Inner mongolia emerges as core of china's AI expansion

Traveling just two hours west of Beijing by train leads to Inner Mongolia, a region known for its vast grasslands and coal mining. In recent years, it has become a key location for AI data centers. Ulanqab, a city in Inner Mongolia with a population of about 1.5 million, has seen nearly 100 data centers opened or under construction since 2016. Chinese companies have committed to building projects with a combined estimated capacity of 12.5 gigawatts, with over 70 percent of these commitments announced in the last year. This makes Ulanqab one of the fastest-growing compute clusters in Asia, according to a Goldman Sachs research note. For comparison, OpenAI’s Stargate Project is expected to reach only 10 gigawatts of total capacity. Chinese companies are investing heavily in Ulanqab due to its high elevation, long cold winters, and relatively low electricity costs. The region benefits from a strong growth in wind and solar energy, as well as an abundant supply of coal. Notably, Chinese AI firms are now building their own data centers rather than renting from cloud providers. Companies such as DeepSeek, ByteDance, Alibaba, and Xiaohongshu are all investing in Ulanqab. This marks a shift from previous years, when Chinese AI companies spent less on physical infrastructure compared to their American counterparts. One major challenge is the scarcity of water. Ulanqab receives only about 14 inches of rain annually, and the local government is already struggling to meet the water needs of its residents. Last month, the city’s water company had to shut down several waterworks for seven hours each night to manage peak demand. While data centers will require less water in the winter, the new infrastructure could still pose a significant environmental challenge. Inner Mongolia has been a data center hub for over a decade, with Huawei establishing its first center in Ulanqab in 2016 and Apple following three years later. In 2021, the region was designated as a key hub for China’s 'Eastern Data, Western Compute' initiative. Initially, these data centers faced high latency due to their remote location, limiting their use to backup storage. However, the rise of AI in 2022 changed this, as model training can take months and does not require real-time adjustments, making latency less of an issue. Ulanqab is relatively close to Beijing and other major cities in China, and two dedicated fiber optic cables built in 2017 and 2019 have reduced average latency to under five milliseconds, supporting real-time data exchanges. According to Andrew Stokols, a professor at Singapore Management University, Ulanqab’s growth is driven more by commercial demand than government-led investment. As Chinese AI startups like DeepSeek, Moonshot AI, and Zhiput AI attract more users, they have found a viable business case for inference data centers. The Chinese government sees data centers in Inner Mongolia as a way to absorb excess renewable energy. Damien Ma, director of Carnegie China, notes a positive correlation between regions with the most unused renewable energy and those with the most data centers. This is a win-win strategy, allowing China to catch up with the U.S. in data center construction while increasing demand for renewable energy. Envision, a major Chinese wind turbine manufacturer, recently announced plans to build a 2 gigawatt AI data center in Ulanqab connected directly to its own clean power supply. However, experts caution that the link between AI data centers and renewable energy in China is not as straightforward. Stokols found that about 37 percent of electricity in Ulanqab still comes from coal, which is preferred for its reliability. Damien Ma describes Inner Mongolia as the 'West Virginia of China,' a coal region now racing to replace coal with wind and solar. While the transition remains uncertain, data centers in the region will likely continue to rely on coal for now. Ma suggests that in three years, the region could be fully powered by renewables.