China’s effort to run its rapidly expanding AI data centers industry on renewable energy is running into a series of practical obstacles, with industry experts warning that the sector’s power needs are too unpredictable to mesh well with green energy suppliers, while grid operators are pushing back against arrangements that could cut into their revenues.
Reliable electricity supply for AI-focused data centers has been elevated to a strategic national priority. China’s 2026 government work report, released earlier this year, committed to deeper integration between the country’s computing infrastructure and its power supply networks. Central to that goal is an ambitious target requiring renewables to cover 80% of the data center sector’s total electricity consumption by 2030, a dramatic jump from the 11% share recorded in 2023.
The scale of the challenge is considerable. Power demand from Chinese data centers is projected to grow by between 300 billion and 500 billion kilowatt-hours from 2026 to 2030, a figure that would represent 18% of all electricity demand growth in China over that period, according to Pei Shanpeng, a director at State Power Investment Corp, one of China’s largest state-owned energy companies. To put the lower end of that range in perspective, it is roughly equivalent to the United Kingdom’s total annual power consumption.
Despite this enormous and growing appetite for electricity, experts say AI data centers are a poor match for renewable energy providers compared with other heavy industrial users like aluminum smelters. The core problem is unpredictability. Unlike smelters, which draw power at relatively steady and foreseeable rates, data centers are difficult to schedule around because their peak loads can shift sharply depending on computational workloads.
“At least for now, they do not appear to be very flexible in managing power demand,” Pei said at an industry conference in Beijing last week. “From what we understand, they cannot really adjust power consumption load much. GPUs are very expensive, so once they are purchased, operators want to use them as quickly and as intensively as possible.”
He added that the drive to connect data centers to green power sources is motivated primarily by the goal of reducing emissions rather than cutting electricity costs for operators.
A second set of obstacles involves the grid operators themselves. Experts say that broader adoption of direct green-power supply arrangements, which would link renewable generators straight to data centers and bypass parts of the traditional transmission network, is meeting resistance from grid companies. Those operators worry that such setups would reduce electricity sales flowing through their networks, making it harder for them to recoup the substantial investments they have made in transmission and distribution infrastructure. If demand through conventional grid channels slows or declines, recovery of those costs becomes harder to guarantee.
China’s push to build dedicated grid connections for AI workloads is also coming at a moment when its rapid data center construction has already started straining the broader power system in certain regions. The surge in facilities is raising both average and peak grid loads, forcing operators to find ways to balance rising demand with the need to maintain reliability.
Wang Zelin, deputy director at State Grid Jibei Electric Power Research Institute, framed the scale of the reliability challenge. “If 15% of the power consumption loads can be adjusted, it will significantly reduce capacity expansion pressure on the grid over the next three to five years,” he said, underscoring how much room there is for improvement in demand flexibility.
