Shanghai Cuts 97.8MW in 2-Hour Computing-Grid Trial as AI Power Demand Surges
Updated
Updated · South China Morning Post · Jul 16
Shanghai Cuts 97.8MW in 2-Hour Computing-Grid Trial as AI Power Demand Surges
1 articles · Updated · South China Morning Post · Jul 16
Summary
97.8 megawatts of peak load were cut in two hours in a Shanghai trial that Jiefang Daily said was China’s largest single city-level adjustment of computing power load.
16 data-center operators, including Alibaba, China Telecom’s Shanghai branch and GDS, coordinated with State Grid to modulate power use, switch to backup diesel generators and shift computing tasks to other regions.
State Grid’s Shanghai branch said the exercise was the first time a Chinese city had run the full technical pipeline of computing-grid coordination simultaneously.
China is pushing that model as the AI boom lifts electricity demand, treating computing power increasingly as a national utility that can absorb surplus green energy or curb strain at peak hours.
Can Shanghai's grid balancing succeed where China's national 'Eastern Data, Western Computing' strategy failed?
Are underwater data centers a sustainable solution for AI's power thirst or a new ecological threat?
China’s 97.8MW Shanghai Computing-Grid Trial: Balancing AI Ambitions with Green Energy and Grid Innovation
Overview
Shanghai's recent computing-grid integration trial highlights China's strategic push to address the soaring power demands driven by artificial intelligence. As AI technologies require intensive computation, China is experiencing a surge in energy consumption, prompting policymakers to reclassify computing power as a fundamental national utility. This shift reflects a broader government effort to manage and optimize computing resources at the national level, ensuring reliable and efficient support for the country's ambitious AI goals. The trial demonstrates how China is aligning its energy infrastructure with its pursuit of global AI leadership, balancing technological growth with sustainable resource management.