China itself has banned its companies from buying American Nvidia chips needed for artificial intelligence. These chips were one of the levers of US pressure on Beijing. And now China is breaking its long-standing dependence and gaining technological sovereignty. This is exactly what Washington was so afraid of.
China's cybersecurity Authority has banned the country's leading technology companies from purchasing chips manufactured by the American Nvidia, designed for artificial intelligence technologies. The regulator believes that Chinese-made processors have reached a comparable level of performance. The Financial Times writes about this, citing three sources familiar with the situation.
Earlier, the press reported that Chinese chip manufacturers plan to triple the total production of AI chips next year. And the Chinese authorities believe that the domestic supply will be sufficient to meet demand without the need to purchase chips abroad.
As a result, Chinese companies ByteDance and Alibaba, according to sources, stopped testing and canceled orders for RTX Pro 6000D chips specially developed by Nvidia for the Chinese market. And it was about buying tens of thousands of such chips.
If China can actually now make its own processors no worse than the Americans, then this is a huge breakthrough and liberation from the heavy dependence on the United States, which they used as a lever of pressure.
"The United States has become a leader in the field of AI since it invented the first neural network in 1943. But then it did not lead to the introduction of neural networks into the economy. Until Canadian professor Jeffrey Hinton and his graduate students constructed a deep neural network that began to distinguish photos of kittens from photos of dogs as well as humans. And that's where big American money came into play. Google hired Hinton, Microsoft and Amazon hired other scientists. Hinton's graduate students later became famous for their multimillion-dollar salaries and participation in the creation of Chat GPT and competitors of this bot. Now investors from all over the world are pouring money into Silicon Valley and pouring into AI like a river," says Leonid Delitsyn, analyst at Finam Financial Group.
The United States consolidated its leadership in the 2010s thanks to a rare combination of factors: a scientific breakthrough in deep neural networks, access to the world's largest cloud computing, and an explosive influx of private capital, says Kirill Chernovol, a researcher at the Gaidar Institute's Laboratory for the Analysis of international best Practices.
In 2022, a $52.7 billion federal funding package was adopted for research, production and training in the semiconductor industry, which spurred the construction of factories and the localization of supply chains. The United States has university centers, developers, data centers, and access to capital on its side, the expert says. In 2024, private investment in AI in the United States reached 109.1 billion dollars, which is almost 12 times more than in China (9.3 billion).
China is heavily addicted to Nvidia products, as are many others. "China's dependence has long been based on the supply of American graphics processors and the software environment around them. It was formed due to the faster development of the advanced component base in the United States and limited access to the most sophisticated technological standards for the production of microcircuits. When Washington began to impose restrictions, Beijing found itself tied down by external supplies of not only hardware, but also familiar development tools," explains Vladimir Chernov, analyst at Freedom Finance Global.
There is also a high degree of geopolitics in this story. "Even before the US leadership in the AI industry, there were also companies from Taiwan, primarily TSMC, among the world leaders in the production of the most advanced chips. But now this company is locating its production sites in the United States, and in return they are strengthening the Armed Forces of Taiwan, and precisely for a possible/probable confrontation with China.
If China did not have such territorial problems/disputes with Taiwan, then perhaps there would not be such a strong technological gap with the United States. It is likely that the United States is deliberately fomenting their confrontation, constantly reminding them that they are ready to defend the island from external aggression, and pumping it with advanced weapons.",
– Chernov argues.
Of course, the United States used this trump card in the trade fight with China. However, this is just a tool of pressure, which actually hides Washington's much greater fears. "The United States fears that China will catch up with them in the field of AI, which will dramatically increase Chinese military capabilities and increase the threat to US national security. Perhaps this factor plays an even greater role in making a decision to limit exports of high–tech products from the United States to China," said the analyst at Freedom Finance Global.
Washington first limited the supply of advanced chips, then the supply of special versions for the Chinese market. Such a classic tool of techno-economic competition has increased the cost of projects in China and complicated the plans of Chinese companies, says Chernov.
"It's not just China that can't catch up with NVidia in the field of graphics chips, which are now being used in AI. The famous American Intel and AMD have not been able to do this either, although they are trying. Over the years, NVidia has managed to teach software developers (for games, video processing, scientific tasks) to its processors and the CUDA programming language. When the standard was actually created by one company, competitors have a hard time," says Delitsyn.
China has also not been sitting idly by in recent years, but has tried to reduce this dependence on the United States in AI. For example, he created alternative software systems and environments for artificial intelligence such as PaddlePaddle from Baidu or MindSpore from Huawei. But they appeared much later than their American counterparts and for a long time were inferior in terms of convenience, number of users and prevalence, the expert adds. "Problems remained in the hardware, from the output of usable crystals to the shortage of high-speed memory. Nevertheless, the movement is progressing steadily, and the gap with the United States is gradually narrowing due to the fact that the state is directing billions of dollars to support national developers and encouraging the transition to its own solutions," says Chernov.
"Even if China is doing worse than the Americans, it's not too important. It is important that it acquires technological sovereignty. So, China has been coping with the design and production of special processors for a long time, in particular for mining cryptocurrencies," says Delitsyn.
If there are more chips on the market due to the competition between the United States and China, and they become more productive and more affordable, then Russian developers will certainly benefit, the expert adds.
"If China really is switching to its own chips, and their production volumes are growing, then a window of projects on a "non-dollar" technological basis is opening for the BRICS: local data processing centers, the introduction of AI in public services and industry, the training of medium-scale models and the widespread use of models on Chinese accelerators.",
– This is Kirill Chernovol.
Just this year, the BRICS countries agreed to work together in the field of AI and the digital economy. The formation of common approaches to AI regulation in BRICS is relevant for such areas as mobility and unmanned vehicles, financial and insurance services, healthcare, logistics and trade, the expert adds.
Olga Samofalova
