Regarding the impact of digitization on the environment, a definitive consensus remains elusive in the scholarly discourse. This article evaluates the carbon emissions impact of China's National Supercomputing Centers (NSCs) in terms of computational power, aiming to determine whether NSCs exhibit significant carbon emissions effects and to identify the underlying factors driving them. Theoretically, we integrate both direct and indirect effect into a comprehensive research framework, comprising the dimensions of energy consumption and energy structure. Empirically, the synthetic control method (SCM) is employed, constructing a group of synthetic control cities that closely resemble the NSC cities. The research findings show that NSC cities exhibit statistically significant differences in carbon emissions compared to synthetically non-NSC cities, yet this phenomenon manifests with intercity heterogeneity. Upon scrutinizing the driving factors, it becomes apparent that NSCs primarily elevate urban energy consumption scale through increased industrial electricity usage, thus fostering the escalation of local carbon emissions. Interestingly, in certain cities, despite witnessing a surge in energy consumption scale attributable to NSC, carbon emissions effects were not observed, which could be predominantly ascribed to a higher proportion of clean energy within the energy structure. The above results unveil the existence of NSC carbon emissions effect as well as its driving factors (high energy consumption and dirty energy using), providing policy guidance for public sectors aiming to enhance NSC operational efficiency, optimize energy structure, and broaden the spillover effect.