China’sArtificial Intelligence Industry as a Carbon-Linked Production System: Embodied Emissions, Structural Paths and Demand-Side Drivers

Artificial intelligence (AI) is commonly assessed through the electricity used by models and data centres, leaving the carbon transferred through the wider production system insufficiently resolved. We disaggregate China’s AI industry from the broader information and communication technology sector and construct comparable 31-sector environmentally extended input–output accounts for 2015, 2017, 2020 and 2023. Embodied-emission accounting is integrated with linkage analysis, structural path analysis and structural decomposition analysis to quantify the footprint, locate critical supply-chain pathways and explain temporal change. AI-related embodied CO2 emissions increased from 140.03 Mt in 2015 to 324.37 Mt in 2023. Indirect emissions reached 187.01 Mt in 2023 and remained higher than direct emissions, while backward linkages consistently exceeded forward linkages. Approximately half of the footprint was concentrated within the first three production tiers, with other ICT, electricity and heat, transport, metals and non-metallic minerals forming the largest short paths. In the current-price SDA, domestic final demand was associated with a 138.82 Mt increase between 2020 and 2023, outweighing the 20.24 Mt reduction from production-structure change. The results recast AI decarbonisation as a systems-governance problem: operational efficiency must be coupled with demand management, low-carbon electricity, hardware circularity and targeted upstream procurement.