China's AI boom could quadruple data-center electricity consumption to 774 terawatt-hours by 2030, while reshaping where computing capacity is concentrated, Wood Mackenzie said in a note on Tuesday.
Wood Mackenzie expects data centers to account for 17% of China's total electricity consumption by 2060 as AI workloads drive sustained growth in computing-related power demand.
"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-center capacity is developed," said Wanting Zhao, research analyst at Wood Mackenzie.
AI training is pushing operators toward more energy-intensive infrastructure, while the rapid expansion of AI inference is creating a larger base of ongoing electricity demand.
China's government is linking computing expansion with green power through the 15th Five-Year Plan and promoting "compute-power synergy."
Data centers have historically clustered around major cities and technology hubs to reach customers, skilled workers, network infrastructure and low-latency connections.
AI growth is now making reliable, affordable and lower-carbon power increasingly important, the note said.
That shift could move latency-tolerant workloads to western China, where abundant land and growing renewable capacity support AI training, batch processing and data storage.
Eastern hubs should retain latency-sensitive workloads such as AI inference and financial services, while western regions attract more power-intensive computing.
Wood Mackenzie expects eight national hubs to anchor China's computing capacity through 2060, with renewable-rich regions gaining capacity as grid constraints persist.
Battery storage, new power systems and smart workload scheduling could let data centers shift non-urgent computing to periods of abundant renewable power or lower prices, the note said.
Wood Mackenzie expects data-center demand for green electricity to rise through 2060, while emissions could peak by 2035 as renewables expand.
Zhao said the shift toward compute-power synergy will require more than grid expansion, with data security and service-level agreement concerns complicating cross-regional workload migration.
He said operators and tenants must also participate more actively, while targeted measures should ease the heavy initial investment burden.
"The direction of travel is clear: computing is beginning to follow power, but our commercial, security, and institutional architectures must urgently evolve to make this commercially viable," Zhao said.