While data center developers in Europe and the US wrestle with NIMBYism, grid bottlenecks, and eye-watering land costs near major metros, China has taken a different approach: build where almost nobody lives.
Over 50% of China’s forthcoming data center projects are now located in the country’s northern and northwestern regions, according to BloombergNEF data as of September 2026. The shift is the product of a deliberate government initiative called “Eastern Data, Western Computing,” launched in the 2021-2022 period, which funnels new computing capacity toward areas with cheap power, cooler climates, and vast tracts of open land.
The Ulanqab phenomenon
No city better illustrates the strategy than Ulanqab, a prefecture-level city in Inner Mongolia. Since 2016, nearly 100 data center facilities have either been completed or are under construction there, collectively representing 12.5 gigawatts of pledged capacity. A substantial chunk of those commitments came in just the past year alone.
The list of tenants reads like a who’s who of Chinese tech: Alibaba, Tencent, and ByteDance are all expanding in the area, with ByteDance alone scaling up by 5-6 GW in Ulanqab. DeepSeek, the AI startup that rattled global markets earlier in 2025 with its efficient large language models, is targeting roughly 1 GW of capacity. Envision, a renewable energy company, is planning a 2 GW “Galaxy Campus” in partnership with Tencent.
Electricity prices in western regions run approximately 50% lower than in eastern areas, with rates as low as 0.34-0.38 yuan per kilowatt-hour.
Why the West and Europe can’t easily replicate this
American and European data center developers face a fundamentally different set of constraints. In Northern Virginia, the world’s largest data center market, new projects have been delayed by years because the local utility simply cannot deliver enough electricity fast enough. In the UK, grid connection wait times stretch beyond a decade in some areas. In Germany, community opposition to large industrial facilities near residential zones has slowed permitting.
China’s remote regions sidestep most of these problems simultaneously. The land is cheap and abundant. The grids in places like Inner Mongolia were built to export power from wind and solar farms that often had nowhere to send their electrons. Data centers give those electrons a purpose, turning a grid management headache into an economic development tool.
There is a tradeoff. These remote sites are hundreds of miles from the major population centers generating the data. For latency-sensitive applications like real-time trading or video conferencing, that distance matters. But for the batch-processing workloads that dominate AI model training, a few extra milliseconds of network latency is a rounding error.
The capacity trajectory
China’s total data center capacity is projected to nearly double over the next five years, reaching 329 terawatt-hours by 2030. By 2028, the northern and northwestern regions are expected to outpace China’s major cities in total computing capacity.
China is also targeting having 80% of data center power sourced from renewable energy by 2030. Many western regions have surplus wind and solar capacity that was previously curtailed because demand didn’t exist nearby. Data centers provide baseload demand that improves the economics of renewable projects, creating a feedback loop: more data centers justify more renewable investment, which further lowers power costs, which attracts more data centers.