Data centres in China have been told to stick with NVIDIA chips due to difficulties in migrating to domestic chip and software! Here is what you need to know…
China told to stick with NVIDIA chips amid domestic challenges
China’s push for self-reliance in advanced semiconductor technology has faced a significant roadblock due to the ongoing US export restrictions targeting critical components like GPUs. These sanctions have severely limited China’s access to NVIDIA’s cutting-edge chips, key hardware for powering artificial intelligence (AI) development and other high-performance computing applications.
As a result, China has been accelerating its efforts to develop domestic GPU technologies. However, as the country strives for self-sufficiency, it is encountering significant hurdles due to compatibility issues, engineering challenges, and high costs associated with shifting away from NVIDIA chips and software.
The Challenges of Switching to Domestic Chips
While Chinese startups and government-backed enterprises have been developing their own GPU alternatives, the transition from NVIDIA chips to homegrown solutions has proven to be more complex and costly than expected.
For one, the differences in both hardware and software between NVIDIA’s products and domestic alternatives make it difficult for Chinese data centers to seamlessly switch. These data centers rely on NVIDIA’s highly optimised ecosystem of hardware and software, and moving to domestic chips requires extensive engineering work to ensure compatibility, stability, and efficiency.
This transition process is not only technically challenging but also expensive. Many Chinese data centers are heavily invested in NVIDIA hardware and software infrastructure, which means that switching to domestic alternatives could result in significant costs.
Recommended : China Still Has Access To High-Speed NVIDIA AI Chips!
According to a report from the China Academy of Information and Communications Technology (CAICT), shifting to domestic GPU solutions would require considerable investment in both hardware and software engineering. The think tank’s research emphasized that for many data centers, continuing to use NVIDIA chips—despite the export restrictions—may remain the more cost-effective option, at least in the short to medium term.
Domestic Solutions as a Last Resort
The CAICT report suggests that Chinese data centers should stick with NVIDIA GPUs wherever possible due to the high costs and complexity involved in switching to domestic alternatives. Despite the rapid growth of China’s AI computing capacity and an increase in the number of data centers under construction, the report cautioned that many Chinese data centers are struggling with fragmentation in computing power. This fragmentation often leads to underutilization of resources, with GPU average use rates often falling below 40 percent.
If the conditions allow, [data centres] can choose [NVIDIA’s] A100 and H100 high-performance computing units. If the need for computing power is limited, they can also choose H20 or alternative domestic solution.
The trend of computing power fragmentation is increasingly severe, with GPU average use rates less than 40 per cent…There are big discrepancies on hardware in IDCs, such as in GPUs, AI accelerators and network structures, which made it harder to manage and dispatch hardware resources to accommodate for differential computing needs of AI tasks, further impeding the use rate.
Recommended : How NVIDIA A800 Bypasses US Chip Ban On China!
NVIDIA’s A100 and H100 GPUs, in particular, have been central to the efficient functioning of AI workloads in China’s data centers. These GPUs are specifically designed for high-performance computing tasks, such as deep learning and AI model training, providing the necessary computational power for modern AI systems. However, the ongoing restrictions on these chips, paired with the engineering complexity of transitioning to domestic alternatives, have forced some data centers to reconsider their hardware choices.
NVIDIA attempted to mitigate the impact by introducing modified versions of these chips—namely the A800 and H800. However, these detuned chips also fell victim to US sanctions in October 2023, leaving China with limited access to powerful AI chips.
The report highlighted that if computing needs are more limited, Chinese data centers could consider alternative domestic solutions, such as the H20 GPU developed by the Chinese tech giant HUAWEI. However, these alternatives are still a far cry from the performance of NVIDIA’s flagship products, and they may not meet the demands of large-scale AI applications or high-performance computing tasks.
As a result, the need for a sustainable domestic solution remains pressing. China’s rapidly developing domestic GPU market has made some strides, with a number of Chinese startups working on new chip technologies that could eventually rival NVIDIA’s dominance. However, China is stuck with NVIDIA chips for the foreseeable future.
Please Support My Work!
Support my work through a bank transfer / PayPal / credit card!
Name : Adrian Wong
Bank Transfer : CIMB 7064555917 (Swift Code : CIBBMYKL)
Credit Card / Paypal : https://paypal.me/techarp
Dr. Adrian Wong has been writing about tech and science since 1997, even publishing a book with Prentice Hall called Breaking Through The BIOS Barrier (ISBN 978-0131455368) while in medical school.
He continues to devote countless hours every day writing about tech, medicine and science, in his pursuit of facts in a post-truth world.
Recommended Reading
Go Back To > Business | Computer | Tech ARP
Support Tech ARP!
Please support us by visiting our sponsors, participating in the Tech ARP Forums, or donating to our fund. Thank you!