Triple Helix synergy at the subsidy threshold: Government support, IUR collaboration, and knowledge spillovers in enterprise AI innovation

This study examines how university–industry collaboration, government subsidies, and knowledge spillovers jointly influence enterprise AI innovation using panel data from Chinese listed companies (2010−2023). While each Triple Helix component independently enhances innovation, their interactions reveal strong complementarities. We identify a critical subsidy intensity threshold at 1.39% of firm revenue: below this level, collaborative synergies remain dormant; within the optimal range (1.39%–3.93%), interaction effects are nearly four times stronger than in the low-subsidy regime. However, excessive subsidies yield diminishing returns, revealing an inverted U-shaped relationship. This nonlinear pattern helps explain inconsistent findings in prior research. Heterogeneity analysis shows these effects are stronger for non-state-owned enterprises and digital sector firms. Government subsidies benefit low R&D intensity firms most, while knowledge spillovers favor high R&D intensity firms. Using Heckman selection models and propensity score matching to address endogeneity, our findings introduce conditional complementarity to Triple Helix theory and demonstrate that appropriately calibrated subsidies combined with university partnerships substantially outperform isolated policy interventions.