China is investing heavily in electricity generation and battery storage to power a rapidly expanding data center sector [1, 2].
This energy surge is critical because artificial intelligence and high-performance computing require massive, reliable electricity. By reducing the supply constraints that could limit growth, China may secure a competitive edge over the U.S. in the global AI race [1, 2].
Data center capacity in China is expected to nearly double by 2030 [2]. During this period, power demand for these facilities could reach 289 TWh [2]. This growth is driven by an annual increase in AI-driven power demand of 19% per year [2].
To sustain this trajectory, the Chinese government is prioritizing large-scale investments in renewables, and transmission infrastructure [1, 2]. These efforts aim to ensure that the energy grid can support the heavy loads required by next-generation computing—a challenge that currently complicates AI scaling in other regions [1].
By 2030, power for data centers is projected to represent 2.3% of China's total electricity demand [2]. The strategy involves a coordinated rollout of electricity generation, and battery storage to mitigate the intermittency of renewable sources [1, 2].
Former Treasury Secretary Hank Paulson and former U.S. Ambassador to China Nicholas Burns said these infrastructure developments are key factors in the bilateral tech competition [1]. While the U.S. leads in many software domains, the physical layer of energy availability is becoming a primary bottleneck for AI expansion [1].
“China’s energy expansion reduces the supply constraints that could limit U.S. AI growth.”
The competition for AI supremacy is shifting from a battle of algorithms and chips to a battle of infrastructure. If China successfully integrates its massive renewable energy output with data center demand, it removes a critical physical barrier to scaling AI. For the U.S., this underscores that the AI race is not just a software challenge but a power-grid challenge, where the ability to generate and transmit electricity determines the ceiling for computational growth.
