Updated October 7 Semiconductor Export Controls
On October 7, 2022, the United States issued a sweeping set of export controls aimed at restraining Chinese military modernization efforts by controlling advanced AI chips made with U.S. inputs. A little over one year later, the Bureau of Industry and Security has issued an updated set of rules that revises the original ones in an attempt to close several loopholes that weakened the efficacy of the rules. These rules, which go into effect 30 days after publication, do not represent a major departure from the original intention of the export controls, but they do contain significant changes that will affect how they function and where they aim.
The Three Rules
The updates contain three main rules. The first, the Advanced Computing/Supercomputing Interim Final Rule (AC/S IFR), contains two primary elements. First, it adjusts the thresholds for which chips are covered under the new regulations. The administration is foregoing interconnect speed—the speed at which chips talk to each other—in exchange for performance density as a parameter. Exceeding performance parameters or the new performance density threshold would clearly fall inside of the line. However, the administration is also including a “gray zone” for chips that fall outside of this technical threshold but contravene the original intention of the rule. This is largely aimed at preventing the proliferation of Nvidia-style A800 and H800 chips, which the company engineered to comply with the rules, along with Intel’s Gaudi2 chips. While most chips for consumer items, such as vehicles or mobile phones, will be permitted under a “presumption of approval,” others may require prior notification and could potentially become subject to licensing restrictions. This will undoubtedly capture a far greater number of chips, although the number of impacted firms remains low.
The second part of the AC/S IFR is a significant update intended to combat circumvention efforts. A main pillar of this portion of the rule is to address problems related to China’s use of subsidiary firms to obtain otherwise controlled chips. Under the updated rules, exports of chips will be restricted to companies headquartered in China, Macau, and countries with whom the United States maintains an arms embargo. The effect of this update is that an additional 43 countries, spanning the D:1, D:4, and D:5 country designations, will now be subject to expanded licensing requirements. The D:1 category covers countries with national security risks, whereas D:4 covers missile technology, and D:5 is the list of arms-embargoed countries. The updated set of countries includes states with a high transshipment risk, such as Kyrgyzstan, which has shown an anomalous uptick in chips trade, despite having a relatively limited import and export of chips in past years. Overall, this update represents a significant geographic expansion but is intended to make it more difficult for potential Chinese military end users to obtain controlled technology.
The second rule relates to semiconductor manufacturing equipment (SME). This rule adds several dozen items to the list of controlled equipment, most of which are used for fabricating logic chips under a 16-nanometer threshold. Additions include equipment used in wet chemical processing, dry etching, and various types of deposition, among others. These are intended to achieve parity with recent Japanese and Dutch licensing restrictions as part of a trilateral arrangement achieved with the United States last January, although on certain deep ultraviolet (DUV) lithography equipment, the updated U.S. rules exceed the Dutch thresholds, potentially ushering in a fresh round of diplomatic discussions.
The new SME restrictions will also involve licensing restrictions to China, Macau, and countries with whom the United States maintains an arms embargo. However, the SME rule only applies to D:5 listed countries, representing a geographic expansion of 22 countries. The SME rule also significantly updates the restrictions on U.S. persons, clarifying covered end uses and facilities. However, the U.S. persons controls also include certain exceptions for individuals working in allied jurisdictions. Per the rules, the U.S. persons updates are “calibrated to address the national security concerns described above without unduly undermining the ability of U.S. persons to work for companies headquartered in the United States and closely allied countries.”
The third rule expands the number of firms on the Entity List, the U.S. equivalent of a blacklist of firms, to which the export of certain technology is prohibited. This new rule adds 13 entities to this list. These entities are involved in the development of artificial intelligence that threatens U.S. national security and foreign policy objectives. Specific firms listed include Beijing Biren Technology Development Co. Ltd., Light Cloud (Hangzhou) Technology Co. Ltd., Moore Thread Intelligent Technology (Chengdu) Co. Ltd., and Superburning Semiconductor (Nanjing) Co. Ltd.
Foundries—chip producers that supply other companies—outside of China and also outside of the United States that send chips to listed parties will also require a license, representing an extraterritorial expansion of these rules. However, both the SME and the chip rules will remain open for public comment for 60 days after they are published in the Federal Register. The administration will incorporate these changes in the coming months.
In terms of a likely response from Beijing, it is unlikely that Beijing will respond in kind to these updated restrictions. First, the United States has socialized—and even briefed—Chinese officials on the restrictions. In other words, these updates should not come as a surprise to the Chinese government, which has long acknowledged that an update to the October 2022 rules was pending. Deteriorating macroeconomic conditions in China also restrict its ability to engage in economic coercion that would risk deterring future investment in China. However, the last tranche of controls did result in China banning Micron products from its critical infrastructure and also in new licensing restrictions on Chinese exports of gallium and germanium, two inputs critical to global semiconductor supply chains. The gallium and germanium restrictions have ultimately proven to function mostly as a warning shot since the Chinese government intends to approve export licenses except in circumstances that would directly harm Chinese national security interests. China has also pursued other measures, such as restricting the ability of economic intelligence firms to operate in-country. Although a Chinese in-kind retaliation remains unlikely in the current context, it is almost certain that China will react in some way.
Because the updates reflect the original intention of the October 2022 rules, this update is unlikely to affect the forthcoming and highly anticipated meeting between Presidents Biden and Xi at the forthcoming Asia-Pacific Economic Cooperation summit in San Francisco in November. However, broader questions persist about whether these controls risk accelerating indigenization in China. Recent revelations relating to Huawei’s seven-nanometer capabilities—announced during Secretary of Commerce Gina Raimondo’s visit to China—denote an ongoing Chinese effort to indigenize advanced technology production capabilities.
China’s indigenization agenda far predates the Biden administration but has come into greater focus following the original October 7 export controls. The latest news captures one chapter in an otherwise long policy pursuit. If China can produce seven-nanometer chips efficiently and at scale—which still needs to be proven definitively—then the effectiveness of the U.S. controls on restraining seven-nanometer production will inevitably be reduced. However, if China cannot produce these chips at scale and efficiently, then the updated rules will continue to make it costlier and more time-intensive for Chinese entities to produce these chips.
Overall, the original intention of the rules—to safeguard U.S. national security interests through the control of advanced AI chips for military end uses—remains unchanged. The major expansion lies in the administration’s means of achieving this goal. Significantly growing the list of countries to which exports will now require a license, while simultaneously controlling additional items and enlarging the Entity List, point to a far taller fence around a yard that is expanding, however incrementally. This expansion will require ongoing allied buy-in and enhanced enforcement capabilities. The administration will also need to contend with how to mitigate workarounds related to cloud computing (a topic currently open for public comment) and whether the proliferation of advanced AI merits the use of other legal levers in the coming months.
The administration continues to affirm the importance of cooperating with allies and realizes fully that the success of these controls depends on allied buy-in and partnership. Moreover, the U.S. government foresees an ongoing effort to update the rules on an annual basis, signaling a realization that shifting targets alongside an evolving technology landscape will be key to the effectiveness of the rules. It also indicates that the administration remains receptive to feedback from industry, foreign partners, and other stakeholders. Perhaps the biggest outstanding question is whether these controls will ultimately succeed in bolstering U.S. and allies’ national security interests or whether additional and complementary tools will be needed to help the United States and its partners “run faster” against the competition.
Emily Benson directs the Project on Trade and Technology at the Center for Strategic and International Studies (CSIS) in Washington, D.C. Emily thanks Catharine Mouradian for her research assistance that enabled this commentary.