Computing Power Drives Surge in China's Electricity Consumption

China's overall electricity consumption exceeded 1 trillion kilowatt-hours in July, marking a substantial increase driven in part by a 40% year-on-year rise in power usage for internet data services. Computing power is becoming a major contributor to this growing demand.

In Changshu, Jiangsu province, an intelligent computing center experienced an August peak load increase of over 70% compared to the previous year. The computing industry's peak load reached 200,000 kilowatts, accounting for approximately 4.1% of the city's total electricity load. According to the local power company, the computing sector contributes nearly half of the city's incremental electricity demand, supplanting seasonal air conditioning loads as the primary driver of summer electricity consumption growth.

Across Jiangsu province, 106 computing center companies consumed over 1.3 billion kilowatt-hours of electricity in July, a 31% increase. Supercomputing and intelligent computing centers represented over 80% of this consumption.

To manage the grid impact, China recently completed its first 100-megawatt-class coordinated regulation between computing power and the electricity grid in the Jibei power grid. During peak hours, 11 computing centers voluntarily participated, freeing up 100.64 megawatts of power supply space in less than 10 minutes, enough to power over 20,000 households.

Projections estimate China's data center electricity consumption could surge to 800 billion kilowatt-hours by 2030. Simultaneously, the country's installed capacity for solar and wind power is expected to exceed 2.8 billion kilowatts. To address the dual challenges of massive computing demand and green power absorption, several regions are developing "computing-electricity coordination" models, aiming to align computing load with green power availability for cost reduction and consumption promotion. For instance, the world's largest intelligent computing center in Ulanqab, Inner Mongolia, is exploring direct green power supply models using its own wind farms and AI-driven power systems.