The race for artificial intelligence supremacy in China is driving tech companies out of the coastal megacities and into the northern steppes. The South China Morning Post reported that RedNote, the popular lifestyle social media platform known domestically as Xiaohongshu, is planning a massive 15 billion yuan ($2.2 billion) data center in Inner Mongolia, a move that underscores a fundamental shift in the Chinese tech sector.
Social media and e-commerce platforms are no longer just software companies; they are becoming heavy infrastructure players. As AI becomes deeply integrated into core product offerings, securing proprietary, large-scale compute is now viewed as an existential necessity rather than a competitive luxury.
The Ulanqab Buildout
RedNote is targeting Ulanqab, a city in Inner Mongolia renowned for its wind energy resources, for the new facility. The company is reportedly in talks with VNET, one of China’s largest data center operators, to manage the buildout, with a targeted delivery date in the first quarter of 2027.
The $2.2 billion budget is notably exclusive of semiconductor costs, meaning the final investment, once outfitted with tens of thousands of AI accelerators, will be significantly higher. A 600-megawatt capacity places the planned facility among the largest dedicated AI data centers in the country, rivaling the infrastructure investments of established cloud providers like Alibaba and Tencent.
RedNote’s move follows a similar announcement from DeepSeek, which is planning a 1-gigawatt facility in the same city. The simultaneous buildout by two of China’s most prominent AI players transforms Ulanqab from a regional industrial center into one of the most strategically important locations in the global AI race.
AI at the Core of RedNote’s Strategy
For RedNote, the massive investment is driven by the rapid integration of AI into its platform. The company, which recently filed for a confidential IPO in Hong Kong, relies heavily on sophisticated recommendation algorithms to serve content to its hundreds of millions of active users.
Beyond recommendations, RedNote is aggressively developing generative AI features. The company recently debuted its Dots-Note 3.0 model, which achieved a perfect score on the International Mathematical Olympiad benchmark, signaling serious ambitions in foundational model development. Training and serving these models at scale requires immense computational power, and relying solely on third-party cloud providers exposes the company to pricing volatility and potential capacity bottlenecks.
The Inner Mongolia Advantage
RedNote’s choice of Ulanqab mirrors a broader industry trend. As power consumption becomes the primary constraint on AI scaling, tech companies are migrating to regions with abundant, cheap electricity. Inner Mongolia’s vast renewable energy projects, particularly wind power, offer a sustainable and cost-effective solution to the staggering energy demands of AI training clusters.
Furthermore, the region’s naturally cool climate significantly reduces the energy required for server cooling, further improving the power usage effectiveness of the facilities. This geographic shift aligns with Beijing’s “East Data, West Computing” initiative, a national strategy designed to transfer computing resources from the energy-constrained east to the resource-rich west.
The Hardware Challenge
While the land and power may be secured, RedNote faces the same critical challenge as every other Chinese AI developer: hardware procurement. Outfitting a 600-megawatt data center requires a massive volume of advanced AI chips.
With US export controls restricting access to Nvidia’s top-tier GPUs, RedNote will have to navigate a complex procurement landscape. The company will likely rely on a mix of domestically produced chips, such as Huawei’s Ascend series, and whatever Nvidia hardware can be legally acquired or sourced through alternative channels. The success of the $2.2 billion Ulanqab project will ultimately depend not just on concrete and power lines, but on RedNote’s ability to secure the silicon needed to bring the facility online at full capacity.
The hardware challenge is, in many ways, the defining constraint of China’s entire AI buildout. Every gigawatt of data center capacity announced by a Chinese tech company is ultimately a bet that domestic chip production, from Huawei’s Ascend series to Cambricon’s latest accelerators, will scale fast enough to fill those racks. RedNote’s Inner Mongolia project is both a statement of ambition and a test of whether China’s semiconductor industry can deliver on the promise that Beijing has staked its AI strategy upon.
For investors watching RedNote’s pending Hong Kong IPO, the data center announcement is a double-edged signal. On one hand, it demonstrates the company’s seriousness about AI and its willingness to make the capital commitments necessary to compete with Alibaba and Tencent at the infrastructure level. On the other hand, a $2.2 billion commitment, exclusive of hardware costs, represents a massive capital outlay for a company that has not yet gone public.
The success of the IPO may well depend on whether investors view this infrastructure bet as a visionary move or an overextension, and that judgment will ultimately hinge on whether RedNote can demonstrate that its AI features are driving measurable improvements in user engagement and monetization. In a market where AI infrastructure spending is under intense scrutiny, RedNote’s willingness to make this bet before going public is either a sign of exceptional confidence or a calculated gamble that the company cannot afford to lose.