity, mitigating regional disparities, and advancing sustainable development.