In Depth: Chinese Tech Giants’ Struggle to Power AI Data Center Boom

China's surging demand for artificial intelligence (AI) computing has significantly increased the energy consumption of its data centers, creating challenges in securing reliable and affordable electricity while striving to meet government green energy targets [para. 1]. One proposed solution involves relocating data centers to the western regions of the country, which are rich in wind and solar energy, but this approach faces multiple obstacles [para. 2].

In 2024, Chinese data centers consumed around 140 billion kilowatt-hours (kWh) of electricity, representing a 31% increase from the previous year, and far exceeding the national total power consumption growth of 6.8% [para. 3]. Projections by the China Academy of Information and Communications Technology (CAICT) suggest that this figure could rise to 400 billion kWh by 2030, raising the share of data centers in national electricity usage from less than 2% to 6% [para. 3]. The primary driver of this demand is the rapid rise of intelligent computing centers, so-called “AI brain factories,” which require significantly more power—20 to 130 kilowatts per server cabinet compared to just 5 to 10 kilowatts in traditional cloud data centers [para. 4]. With the intelligent computing sector projected to grow at 33.9% annually from 2022 to 2027, the pressure on energy infrastructure continues to mount [para. 5].

To reconcile their ambitious carbon neutrality goals for 2030 and 2060 with increasing data center power consumption, Chinese policymakers are incentivizing the move of data centers to western provinces [para. 6]. These areas still have drawbacks: underdeveloped infrastructure, a scarcity of talent, a small local customer base, and the intermittency of renewable power, which does not always translate into lower costs [para. 7]. In 2022, the government launched an initiative to channel computing resources from the energy-hungry east—where renewable energy makes up less than 40% of the supply—to the wind- and solar-rich west, which holds 70% of installed renewable capacity [para. 12][para. 13]. National computing hubs have arisen in regions like Ningxia, attracting investments from leading Chinese tech companies such as Tencent, Baidu, and the major telecom providers [para. 14]. Yet companies often split their operations—using western centers for training energy-intensive AI models and maintaining application-oriented centers in the east for closer customer feedback [para. 16].

A significant challenge is the reliable delivery and storage of green energy. For instance, Tencent’s Huailai data center project, which incorporates a wind and solar-powered microgrid with battery storage, can meet just 15% of annual electricity demand, with the bulk still coming from the state grid at higher prices [para. 29][para. 31]. Rooftop solar installations can usually provide less than 5% of a data center’s electricity needs, much lower than the 20–30% possible in manufacturing plants, due to spatial and policy restrictions [para. 38]. New policies might eventually ease these constraints by allowing off-site renewable generation [para. 41].

To attract data center investment, several western and northern regions are offering heavily subsidized electricity—prices as low as 0.35–0.36 yuan per kWh compared to over 0.6 yuan elsewhere [para. 44][para. 45]. Such investments can significantly boost local economies: in Qingyang, Gansu province, every yuan spent on computing generates 3–4 yuan in local output, with the local digital economy expected to surpass 100 billion yuan in 2024 [para. 49]. Nevertheless, questions linger about the longevity of these subsidies, as authorities try to balance economic growth with national carbon reduction mandates. In Inner Mongolia, for example, electricity subsidies were revoked in 2022 to comply with central government directives and tackle environmental concerns [para. 54][para. 55].