According to a report by Reuters and Bloomberg on July 14, Jeffrey Czesler, an official from the Department of Commerce's Bureau of Industry and Security who is responsible for U.S. export controls, confirmed that only 'a very small quantity' of NVIDIA H200 artificial intelligence chips will be shipped to Chinese customers, subject to approval by the U.S. government.
Kessler stated during a hearing on Foreign Affairs Committee of the U.S. House of Representatives on the 14th that the number of H200 and similar products shipped with export licenses is “very small, almost negligible” (“Very small quantity of chips, so it’s trivial.”).
Meanwhile, US Republican Congressman Bill Huçin attacked Kessler. On May 31, the US Department of Commerce issued a guideline regarding a so-called 'potential loophole' that could allow Chinese companies' overseas subsidiaries to obtain more advanced NVIDIA Blackwell chips than the H200.
Hui Cen questioned why the aforementioned guidelines stated that Chinese companies could retain the Blackwell chips obtained through loopholes.
Kessler replied that if a company does not obtain a chip license, it should disclose its violations on its own.
Huichen is relentless, demanding that it should be made clear to the Chinese side that Blackwell chips cannot be retained.
This is just a bunch of meaningless roundabout talk, I'm sorry, this is unacceptable," he said.
Reports suggest that the sale of NVIDIA's H200 chip to China has become a major point of controversy in the technology competition between the United States and China. In recent years, the U.S. government has continuously restricted China from accessing advanced semiconductor technologies, claiming that these chips could be used in military applications.
In December last year, Trump approved the export of H200 chips to China, marking a significant relaxation of the US restrictions on the export of AI chips to China. In January this year, the US Department of Commerce formalized this decision, allowing sales only to verified Chinese buyers and imposing a 25% US tax on such exports. Companies applying for this opportunity must confirm that the chips will not be used for military purposes in China.
In April, U.S. Commerce Secretary Wilbur Ross revealed during congressional questioning that although the Trump administration has tried to maintain a so-called 'subtle balance' in technology transfers with China, China has yet to purchase any H200 chips, as it wishes to focus its investments on the independent development of domestic industries.
China's self-sufficiency in AI chips is rapidly increasing. According to data from Morgan Stanley, China's self-sufficiency in AI chips has risen from about 10% in 2021 to 41% by 2026, an increase of more than four times over five years. The institution predicts that this proportion will further rise to about 86% by 2030.
American research firm Bernstein predicts that NVIDIA's share of the Chinese AI semiconductor market will drop sharply to 8% by 2026, while the total market share of Chinese companies will reach 80%.
The Chinese chip industry is proving with its actions that core technologies cannot be bought or obtained through external means. Independent innovation is the fundamental solution. Against the backdrop of increasingly strict U.S. export controls, China's semiconductor industry has not been hindered at all. Instead, it has forged ways to break through under pressure.
