Washington has intensified its efforts to put a speed limiter on China’s technological capabilities. In its latest attempt to impose a “technological blockade” on China, it has informed Nvidia and Advanced Micro Devices that they cannot export their newest data-center graphics processing units to China without obtaining a license.
The latest architecture GPUs, namely Nvidia’s A100 and H100 and AMD’s MI250, are a key technology for advanced computing tasks such as image and speech recognition. They are designed to speed up machine deep learning, a key step in developing high-end artificial intelligence technologies and supporting cloud service platforms.
In once again “stretching the concept of national security”, the Joe Biden administration is playing an old trick from 2015 when the US government blocked Intel from supplying high-end Xeon Phi processors to China’s supercomputer builders. An embargo that only added impetus to the development of the Chinese mainland’s processor industry.
The production of integrated circuits on the Chinese mainland grew by 33.3 percent in 2021 from 2020, while the whole year’s sales exceeded 1 trillion yuan ($144.96 billion) for the first time. The latest US move will only spur China to make greater efforts to develop its own chips and systems as it doesn’t want its development of technology to be at the mercy of vendors controlled by the US government.
In an article published in The Washington Post after the Intel ban, Michael Hayden, former head of the US National Security Agency, said that the US had previously restricted the sale of high-end chips capable of millions of theoretical operations per second (MTOPS) as it wanted an MTOPS advantage, but it quickly became clear that such actions “were undermining the global competitiveness of the US computer industry” and that “the overall health of that industry was more important than any MTOPS advantage against a specific target country”.
It seems that in its eagerness to turn the screws on China, the Biden administration has failed to heed that lesson. Apart from reducing the revenue of the two companies, the latest restrictions are expected to have an impact on the development of their next generation technology. The US Securities and Exchange Commission has warned that the new license requirement may impact the development of Nvidia’s Hopper accelerated computing platform, which reportedly delivers “an order of magnitude performance leap” over its predecessor.
It is simply make-believe for US politicians to expect China to give up the high-tech sector under their pressure. Intel was prohibited from exporting Xeon Phi to China as early as 2015, but Chinese scientists and researchers developed the Shenwei Taihu Light supercomputer that won four successive speed competitions from June 2016 to November 2017. It was not until they “learned” from the Chinese design that the US’ Summit became the world’s fastest computer in 2018.
There are gaps between China and the US in the high-tech sector, but they will shrink with China’s efforts and the US’ self-inflicted blowback on its own high-tech sector. China has already been working to create alternatives to various technologies for which the country has depended on the two companies. This move will only make it redouble its efforts to be self-reliant.
Editorial, China Daily