Threads
- Chinese AI companies are undertaking a major project to build 36 data centers, primarily in the country’s western deserts.
- The plan is to install a massive number of high-end NVIDIA AI chips, totaling over 115,000 units, to create a “hyperscale infrastructure.”
- Despite a US ban on selling high-performance GPUs to China, companies are finding ways to acquire them through third-party nations.
A report by Bloomberg shows that the Chinese AI companies are building a massive AI facility in Yiwu, China. They want to install over 115,000 of NVIDIA’s high-end AI chips in 36 data centers across the country’s western deserts.
China started to alter its approach after the US introduced limitations and penalties aimed at it. It decided to make significant investments in its own equipment and technology rather than depending on the West for its chips. The goal was to close the initial gap between China and the US and build increasingly advanced chips.
China managed to make a Huawei mobile chip that was manufactured at 7nm and is now rumored to be moving towards 5nm in 2026. Meanwhile, this sounds impressive, but in terms of performance, these chips are pretty far behind what the Americans offer.
On the other hand, China cannot compete in GPUs and although it is prohibited from buying NVIDIA graphics cards for AI. But we have already seen that they have used other methods to acquire them.
Now that Chinese AI companies want to build data centers with more than 100,000 NVIDIA H100 and H200 GPUs, they will have to figure out how to get their hands on this enormous amount of hardware.
According to reports, Chinese AI businesses are constructing 36 data centres in order to establish a “hyperscale infrastructure.” They anticipate buying and setting up more than 115,000 NVIDIA GPUs in total, which is more than Elon Musk utilised to train Grok 3 for xAI.
It’s important to note that China has been purchasing graphics cards that are meant to be prohibited for a while, even if US prohibitions are still in place. They accomplish this by buying them from third-party countries, such as the United States, and then shipping them to China. For Deepseek R1, they used the Nvidia chips to train the model, but for R1, they will use mostly Huawei chips instead of Nvidia chips.
Thank you! Please share your positive feedback. 🔋
How could we improve this post? Please Help us. 😔
[Editor-in-Chief]
Sajjad Hussain is the Founder and Editor-in-Chief of Tech4Gamers.com. Apart from the Tech and Gaming scene, Sajjad is a Seasonal banker who has delivered multi-million dollar projects as an IT Project Manager and works as a freelancer to provide professional services to corporate giants and emerging startups in the IT space.
Majored in Computer Science
13+ years of Experience as a PC Hardware Reviewer.
8+ years of Experience as an IT Project Manager in the Corporate Sector.
Certified in Google IT Support Specialization.
Admin of PPG, the largest local Community of gamers with 130k+ members.
Sajjad is a passionate and knowledgeable individual with many skills and experience in the tech industry and the gaming community. He is committed to providing honest, in-depth product reviews and analysis and building and maintaining a strong gaming community.
Join Our Community
