Shanghai has become the first city in China to synergise computing centres with the power grid for peak-load management, a national milestone in the country’s efforts to enhance computing-grid coordination amid the global artificial intelligence race.

Led by State Grid’s Shanghai subsidiary, the trial on July 16, 2026, achieved a peak load reduction of 97.8 megawatts within a two-hour window. This represents the largest single city-level adjustment of computing power load in China, according to a report by Jiefang Daily, a Shanghai-based official newspaper.

The scale of this reduction is equivalent to shutting down nearly 140,000 of Nvidia’s flagship H100 graphics processing units (GPUs) simultaneously, with each unit requiring up to 700 watts of power for maximum performance, as calculated by the South China Morning Post.

A total of 16 data centre operators participated in the trial, including major players such as Alibaba Group Holding, China Telecom’s Shanghai branch, and GDS Holdings. The initiative aims to transform AI data centres into flexible shock absorbers that can increase energy-intensive training during green energy surpluses and reduce operations to ease peak grid demand.

According to State Grid’s Shanghai branch, this trial marked the first time a Chinese city has successfully run through the entire technical pipeline of computing-grid coordination simultaneously. The process involved modulating data centres’ power usage, switching to backup diesel generators when necessary, and shifting computing tasks to other regions to alleviate grid strain.

This development comes as China experiences a surge in power consumption driven by the global AI race, prompting policymakers to increasingly regard computing power as a national utility rather than a resource owned solely by private firms.

Sources