How Computing Power Infrastructure Development Shapes Urban Green Innovation Performance: Evidence from China

Computing power infrastructure (CPI), represented by data centers, cloud computing platforms, edge computing nodes, and intelligent computing clusters, has become a strategic component of new infrastructure and a direct carrier of computational resources. This study adopts a two-stage empirical framework. In the first stage, urban green innovation performance (UGIP) is measured using a global benchmark undesirable-output slacks-based measure (GB-U-SBM) model that incorporates innovation inputs, desirable patent outputs, and undesirable carbon-emission outputs. In the second stage, the estimated UGIP scores are used as the dependent variable in two-way fixed-effects panel regressions to examine the association between CPI development and UGIP. Using a city-year panel for Chinese prefecture-level cities from 2006 to 2023 and a main regression sample of 5166 observations, the results show that CPI development is positively associated with UGIP after controlling for city fixed effects, year fixed effects, and city-level covariates. The results are robust to alternative measures of UGIP and CPI, sample treatments, winsorization, lagged specifications, and instrumental-variable estimates. Channel-consistency tests indicate that CPI development is associated with human capital accumulation and informatization, two conditions that support green innovation. The study clarifies that the evidence should be interpreted primarily as conditional association rather than definitive causal mediation. The findings suggest that CPI can become a productive foundation for green innovation only when it is embedded in skilled-labor formation, data governance, high utilization, and low-carbon operation systems.