Journey simply two hours west of Beijing by practice, and also you’ll end up surrounded by the rolling grasslands and historic cinder cones of Internal Mongolia. This huge, arid land has lengthy been China’s capital of sheep farming and coal mining, however over the previous couple of years, it has grow to be the most well liked place within the nation to construct an AI knowledge middle.
In Ulanqab, a metropolis in Internal Mongolia residence to about 1.5 million folks, almost 100 knowledge facilities have been opened or begun development since 2016. Chinese language firms have pledged to construct initiatives with a mixed estimated capability of 12.5 gigawatts within the metropolis, and over 70 % of the entire commitments have been introduced in simply the final 12 months, making it one of many quickest rising compute clusters in Asia, based on a analysis notice printed by Goldman Sachs final week. For comparability, OpenAI’s $500 billion Stargate Challenge is ready to succeed in solely 10 gigawatts of complete capability when it’s full.
Chinese language firms are flocking to Ulanqab for quite a few causes. Town sits at excessive elevation on the Internal Mongolian Plateau and has lengthy, chilly winters, which suggests knowledge facilities there don’t want to make use of as a lot vitality to remain cool. It’s additionally comparatively near Beijing, so knowledge may be transmitted to China’s populous areas with minimal latency. However essentially the most engaging issue has to do with prices. Electrical energy is cheaper in Internal Mongolia than nearly wherever else in China, pushed by each the sturdy development of wind and photo voltaic vitality and an plentiful provide of coal.
What’s additionally fascinating is who is constructing these knowledge facilities. For the primary time, Chinese language AI firms are making huge investments in their very own infrastructure, moderately than renting compute from cloud firms. DeepSeek is reportedly constructing an enormous AI knowledge middle in Ulanqab, as are ByteDance, Alibaba, and Xiaohongshu. For years, Chinese language AI firms have spent far much less on constructing bodily infrastructure than their American friends, regardless of creating quite a few well-liked AI fashions with spectacular capabilities. The Ulanqab knowledge middle growth indicators that now they’re lastly beginning to catch up.
There’s only one drawback: discovering sufficient water. Ulanqab is about as dry as Denver, getting solely roughly 14 inches of rain every year. The native authorities is already struggling to supply sufficient water to fulfill resident demand—earlier than most of the deliberate knowledge middle initiatives are even up and operating. Final month, the native water firm in Ulanqab was compelled to turn off a number of waterworks for seven hours every evening to mitigate peak demand. The info facilities being constructed within the metropolis will want much less water within the winter—climate knowledge from the native authorities of Ulanqab reveals they solely require further water for cooling throughout two months out of the 12 months—however the entire new infrastructure might nonetheless pose a big environmental problem for the area.
The Boonies
Internal Mongolia has been a knowledge middle sizzling spot for at the very least a decade, lengthy earlier than the present AI growth. Huawei constructed its first one in Ulanqab in 2016, and Apple adopted go well with three years later. In 2021, the world was designated as one of many fundamental hubs of a country-wide authorities challenge dubbed “Japanese Information, Western Compute,” which goals to construct knowledge facilities within the Western hinterlands of China.
There was one main downside, although. As a result of they’re situated removed from China’s populous japanese coast, these knowledge facilities initially confronted excessive latency charges when transferring knowledge to nearly all of customers. In consequence, they had been initially largely relegated to backup storage—till AI gave them a brand new goal. “With the rise of AI in 2022, there was the conclusion that really, these distant knowledge facilities could possibly be well-utilized for mannequin coaching,” says Andrew Stokols, a professor at Singapore Administration College who research China’s compute infrastructure. A coaching run for an AI mannequin can take months and doesn’t require a lot real-time tinkering, so latency is much less of a problem.
Comparatively talking, Ulanqab can also be not likely that far-off. Internal Mongolia is far nearer to Beijing and different main metropolitan areas in China than any Western knowledge middle hub is. And it’s now related by two devoted fiber optics cables in-built 2017 and 2019 that diminished common latency speeds to lower than 5 milliseconds, quick sufficient to help real-time knowledge exchanges like AI inference.