Data Brief: Energy Economics AI Emission Reduction Dataset

This dataset comprises a balanced panel of 285 prefecture-level cities in China (including autonomous prefectures, prefectures, and leagues) over the period 2000–2019, yielding 5,700 city-year observations. It integrates four core categories of indicators: ① Carbon emissions data (traditional scope 1 & 2 emissions, and novel net carbon emissions that deduct AI’s own embodied carbon costs); ② Energy data (total energy consumption, electricity use by sector, and calibrated data center energy footprints); ③ AI-related data (number of AI enterprises, AI invention patents, and robot import exposure); ④ Socioeconomic and control data (GDP, population, industrial structure, financial development, environmental regulation, etc.).
The dataset was constructed to empirically test two core theoretical propositions: the “net effect principle” for evaluating general-purpose technologies and the “absorptive capacity threshold theory” for digital green transformation. It supports the accompanying research article “Unlocking AI's Green Potential: Net Emission Reduction Pathways and Regional Absorptive Capacity Thresholds” and is reusable for studies on urban green transition, AI environmental impacts, spatial spillovers of green technology, and energy–digital policy coordination.