Expansion of US export controls; from chips to AI

 

What had been negative speculations about the impact of U.S. export controls against China’s access to advanced semiconductor technology has turned out to be a reality. Despite the U.S.-led export restrictions, which blocked not only U.S. technology but those of allies that produce essential manufacturing equipment, China has been able to achieve a notable increase in its production of semiconductors in the first three months of this year. According to reports by some Chinese sources, China’s chip production surged by 40 percent to reach 98 billion in semiconductor units in the first quarter of 2024. In March alone, China produced a record high of 36 billion units, growing by 28.4 percent. While such growth has been partly due to strong demand for chips in its domestic downstream industries, such as electric vehicles and smartphones, more significant attribution has to be made to China’s expanding chip production capacity due to an increase in semiconductor plants built across various regions in the country.

Perhaps more shocking was the news last year that China’s telecommunications equipment company, Huawei, had made a breakthrough in its smartphone production in August 2023. China’s Semiconductor Manufacturing International Corporation (SMIC) had reportedly succeeded in mass production of Huawei’s Mate 60 Pro smartphone, developed with second-generation 7-nanometer process technology, using a custom chip architecture designed by Huawei’s subsidiary, HiSilicon. Based on such breakthroughs in its chip manufacturing technology, Huawei is reportedly expected to push forward its production of 5-nanometer semiconductors within the next two to three years. Apparently, barriers placed by the United States against China’s access to semiconductor hardware and software have pushed China to double up its efforts to move up the semiconductor technology value chain and be able to produce advanced chips for its own use.

The unintended consequence of U.S. export controls has been China’s doubled efforts to increase government funding and investment to obtain advanced semiconductor process technology to become more self-reliant and less dependent on foreign input sources for advanced chip manufacturing. Indeed, China enjoys the world’s largest consumer market for semiconductors, purchasing more than half of the world’s chip production. However, being highly dependent on foreign providers, such as Taiwan Semiconductor Manufacturing Company for advanced chips and ASML for process technology, China has been vulnerable to supply chain shocks, particularly to those imposed by the U.S., Japan, Taiwan and the Netherlands, which provide essential software and hardware for China’s chip production. China has been stuck in its position in the global semiconductor value chain, with concentrated dominance only in the lower-end chip market. However, it is now obvious that China has succeeded in developing advanced semiconductors for use in its production of telecommunication products, in addition to its increased capacity in terms of the overall production volume of semiconductors.

Despite such apparent failure in the efficacy of export controls, the U.S. Biden Administration has been recently considering new regulatory measures to restrict the export of advanced Artificial Intelligence (AI) models, in addition to the previously imposed export controls on semiconductors, including AI chips. According to the 2024 report on homeland threat assessment released by the U.S. government, advanced AI capabilities can be used to develop new tools that enable “larger-scale, faster, efficient, and more evasive cyberattacks.” There are also concerns that AI capabilities can lead to the potential development of biological weapons using the information provided by advanced AI models, in addition to reports that China is using AI advancements to prepare for military conflict with Taiwan. Such concerns have led to the legislative proposal to tighten AI export controls, targeting not only China but also other adversaries, including Russia, North Korea and Iran. This bipartisan bill aims to strengthen the authority of the Biden administration to impose export controls on AI models and also bar Americans from working with foreigners to develop AI systems that pose risks to U.S. national security.

However, criticisms regarding the efficacy of export restrictions on AI models remain since the threshold to be applied will be critical to the effectiveness of the measure. Another issue will be whether open-source AI models are subject to export controls, since China has reportedly been highly dependent on using open-source models for developing its AI model. However, such a broad scope of export controls could also potentially hurt U.S. businesses due to financial burdens, which will be disproportionately high for smaller companies and startups in the AI industry. Perhaps the spread of export controls into the AI sector as a major tool to be used by the U.S. government amid the intensifying U.S.-China technology rivalry may not be surprising at all in this current AI era. The consequential task remains for other countries that are home to major or potential players in the global AI market to assess the implications of the new legislative actions that the U.S. government is contemplating.

Lee Hyo-young is currently associate professor at the Korea National Diplomatic Academy (KNDA), teaching international trade & diplomacy. Before joining the KNDA in March 2017, she worked as a research fellow at the Korea Institute for International Economic Policy, during which she also worked as assistant secretary for trade, industry and energy at the Presidential Office (2014-2015).