- Enhanced incentives target tech giants including ByteDance, Alibaba and Tencent to reduce dependence on US suppliers.
China has ramped up subsidies that reduce electricity bills by up to 50 per cent for its leading data centres, in a significant push to promote domestic semiconductor usage and reduce the country’s dependence on US suppliers like Nvidia, according to the Financial Times.
The enhanced incentives by local governments target tech giants including ByteDance, Alibaba, and Tencent—companies that have faced rising electricity costs following Beijing’s ban on the purchase of Nvidia’s advanced artificial intelligence chips.
The new subsidies are intended to offset the higher operating costs associated with Chinese-made processors from firms such as Huawei and Cambricon, which industry experts say consume 30–50 per cent more power than Nvidia’s most advanced H20 chips.
Provinces with dense data centre clusters—such as Gansu, Guizhou, and Inner Mongolia—have unveiled plans to slash power bills by up to half for large data centres, provided they exclusively use approved Chinese semiconductors. Data centres running foreign chips, including those from Nvidia, are explicitly excluded from these programs, reinforcing China’s drive for self-sufficiency in advanced technology.
Local officials confirmed that aggressive energy subsidies, in some cases large enough to cover a data centre’s entire operating cost for about a year, are now part of fierce interprovincial competition to attract the next wave of AI infrastructure projects.
Industrial power rates in these regions are already about 30 per cent below those in eastern coastal provinces, and the new incentives lower them further to around 0.4 yuan ($5.6 cents) per kilowatt-hour. By comparison, similar electricity in the United States averages roughly 9.1 cents per kWh, with prices varying widely by state due to a fragmented grid.
While Chinese chips still lag behind Nvidia’s in single-chip compute performance, industry leaders such as Huawei have adopted clustering strategies to bridge the gap. This approach increases total power consumption but enables tech companies to keep pace with surging demand for AI applications.
Despite higher energy needs, China’s more centralised and resource-rich grid offers both cheaper and greener electricity than the US, ensuring no imminent supply shortages for rapidly scaling data centre operations. Mega-project incentives underscore the country’s determination to accelerate semiconductor self-reliance and compete head-to-head with the US in the global race for AI supremacy.