China proposes establishing 39 AI data centers equipped with 115,000 limited Nvidia Hopper GPUs, sparking concerns about procurement and the potency of trade restrictions.
The U.S. government has confirmed the presence of unauthorized Nvidia stock in China, but doubts the possibility of supplying over 100,000 restricted processors to one country. However, the origin of these GPUs remains a mystery, with plenty of smuggled GPU servers found in China.
One of the planned data centers, located in Yiwu County, Xinjiang, could become one of the most powerful AI data centers in China. This data center is expected to house around 80,500 NVDA GPUs, potentially used for training advanced large language models and large reasoning models.
The data center's location was chosen for its access to solar and wind energy, inexpensive land, and high elevation, which helps to cool hardware. A large solar power tower has been erected in Yiwu to provide consistent electricity for the data centers.
Chinese government statements indicate that Xinjiang already has a data center providing 24,000 PetaFLOPS of processing power, equivalent to around 12,000 NVDA H100 GPUs. The planned data center in Yiwu could significantly add to this power.
The planned data center projects, which include 39 new AI data centers mostly in Xinjiang and Qinghai, would require the procurement of more than 14,000 servers using either H100 or H200 processors, worth billions of dollars on China's black market.
U.S. officials have asked Malaysian authorities to take action against unauthorized technology transfers, and Malaysia has stated it will act if presented with solid evidence. Singapore is currently prosecuting individuals for allegedly exporting AI servers containing restricted components to Malaysia, potentially ending up in China.
NVDA has repeatedly stressed that there is no indication of large-scale diversion of its GPUs to China from other countries. However, smuggling through Southeast Asia has become a point of concern, especially in Malaysia and Singapore.
Despite the lack of direct proof, work on the new AI data centers, particularly in Yiwu, continues. NVDA has stated that building functioning AI infrastructure with unofficial and/or used parts is both risky and impractical, and that it provides no operational or technical support for restricted products in China or elsewhere.
No credible public information confirms any Chinese companies owning over 80,000 NVDA Hopper H100 and H200 GPUs, and acquisition despite US export restrictions remains unverified. Estimates on the total number of restricted GPUs in China vary, with some senior officials in the Biden administration mentioning a figure closer to 25,000.
Chinese companies are planning to equip these data centers with high-performance NVDA Hopper GPUs, with one of the larger projects run by Nyocor, which plans to install 625 H100 DGX servers with around 5,000 H100 accelerators. The H100 GPUs are significantly faster than cut-down H20 GPUs, with H100 being 3.3 - 6.69 times faster than H20 with AI data formats and 1.52 - 64 times faster with HPC data formats.
Xinjiang, particularly the Hami region, is already a hub for renewable power and the local government is promoting the development of computing infrastructure using its vast energy reserves. The remaining 30% of the GPUs will be spread across at least 38 additional data center projects, primarily in Xinjiang and Qinghai.
The development of these AI data centers in China is a topic of ongoing investigation and concern for U.S. officials. As more information becomes available, the situation is likely to evolve.
Read also:
- Exploring Harry Potter's Lineage: Decoding the Enigma of His Half-Blood Ancestry
- Elon Musk Acquires 26,400 Megawatt Gas Turbines for Powering His AI Project, Overlooks Necessary Permits for Operation!
- U Power's strategic collaborator UNEX EV has inked a Letter of Intent with Didi Mobility to deploy UOTTA(TM) battery-swapping electric vehicles in Mexico.
- Commercial-grade hydrogen enhancement systems manufacturing initiated by H2i Technology
