DeepSeek, China's leading AI inference company, plans construction of a 1-gigawatt AI data center in Inner Mongolia.
DeepSeek is developing a one-gigawatt AI data center in northern China, a project that would substantially expand the startup's computing infrastructure as it competes with better-funded US rivals.
The campus is planned for Ulanqab in China's Inner Mongolia region, approximately 350 kilometers (217 miles) from Beijing, Bloomberg News reported, citing people familiar with the project. Part of the facility is expected to begin operating by the end of 2027. DeepSeek has not publicly announced the project, and details remain unclear regarding cost, chip suppliers, and construction timeline.
A one-gigawatt power allocation would place the facility among the largest AI computing developments undertaken by a Chinese model developer, representing a significant shift from the lean infrastructure strategy that helped DeepSeek gain global attention after its R1 reasoning model demonstrated competitive performance at a fraction of reported US development costs.
The project builds on earlier signals of DeepSeek's Ulanqab presence. In April, the company advertised positions for a data-center engineer and delivery manager in the city, with responsibilities encompassing AI server cluster maintenance, construction oversight, and facility commissioning.
Ulanqab has become a preferred location for large computing projects due to its cool climate, available land, and access to wind and solar power. The city's average annual temperature is approximately 4.3 degrees Celsius, enabling operators to use outside air for cooling throughout much of the year and reducing data center electricity consumption significantly.
The location aligns with China's "East Data, West Computing" strategy, which relocates energy-intensive computing from densely populated coastal regions to areas with cheaper land and greater power resources. China's National Development and Reform Commission has designated Inner Mongolia as a national computing hub, calling for power-usage-effectiveness ratios below 1.2 and increased renewable energy adoption. The regional hub is intended to support computing demand in Beijing, Tianjin, and other eastern markets.
Access to advanced processors remains a central constraint for Chinese AI companies. DeepSeek's V3 model was trained on a cluster of 2,048 Nvidia H800 processors, according to the company's technical report. Washington has subsequently tightened controls on advanced Nvidia chip shipments to China.
DeepSeek is developing its own inference processor, though the effort remains in early stages. The company has also partnered with Huawei Technologies as Chinese developers seek to reduce reliance on US hardware.
Building a gigawatt-scale campus would require DeepSeek to secure large volumes of processors, networking equipment, and electricity—potentially testing China's domestic AI-chip supply capacity and efficiency. The facility's computing performance will depend on processor availability, installation decisions, and effective linkage into large training clusters.
This expansion follows DeepSeek's substantial funding growth. The company completed its Series A in June, raising more than $7 billion at a valuation exceeding $50 billion. It is now preparing for a potential mainland Chinese initial public offering and could submit an application this year for a 2027 debut, Bloomberg reported.
The capital enables DeepSeek to own the computing infrastructure underlying its models rather than relying primarily on leased capacity. It also reflects how competition among Chinese AI developers is increasingly shifting from model design toward control of chips, electricity, and data-center capacity.