China’s strategy in AI race with US — big chip clusters, cheap energy
China is focusing on large language models in artificial intelligence.
Blackdovfx | Istock | Getty Images
It is well known that Chinese semiconductors designed for artificial intelligence cannot compete with the American firm Nvidia. But China has managed to continue developing highly advanced AI models, many of which are powered by chips it developed itself.
China’s secret? Lots of cheap energy and giant chip clusters from China’s tech champion Huawei, supporting the country’s AI advances in its race against the US
“China is striving for self-sufficiency in the AI stack because it views AI as a strategic technology for national and economic security,” Wendy Chang, senior analyst at the Mercator Institute for China Studies (MERICS), told CNBC.
The world’s second largest economy is deprived of certain technologies due to US restrictions and Beijing’s prefer to stay away Questions are swirling about Nvidia chips’ ability to compete in the AI space.
Despite these geopolitical challenges, domestic technology companies Alibaba’s DeepSeek has managed to develop and launch high-performance AI models; many were trained on home-developed chips.
Huawei vs Nvidia
Nvidia’s graphics processing units (GPUs) are considered the gold standard when it comes to semiconductors needed to train and run AI models and applications. But US export controls prevented Nvidia from shipping its most advanced chips to China.
Nvidia under an agreement with the White House this year It was given the green light to market its H20 product, a downgraded chip designed for China. However, Beijing has reportedly encouraged Chinese firms to stay away from Nvidia products and instead use chips designed for domestic companies.
Next comes Huawei, one of China’s most famous technology giants, which developed the Ascend series chips. But on a chip-per-chip basis, Huawei isn’t competing with Nvidia. Instead, Huawei’s advantage comes from its ability to bundle and connect many of these chips into high-performance “clusters” that can compete with Nvidia.
One of these products is the Huawei CloudMatrix 384, which connects 384 of the Ascend 910C chips to deliver performance that rivals the GB200 NVL72, one of Nvidia’s most advanced systems. Nvidia’s system uses 72 of its GPUs, while Huawei’s product uses five times as many of its own Ascend chips.
“This strategy relies on high-speed, potentially optical, interconnects to move data quickly between large clusters, a setup that does not require high-end chips and therefore suits China’s current strengths,” Brady Wang, deputy director of Counterpoint Research, told CNBC.
China’s energy advantage
The disadvantage of the Huawei system is that using more chips also means significantly higher power consumption. This is where China’s energy advantage over the USA comes into play.
“Solutions like CloudMatrix are less power efficient than Nvidia systems, but here China is taking advantage of an abundance of cheap energy,” said MERICS’ Chang.
“China has invested heavily in green energy, including solar, wind and more. It is also rapidly expanding its nuclear energy infrastructure. So it can rely on cheap energy when building its AI infrastructure.”
An overview of the new AI computing system CloudMatrix 384 system was unveiled at the Huawei Booth at Shanghai New Expo Center on the opening day of the World Artificial Intelligence Conference (WAIC) 2025 in Shanghai, China, on July 26, 2025.
YingTang | Nurfoto | Getty Images
Beijing and local governments have tried to support the effort. Many Chinese cities, from Shanghai to tech hub Shenzhen, have offered subsidies or “coupons” to lower costs to companies looking to rent computing power.
Finance Times reported this week that some local governments in China are offering subsidies that reduce the electricity bills of data centers using domestic chips.
“Less advanced process accelerators consume more power, but China offsets this with low rents and financing from a variety of energy sources (renewable options such as nuclear and solar), allowing it to finance and operate large-scale clusters despite chip-level inefficiencies,” Counterpoint Research’s Wang said. he said.
China vs USA: Will the gap widen?
The question is: As semiconductors for AI advance, can Huawei and SMIC keep pace with Nvidia and TSMC, given the Chinese firms’ restrictions on access to critical technologies?
“One of the key constraints to this strategy is China’s capacity to produce enough domestic chips to close the talent gap and keep pace as NVIDIA and others also continue to improve performance,” Hanna Dohmen, senior research analyst at Georgetown’s Center for Security and Emerging Technologies (CSET), told CNBC.
“China is working hard to improve its semiconductor manufacturing capabilities and capacity, but is still significantly behind due to export controls imposed by the United States and its allies on semiconductor manufacturing equipment.”
