DSET Energy Resilience Program Deputy Director Cartus Bo-Xiang You and Policy Analyst Angela Glowacki recently contributed an op-ed to The Strategist, analyzing the challenges facing China’s rapid expansion of AI-focused data center infrastructure under the “Eastern Data, Western Computing” (EDWC) initiative.
The authors argue that abundant electricity alone does not guarantee AI competitiveness. While China has dramatically expanded its computing infrastructure and renewable energy capacity, many newly built data centers in western regions remain underutilized due to transmission bottlenecks, latency constraints, fragmented standards, and a mismatch between infrastructure placement and real-world AI demand.
According to the authors, China’s drive to develop a nationwide computing network has unintentionally fueled speculative overbuilding and inter-provincial competition. Local governments, spurred by economic growth targets and subsidy incentives, accelerated data center construction—even in regions lacking adequate connectivity, talent pools, or customer demand. As a result, some facilities reportedly operate at utilization rates as low as 20–30 percent, far below Beijing’s original targets.
The analysis highlights that AI infrastructure depends not only on energy supply, but also on broader ecosystem coordination—such as fiber-optic transmission, low-latency networking, hardware interoperability, and skilled operators. While western provinces benefit from cheaper renewable energy and colder climates, their geographic remoteness creates operational limitations for latency-sensitive AI inference workloads concentrated in eastern economic hubs.
The authors further observe that Beijing has begun to reassess the EDWC initiative by tightening oversight of new data center projects and introducing utilization benchmarks to curb inefficient expansion. China has also launched electricity subsidies linked to domestically produced semiconductors, aiming to boost facility usage following restrictions on advanced foreign AI chips.
The article concludes that China’s AI ambitions continue to face structural coordination challenges, despite advantages in electricity generation and state-backed infrastructure investment. The researchers caution that without significant improvements in network integration, workload distribution, and governance coordination, many data centers risk becoming stranded assets rather than productive AI infrastructure.
