The Carbon Reduction Effect of AI Policy: Quasi-Experimental Evidence from China's National AI Innovation Pilot Zones

The global low-carbon transition necessitates innovative policy interventions. Using staggered difference-in-differences estimation on a panel of 282 Chinese cities (2010–2023), this study provides causal evidence that China's National AI Innovation Pilot Zones (AIPZ) policy significantly reduces urban carbon emissions by 6.3% on average. Spatial econometric models reveal substantial negative spillovers, inducing an additional 8.6% reduction in contiguous cities, leading to a total abatement effect of 14.3%. Mechanism and heterogeneity analyses show that industrial upgrading and green innovation are key channels, with effects pronounced in the Pearl River Delta and non-resource-based cities, but short-run rebound effects occur in resource-dependent areas. This study demonstrates demonstrate that AI policies generate carbon co-benefits, yet their efficacy depends critically on local industrial context and spatial linkages, underscoring the importance of regional coordination in climate governance. Our findings underscore the importance of integrating AI policies into regional climate strategies to maximize carbon co-benefits.