Slashing Electricity Bills: How China Is Subsidising Its AI Chip Self-Reliance

China has expanded subsidies covering up to half of electricity costs for major data centres operated by ByteDance, Alibaba and Tencent, as Beijing intensifies efforts to promote domestic artificial intelligence chips and reduce reliance on American technology.

Local governments in Gansu, Guizhou and Inner Mongolia have expanded subsidies covering up to 50 per cent of electricity costs for data centres that use domestically produced AI processors, such as those developed by Huawei and Cambricon. The incentives specifically exclude facilities using Nvidia's advanced chips, which remain under US export restrictions.

The subsidies respond to complaints from technology groups about increased operational costs following Beijing's restrictions on purchasing Nvidia's chips.

Several technology groups complained to regulators about the increased costs of using domestic semiconductors from companies such as Huawei and Cambricon, most of which are less energy efficient than Nvidia's. With the new subsidies, electricity costs will be reduced further to about 0.4 yuan, or 5.6 cents, per kWh.

Some packages are generous enough to cover a data centre's operating costs for nearly an entire year.

Restrictions on the export of AI chips to China, while creating challenges, have not prevented Chinese laboratories from producing highly competitive models. Chinese companies have demonstrated remarkable innovation despite hardware limitations, with start-ups such as DeepSeek producing AI models that rival leading American systems at a fraction of the cost.

The competitive landscape has shifted dramatically since DeepSeek's January release of its R1 model, which matched performance with OpenAI's offerings whilst reportedly costing just 6 million dollars to develop.

Since then, Chinese AI development has expanded rapidly, with companies large and small rushing to unveil increasingly advanced models. Most releases are open source. This has prompted major Chinese technology firms including Tencent, Baidu, and ByteDance to accelerate adoption of domestic chips, with Tencent Cloud now supporting various types of Chinese processors.

The energy subsidies form part of China's comprehensive strategy to build an independent AI ecosystem whilst managing the country's growing computational demands.

Research firms have said that more than half of all AI chips used in China will be produced locally, compared with only 17 per cent in 2023. Whilst domestic chips currently lag behind Nvidia's in performance and energy efficiency, the combination of government support, competitive pricing and forced adoption due to US sanctions is rapidly accelerating improvements in Chinese semiconductor capabilities.