Zhejiang's low-carbon power-computing coordination cuts AI data center emissions by 62% across China

Zhejiang province in China has launched a pilot program that coordinates computing power with low-carbon electricity scheduling, aiming to strengthen grid resilience and cut emissions from AI data centers. The initiative uses real-time electricity carbon emission factors certified by SGS, CCS, DNV, and TUV NORD to guide where and when computing workloads run. By shifting AI training tasks from high-carbon eastern data centers to cleaner western regions with more renewable energy, the system reduced carbon emissions per AI token by 62% and increased resource utilization by 30%. Temporal scheduling also moves workloads to midday when solar generation peaks and carbon factors are lowest, cutting emissions by roughly 36%. The program integrates over 98 billion energy data records covering electricity, coal, oil, and natural gas. State Grid Zhejiang Electric Power plans to expand these electricity-carbon data services to industries, parks, and products, providing a replicable model for balancing AI growth with decarbonization across the Asia-Pacific region.