China is stepping up efforts to develop viable alternatives to Nvidia’s chips that run artificial intelligence (AI) as the government continues to try to move away from U.S. technology.
The U.S. sanctions imposed on China over the past few years and NVIDIAAnalysts told CNBC that China’s dominance in the sector poses a major challenge to Beijing’s efforts, at least in the short term.
Nvidia’s well-known rapid growth has been driven by major cloud-computing companies buying server products equipped with the company’s graphics processing units (GPUs). These chips have allowed companies such as ChatGPT maker OpenAI to train huge AI models with vast amounts of data.
These AI models will serve as the basis for applications such as chatbots and other emerging AI applications.
The U.S. government has been restricting exports of Nvidia’s most advanced chips to China since 2022, with restrictions tightening last year.
These semiconductors are key to China’s ambitions to become a leader in AI.
CNBC spoke to analysts who have identified some of the leading Chinese contenders looking to challenge Nvidia, including tech giant Huawei. Alibaba and Baidu Startups such as Biren Technology and Enflame are also participating.
The overall view is that they are lagging behind Nvidia at the moment.
“These companies have made impressive progress in developing AI chips (ASICs) tailored for specific applications,” Wei Sun, senior analyst at Counterpoint Research, told CNBC.
“However, competing with Nvidia remains a big challenge, especially in the technology gap for general-purpose GPUs. It will be difficult to match Nvidia in the short term.”
China’s main challenges
According to Sun, one of the challenges facing Chinese companies is a “lack of technological expertise.”
But the biggest obstacle to China’s ambitions is US sanctions and their knock-on effects.
Some of Nvidia’s major Chinese competitors are on the U.S. Entity List, a blacklist that limits their access to U.S. technology. A number of U.S. regulations limit the export of key AI-related semiconductors and machinery to China.
Chinese GPU makers do all the chip design and rely on manufacturers to make the chips. For a while, this Taiwan Semiconductor Manufacturing Co., Ltd.TSMC, or TSMC, but due to U.S. regulations, many of those companies don’t have access to chips made by TSMC.
Therefore, they SMICChina’s largest chipmaker, SMIC, is several generations behind TSMC in technology, partly because Washington has restricted SMIC’s access to key machinery from the Dutch company. ASMLIt’s what’s needed to manufacture cutting-edge chips.
The smartphone displays the Chinese and American flags, with an NVIDIA chip visible in the background.
Ra | Nurphoto | Getty Images
Meanwhile, Huawei is pushing ahead with developing more advanced chips for smartphones and AI chips, which is putting a strain on SMIC’s production capacity, said Paul Triolo, a partner at consulting firm Albright Stonebridge.
“The main bottleneck will be domestic foundry giant SMIC, which will have a complex task of dividing limited resources for advanced node production between Huawei, which currently holds the lion’s share, GPU startups, and many other Chinese design companies that have been or may be cut off from using global foundry giant TSMC to manufacture their advanced designs,” Triolo told CNBC.
Nvidia is not just about GPUs
Nvidia has thrived not only through its advanced semiconductors but also through its CUDA software platform, which allows developers to write applications that run on the U.S. chipmaker’s hardware, which has led to the development of a so-called ecosystem around Nvidia’s products that is difficult for other companies to replicate.
“That’s key — not just the hardware, but the whole ecosystem, the tools for developers, and the ability to continue to evolve this ecosystem as technology advances,” Triolo said.
Huawei takes the lead
Triolo noted that Huawei is one of the leaders in China for its Ascend series of data center processors.
The company’s current generation of chips is called the Ascend 910B, and the company is preparing to launch the Ascend 910C, which could be on par with Nvidia’s H100 products, he said. The Wall Street Journal reported in August.
In its annual report earlier this year, Nvidia explicitly identified Huawei as a competitor, along with other companies, in areas including chips, software for AI and networking products.
Triolo said Huawei “has a lot of advantages” when it comes to software and building a developer community, but the company faces the same challenges as the rest of the industry as it tries to compete with Nvidia.
“The GPU software support ecosystem is entrenched around Nvidia and, to a lesser extent, AMD, and given U.S. export restrictions that limit SMIC’s advanced semiconductor production capacity, Huawei faces significant challenges both in producing advanced GPUs like some of its Ascend 910C in sufficient quantities and in continuing to innovate and improve the performance of its hardware,” Triolo said.
Is a chip IPO on the horizon?
The challenges facing Nvidia’s Chinese competitors have become clear over the past two years, with Biren Technology implementing a series of job cuts in 2022, followed by Moore Threads the following year, both of which blamed U.S. sanctions.
But the startup remains hopeful and is seeking funding to reach its goals. Bloomberg Last week, it was reported that both Enflame and Biren were considering public offerings to raise capital.
“Viren and other GPU startups have experienced industry players from Nvidia, AMD and other large Western semiconductor companies, but they have the added challenge of not having the deep financial resources of Huawei,” Triolo said.
“As a result, both Biren and Enfram are exploring IPOs in Hong Kong to raise funds for additional employment and business expansion.”
