Chinese Engineers Reportedly Accessing NVIDIA’s High-End AI Chips Through Decentralized “GPU Rental Services”

Chinese engineers have found a rather interesting loophole in US export controls, allowing them to access NVIDIA's cutting-edge AI accelerators.

China Continues To Find Loopholes In US Policies; Despite Sanctions, The Country Still Has Access To Cutting-Edge AI Chips

Despite restrictions by the Biden administration to prevent the access of China to high-end AI equipment, it seems like the nation always finds a workaround, and this time, they have indeed come up with a surprising method of accessing the power of Team Green's chips, and rather than acquiring them, Chinese engineers and firms are more inclined towards "rental services," which we'll discuss later on. There were reports that NVIDIA's H100s were being sold in China's black markets for outrageous prices, but Chinese engineers have found a cost-effective workaround now.

Related Story NVIDIA Introduces NIM Agent Blueprints, Gives Enterprise The Ability To Create Their Own AI Experiences

The Wall Street Journal reports that Chinese engineers are working with specific global brokers to get access to high-end computing power, and they are using cryptocurrency as a means for financial payments to maintain anonymity. While the presence of NVIDIA's high-end AI accelerators is still there in the Chinese markets, individuals see the rental option as more viable, given that the costs for acquiring components are significantly larger than just using it for a short time.

WSJ discloses that Chinese customers are reportedly more inclined towards a Bitcoin miner, Derek Aw, for computing power rental services since Aw has managed to convince investors to set up large-scale AI clusters utilizing NVIDIA's H100 AI chips. The entrepreneur is said to have set up 300 servers in a data center in Brisbane, Australia, and is on his way to further expand his rental business, given the massive interest shown by the markets.

Interestingly, this mode of business is being adopted by the mainstream tech giants out there, as we previously