Ulanqab is expanding its role as an artificial-intelligence computing hub by combining relatively low electricity costs, a cool climate and network access to Beijing. The city is roughly 350 kilometers from Beijing, and reported round-trip network latency is about four milliseconds.
The location’s appeal is also reflected in the scale of construction. More than 30 data centers can be seen within a 10-minute drive in Ulanqab’s Chahar High-Tech Zone, where additional facilities are still being built. Alibaba Cloud selected the city as a primary base for its AI-era operations after previously relying on Zhangjiakou as its largest cloud-computing site. The company said a data center of the same scale could save about 5 billion yuan in annual electricity costs in Ulanqab compared with other cities.
Cooling is a second source of potential efficiency. Lower temperatures allow data centers to use outside air and other natural-cooling methods for much of the year, reducing the energy required to remove heat from servers.
Construction speed is becoming another competitive factor as AI hardware demand grows. An Alibaba Cloud facility in Ulanqab uses a modular design that reduces delivery time to about 100 days, compared with a standard six-to-12-month construction period. Faster completion can bring computing equipment into service sooner, but it does not by itself establish that every project will achieve the same utilization, operating cost or financial return.
The leasing market was broadly quiet from Oct. 12 to 19, with 18 of 19 monitored points unchanged. The only adjustment was for Yangtze River Delta A100-80G capacity: the lowest monthly quote fell from 38,000 yuan to 34,000 yuan per unit, the average declined from 39,000 yuan to 37,000 yuan, and per-card-hour pricing dropped about 5. 2%, while the highest quote stayed at 40,000 yuan. Those are tax-inclusive market quotations rather than confirmed transaction prices, so they show movement at the lower end without proving a broad decline across high-end GPU leasing.
The economics therefore depend on more than wind power, temperature or proximity alone. Electricity supply arrangements, cooling performance, network stability, construction delivery, equipment utilization and customer demand will determine whether Ulanqab’s resource advantages become durable computing-service competitiveness. The next test is whether the expanding capacity can translate into actual orders and sustained use rather than only signed projects or planned racks.