Greening China's digital economy: exploring the contribution of the East-West Computing Resources Transmission Project to CO2 reduction

The study addresses how China can expand its rapidly growing digital economy while reducing carbon emissions from data centers, a critical component of digital infrastructure. Globally, data centers and data transmission networks accounted for roughly 300 Mt CO2 in 2020 (~1% of energy-related GHG emissions), and halving these emissions by 2030 is a major challenge. China’s digital economy and associated computing capacity (racks) have expanded swiftly, leading to rising energy use and CO2 emissions, with fossil fuels supplying most electricity to data centers. Existing policies aim to curb this growth in emissions. However, prior research provides limited direct evidence on long-term, national-scale pathways to decarbonize China’s data centers. This paper focuses on the EWCRT Project and estimates its impact on emissions reduction between 2020 and 2050 through multiple scenarios, and explores regional configurations that can lower the carbon intensity of the digital economy using fsQCA.

Digital infrastructure, encompassing data centers and networks, has transformed industries and increasingly influences carbon emissions. Prior studies have examined emissions at regional, sectoral, city, and household levels, but often overlook the central role of data centers and lack robust, validated methods for modeling data center energy use and carbon emissions at national scales, especially in emerging economies. The literature reveals wide estimate ranges, limited data provenance, and a scarcity of methods for long-term national projections. In China, most research has inferred environmental impacts from digital policies (e.g., e-commerce pilots, Broadband China) and has not directly estimated data center carbon reductions; moreover, impacts differ regionally due to heterogeneous resource endowments, energy mixes, and economic structures. Project background: Launched in February 2022, the EWCRT Project addresses imbalances in data demand (eastern regions) and green energy supply (western regions) by redistributing computing workloads westward while maintaining critical services in the east. It plans eight national hubs and 10 mega-clusters and includes measures to improve PUE and increase clean energy use. Despite its significance, no prior studies had quantified the EWCRT’s carbon implications, motivating this analysis.

The study integrates a scenarios approach with fsQCA. Scenarios approach: Using LEAP (version 2020.1.0.64) and a bottom-up method, the authors estimate electricity consumption and CO2 emissions for data centers across eight national hubs. Electricity consumption is modeled as the product of number of racks, designed rack power, IT load utilization, PUE, and operating hours; emissions are then obtained by applying the share of green electricity and relevant emissions factors (green grid and national grid). Three main scenarios are defined: (1) BAU: rapid rack growth consistent with policy (20% annually 2021–2023, decelerating thereafter), rack distribution 8:2 east:west, PUE 1.43–1.58, initial green electricity ~20% rising to 30% in east by 2030 and earlier in west, and provincial-average emissions factors. (2) PRO (EWCRT): same rack growth as BAU but redistribution to 4:6 east:west, improved PUE (<1.25 east, <1.2 west), standard rack power 6 kW, IT load >65%, and staged increases in green electricity (west to 50% by 2035, east to 50% by 2040). (3) ADV: builds on PRO with further improvements: PUE to 1.1 after 2030 and green electricity to 100% in both regions by 2040. Additional sub-scenarios consider PUE-only improvements (P1 under BAU; P2 under PRO), green-electricity-only increases (G1 under BAU; G2 under PRO), and mixed improvements (M1 under BAU; ADV under PRO). Data sources include national policy documents, Greenpeace reports, the 14th Five-Year Plan, and other official statistics (Data S1–S8). fsQCA: Ten data center clusters are treated as cases. The outcome is regional carbon emissions intensity of the digital economy (presence = higher intensity; absence = lower intensity). Five conditions are examined: Green Investment, Racks, Green Energy, Green Attention (governmental focus), and Digital Level (city digitalization index). Continuous indicators are calibrated into fuzzy sets using thresholds at the 75th, 50th, and 25th percentiles. Necessary condition analysis finds none above 0.9. Configuration analysis yields three configurations associated with low-carbon intensity and three with high-carbon intensity, with example clusters identified for each configuration.

• Baseline distribution and early project emissions: In 2020, electricity consumption in regions hosting the eight hubs is ~102.7 billion kWh, generating ~51.58 Mt CO2 (about half of China’s data center emissions that year). In 2022, eastern hubs account for ~73% (52.77 Mt) and western hubs ~27% (19.58 Mt) of emissions. - Scenario results (2020–2050): • PRO vs BAU: Implementing EWCRT (PRO) reduces cumulative emissions by 2125 Mt CO2, driven by lower PUE and higher green electricity shares alongside a 4:6 east:west rack redistribution. • ADV vs no EWCRT: Combining EWCRT with advanced improvements (PUE to 1.1; 100% green power by 2040) cuts emissions by nearly 64% (about 9500 Mt CO2) relative to not implementing EWCRT. • PUE-only improvements: P1/P2 reduce emissions by at least ~6% (≈759 Mt CO2) versus no action. • Green-electricity-only improvements: G2 (EWCRT with 100% green by 2040) achieves 5043 Mt more reduction than G1 (BAU with east 50%/west 100% by 2040). • Mixed improvements: ADV outperforms M1, with additional reductions of 94 Mt (2035), 296 Mt (2040), 266 Mt (2045), and 215 Mt (2050); cumulatively, ADV achieves about 42% lower emissions over three decades compared to not implementing EWCRT. - fsQCA configurations: • Low-carbon intensity configurations (three): (1) GIR~GA~DL (green investment and many racks present; government green attention and city digital level not emphasized) – examples: Wuhu, Shaoguan. (2) ~GI~RGEGA (low racks and low green investment; high green energy and green attention) – examples: Qingyang, Chongqing. (3) GI~RGE~GADL (green investment, green energy, and high digital level; low racks and low green attention) – example: Tianfu. • High-carbon intensity configurations (three): (4) ~GI~R~GEGA – examples: Helingel, Zhongwei. (5) ~GIR~GE~GADL – example: Guian. (6) GIR~GEGA*~DL – example: Zhangjiakou. • No single necessary condition for either low- or high-carbon outcomes (consistency < 0.9 for all single conditions). Overall, green energy availability is a pivotal antecedent for low-carbon configurations; high rack counts can coexist with low carbon intensity given sufficient green investment and/or green energy.

The findings demonstrate that the EWCRT Project can enable China to expand computing capacity while significantly curbing data center emissions through a combination of spatial reallocation (east-to-west), efficiency gains (lower PUE), and decarbonized power supplies (greater green electricity shares). Although emissions may initially rise with rapid capacity growth, the PRO and especially ADV pathways deliver substantial long-term reductions. The fsQCA results complement the scenarios by revealing multiple viable pathways to low-carbon digital economy intensity, highlighting the central role of green energy. Importantly, a high number of racks does not inherently imply high carbon intensity when paired with strong green investment and clean power. Governmental attention and digital maturity can further condition outcomes, particularly in regions with abundant renewables. Collectively, the results underscore that differentiated, region-specific strategies—targeting energy mix, technological efficiency, and policy attention—are essential to greening China’s digital economy.

This study provides direct, long-term estimates of the EWCRT Project’s decarbonization potential using a bottom-up LEAP-based scenarios approach integrated with fsQCA. The PRO scenario reduces 2125 Mt CO2 vs BAU (2020–2050), and adding advanced improvements (PUE 1.1; 100% green electricity by 2040) yields about 9500 Mt cumulative reduction, showing that large-scale digital infrastructure can be aligned with national carbon goals. The fsQCA identifies multiple low-carbon configurations, emphasizing green energy and investment as key levers and illustrating that large rack deployments can be compatible with low emissions intensity. The framework offers a reference model for other emerging economies planning large-scale digital infrastructure. Future research could apply this combined scenarios–fsQCA framework to other national contexts, incorporate richer operational data as it becomes available, and assess policy instruments and market mechanisms to accelerate PUE improvements and green power adoption.