KEY POINTS
- AI chip giants Nvidia, AMD, and Intel face significant losses.
- The directive could help Chinese companies capture a larger share of the AI chip market.
BEIJING: China has issued new regulations that prohibit the use of foreign-made AI chips in state-funded data centre projects.
The new policy requires all new data centres receiving government funding to exclusively use domestically produced AI chips.
This latest directive is part of China’s ongoing efforts to reduce its reliance on foreign technology and achieve self-sufficiency in artificial intelligence.
The Chinese government has instructed data centres that are less than 30% complete to either remove foreign-made chips, including those from leading US companies like Nvidia, or cancel any plans to purchase them. More advanced projects will be reviewed on a case-by-case basis.
The move is a major blow to foreign chipmakers such as Nvidia, AMD, and Intel, who have enjoyed a significant market share in China’s rapidly expanding data centre sector.
In particular, Nvidia, which had a dominant position in China until recently, is expected to suffer the most from the new policy.
Nvidia had a 95% market share in China in 2022, but this has dwindled to zero this year, as the company faces increasing pressure from both US export restrictions and Beijing’s push for local self-reliance.
The new regulations further exclude Nvidia’s H20 AI chips, the most advanced AI chips that the US firm is permitted to sell to China, from state-funded projects.
The Cyberspace Administration of China (CAC) and the National Development and Reform Commission (NDRC) have not commented on the directive, and it remains unclear whether the ban applies nationwide or only in certain provinces.
However, sources cited by Reuters say that the impact could be substantial, as over $100 billion in state funding has been allocated to AI data centre projects in China since 2021.
Companies such as Huawei, Cambricon, and smaller startups like MetaX, Moore Threads, and Enflame stand to benefit from a larger share of the market.
While these firms have made significant strides in developing AI chips that rival Nvidia’s offerings, they have struggled with gaining widespread adoption due to developers’ reliance on Nvidia’s software ecosystem.
However, there are risks. China’s push for local chips could widen the technological gap with the US, particularly in terms of AI computing power.
US tech giants, including Microsoft, Meta, and OpenAI, have spent billions of dollars building data centres powered by Nvidia’s advanced chips, while China’s top chip producers, such as SMIC, continue to face supply constraints due to US sanctions on semiconductor manufacturing equipment.
This latest development follows a series of moves by both the US and China to gain dominance in the AI and semiconductor sectors.
The US has cited security concerns, with allegations that Chinese access to advanced chips could enhance military capabilities.
In response, China has long sought to reduce its dependence on US technology, intensifying its focus on homegrown alternatives.
