Game On: Latest in Gaming News, Reviews & Industry Buzz
- AI investments are driving rapid growth in computing demand and increasing energy requirements across China's data-centre sector.
- Low-latency services remain in cities like Beijing, Shanghai, and Shenzhen; training, backup and storage shift to cheaper, cooler regions.
- Flexible scheduling for latency-tolerant workloads can align demand with renewable generation, making data centres active parts of the power grid.
- Energy capacity limits could constrain growth; analysts warn emissions may peak without accelerated renewable rollout and grid expansion.
- Chinese data centers could use more electricity than the entirety of South Korea by 2030
- Data centers are being relocated in cheaper regions when latency isn’t crucial
- Flexible grid consumption could also play a role for less important work
All the data centers in China are expected to consume a combined 774 TWh of electricity annually by the end of the decade, marking around a 4x increase over their current consumption, as the country continues to drive major AI investments, new reports have claimed.
Wood Mackenzie researchers anticipate China’s data centers to represent around 6% of China’s total electricity consumption by 2030, a figure that’s expected to continue climbing to as much as 17% by 2060.
Data centers aside, a 6% share by 2030 puts the country’s total estimated annual electricity consumption at a staggering 12,900 TWh by the end of the decade, highlighting immense pressure on grids to expand.
Latest Videos FromTechRadar
Data centers could account for 6% of China’s consumption by 2030, 17% by 2060
Energy consumption is set to come from two key areas when it comes to artificial intelligence. Firstly, the predictable inference-related consumption that comes from running models for users and businesses. But secondly, unpredictable and high spikes will come from training as models become increasingly sophisticated.
The country is also seeing a split in where data centers are being located, in a bid to cope with surging power demand. Large facilities near cities like Beijing, Shanghai and Shenzhen offer low-latency connections for immediate performance, but tasks that can accept slightly poorer latency such as training, backup and storage are being moved out of town.
Western and northern regions of China, for example, promise cheaper land, cooler climates and greater access to renewable energy.
“As AI drives sustained growth in computing demand, access to reliable, cost-competitive and lower-carbon electricity will play a growing role in determining where and how new data-centre capacity is developed,” research analyst Wanting Zhao wrote.
Energy is a major data center constraint
Although analysts predict that data center-related energy consumption will continue to rise for decades, Wood Mackenzie expects data center emissions to peak around 2035, implying that renewable sources could be slightly delayed.
Analysts also project that data centers could become more flexible in terms of how they put pressure on the grid, in a similar way to how latency-tolerant workloads could be moving away from the city. For example, non-urgent workloads could be scheduled for periods when renewable production is cheaper or other grid demand is lower.
“This flexibility could allow data centres to move beyond their traditional role as passive base loads and become a more active component of China’s evolving power system,” the company wrote.
As for the 774 TWh figure, that puts China’s data centers markedly ahead of South Korea’s total energy consumption across domestic and industrial measurements. The country as a whole generated around 596 TWh in 2025, per utility statistics.
That 774 TWh figure could also be set to change as we close in on the end of the decade – just a year ago, Wood Mackenzie had predicted that China’s data centers would consume 479 TWh annually by 2030, making this year’s projection around 60% higher.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.
Read the full article on the original site
