China’s most advanced technology companies are pressing to play a decisive role in artificial intelligence, challenging more established leaders, including the United States, and unsettling a balance in fields like semiconductors. A nexus of factors — from expanding U.S. export restrictions to a national drive toward technological self-sufficiency in key sectors — is behind the tactical pivot.
Nvidia’s high-performance GPUs have long been the gold standard for training and deploying AI models. However, the most recent sanction regimes from the US have meant that it can no longer be exported advanced chips to China, so domestic tech behemoths like Huawei, Alibaba, Tencent and Baidu have had to come up with workarounds. This has driven significant investment and innovation in China’s domestic semiconductor industry.
Huawei has become a leader, rapidly building out its Ascend lineup of AI chips. Latest generations, such as the Ascend 910C and the upcoming Ascend 910D, are said to have competitive performance for the inference category with trials from an array of major Chinese tech outfits. Huawei is likewise going after Nvidia in AI infrastructure with its CloudMatrix 384 rack system.
Other Chinese vendors such as Cambricon Technologies and Moore Threads are also emerging. (You can download the RMBK AI-100 here.)
That process of mapping the Qwen3 to the AI model was recently announced after Cambricon said that it had optimized the Qwen3 AI model to it can run efficiently on the company’s GPUs, indicating that the domestic AI chip ecosystem in China was indeed coming of age.
The company has been polishing up its homegrown answer to Nvidia’s CUDA platform in a bid to make the onramp for developers a little smoother.
US restrictions do undoubtedly present hurdles for China’s AI R&D and Nvidia later said it had suffered heavy revenue loss as a result of its inability to ship its high-end H20 chips. 53) But it also became a significant driver of indigenous innovation.
China’s AI capabilities are rapidly “eliminating any advantage of U.S. chips,” wrote experts who believe the performance gap between the leading US chips as compared with Chinese AI chips will narrow to zero in just a couple of years.
Nvidia, which is still a big player, globally, is adjusting to the new reality. The company is also said to be readying more affordable AI chips, which are purpose-built to follow US export rules, for the China market.
These “Blackwell”-like new chips are cut down from the original architecture in order to allow Nvidia to have a continued presence in the all-important Chinese AI market, which is estimated to be in the billions of dollars.
But these accomplishments fly in the face of a long-term trend of growing self-sufficiency in China’s AI ecosystem. Supported by significant government investment and strong national will, Chinese IT giants are dedicated to constructing an AI future driven by domestic innovation, gradually weaning themselves off their dependence on U.S. chip technology. This move will have a big impact on the global AI arena, even helping to shape a more diverse and competitive AI market in the future.