Semiconductor Export Control Trends Under Trump 2.0  — An Interview with Kevin Wolf

What would U.S. export control policies look like under a second Trump administration? Taiwan’s Research Institute for Democracy, Society, and Emerging Technology (DSET) sits down with Kevin Wolf, former Assistant Secretary of Commerce for Export Administration, to unpack the possibilities and their global implications.

Introduction

Kevin Wolf, a seasoned expert in export control compliance and a partner at Akin Gump Strauss Hauer & Feld, has spent over 30 years navigating the complexities of international trade law. During his tenure as Assistant Secretary of Commerce for Export Administration from 2010 to 2017, Wolf played a pivotal role in shaping U.S. dual-use export regulations and was a driving force behind the Obama Administration’s landmark Export Control Reform, which transformed defense trade policies with allied nations.

In an exclusive interview, DSET engages Wolf to explore how U.S. export control policies might evolve under a potential second Trump administration. Below are the full conversation:

A Long-Term Strategic Perspective on the U.S. Export Control Policy

DSET:  Could you walk us through the 16-year policy trajectory across three US presidencies? Are there any key differences in the export-control policy approaches taken by Democratic versus Republican administrations?

Wolf: US Presidents from Clinton, and earlier, to Biden have all implemented export controls to achieve national security objectives, but the idea of what is in the national security interests of the United States has evolved. 

During the Cold War, there was a broader strategic objective in the use of controls to contain the Soviet Union and the East Bloc. When the Soviet Union fell, there was a significant policy debate in the early 90s about what the role of export controls should be, and the US and its allies ultimately agreed on a relatively straightforward non-proliferation focus. 

This meant that, based upon the four multilateral regimes–one for missiles, one for nuclear, one for chemical/biological, and one for conventional military–the types of items (commodities, software, and technology) that were either bespoke for producing, developing, or using weapons of mass destruction or conventional weapons should be controlled.  In addition,the dual-use and commercial items that had some significant, identifiable relevance to the development, production, or use of WMDs or conventional weapons should be regulated. (The WMD and conventional weapons themselves, of course, were also controlled by the regimes.) 

This non-proliferation focus is reflected in the structure of the American administration’s export control system. At the State Department, the Bureau of International Security and Nonproliferation is the lead export control agency.  At Defense, it is the Defense Technology Security Administration.  At Energy, it’s the National Nuclear Security Administration.  The fifth original member of the US Government’s export control policy making structure was the Arms Control and Disarmament Agency, but it was disbanded in 1999. The role of Commerce’s Bureau of Industry and  Security (BIS) (formerly called the Bureau of Export Administration) was and remains  to shepherd this interagency export control system. That is, BIS’s role was and remains basically to consolidate the views of these non-proliferation objectives into the Export  Administration Regulations (EAR), to enforce the regulations. and to cooperate with various departments to work with the regimes to keep the lists of controlled items current.

The system that I inherited in 2010 has not changed much between Democrats and Republicans. Starting in 2016 and little earlier, but was not publicly discussed much until 2017 and 2018, were the changes in Chinese state policy and its fusion of military-civilian use of technology to acquire commercial technologies and modernize the Chinese military. The old (or “classical”)way of export control policy thinking focused on the nature of the item but not on state policies of specific countries or many human rights issues, particularly with respect to mass surveillance activities. This public discussion in 2017 and 2018 resulted in the Export Control Reform Act, with bipartisan support, to require the Commerce Department to think more broadly about the role of export control to identify and control emerging and foundational technologies directly in response to Chinese-specific efforts to use such technologies that did not have a clear, direct relationship to the development, production, or use of a weapon, but were nonetheless important, given the nature of the technology, to China’s broader efforts to advance its industrial base necessary to modernize its military. 

Trump 1.0 did not have a coherent vision of how to define contemporary national security issues and the specific emerging technologies that warranted new controls.  There were many questions asked about what “emerging” and “foundational” technologies should be controlled in addition to those traditionally controlled within the scope of the four regimes.  Many different Trump officials had many different opinions on the topic, but there was no one administration-wide answer to the question. It also took inconsistent positions on several matters, such as the revocation by tweet of sanctions against ZTE and the granting of licenses allowing for exports to Huawei. (Traditionally, exports to listed entities were simply prohibited.)  The Trump Administration, however, did significantly expand the extraterritorial reach of the EAR against Huawei in August 2020, which was a parallel company-specific concern regarding Huawei given its relationship with the Chinese government and ability to engage in acts contrary to national security interests  The Trump Administration also gets credit using the Entity List tool more directly and aggressively to list companies in China engaged in human rights violations, particularly with respect to mass surveillance and the Uyghur concentration camps.  

The Biden administration stayed quiet during its first year on what its export control policies would be.  That changed in 2022 with two major events. The first event was the allied response to the Russian invasion of Ukraine. There are now 38 countries that have come together to use export controls outside the classical multilateral regime system to achieve strategic objectives  far beyond classical non-proliferation objectives to slow the parts of Russia’s industrial base that are needed to support its continued war against Ukraine. . The second major event was a speech National Security Advisor Jake Sullivan gave in September 2022 where he defined our national security interest as including the need to maintain as large of a lead as possible against China in five primary “force-multiplying” emerging technologies, which are essentially those related to (i) advanced-node semiconductors, including memory, (ii) AI-related applications, (iii) the semiconductor production equipment needed to make such items, (iv) supercomputers, and, separately, (v) biotech.  (He also mentioned green energy technology, but that has not been a focus of export control policy thinking.)   This was the first coherent articulation by a senior government official regarding what a new vision of export controls should be to address China-specific national security concerns that were broader than the classical non-proliferation objectives that are the mandates of the four multilateral export control regimes. 

The Commerce Department implemented in October 2022 significant new amendments to the EAR that implemented NSA Sullivan’s vision.   Although the rules are extremely complicated, they are simple in their policy objectives, which are to cut off all the inputs, from the US and abroad, of the inputs needed for Chinese companies to have the indigenous capability to develop and produce in China (i) advanced node integrated circuits; (ii) semiconductor production equipment; (iii) the compute necessary for AI-related applications, particularly large language models, and (v) supercomputers. In other words, the US Government determined with these rules that China’s capability to produce these four technologies is a per se national security threat. After reviewing how those initial controls worked, the Commerce Department has updated the rules each year, including recently on December 2nd, with even more complex amendments, but always with the same four policy objectives.  Whether one agrees with it or not, at least, in my view, the Biden administration articulated a coherent, administration-wide policy vision for how export controls should be used beyond the classical non-proliferation objectives. 

This is the policy vision that the Biden team will leave to the Trump team, which will no doubt expand upon it. The general view is that tariffs will be used to give the Trump team leverage to motivate more domestic manufacturing. With respect to what an export control policy vision will be, I do not really know.  President Trump, individually, has never really mentioned export controls and the policy objectives for export controls. It was not an element of the campaign.  The ultimate vision might be more hawkish because Senator Rubio and Mike Waltz will likely become the Secretary of State and National Security Advisor.  They have each made statements in the past regarding export control policy and China-specific national security issues. I am unaware of any positions or statements on either issue, however, by Howard Lutnick, the current Department of Commerce pick. 

Moreover, it is possible that the Trump team will be more hostile toward allies, based on positions regarding the allies taken during the first Trump administration.  To prepare the slide deck that I sent you, I read all testimonies and speeches from people who might go into a Trump administration. One idea about retaliation against allies was in a Republican-led appropriation bill from a few months ago that said any allied country company that (legally) exports to China items that a US company could not be added to the Unverified List, which is a lighter version of the Entity List I have also heard Republicans in conferences say that the Biden administration was too nice to allies regarding imposing controls against China. Trump, as an individual, can be antagonistic to long-standing arrangements and allies, such as NATO and Taiwan. Also, Trump is widely reported to take a “transactional” approach toward policy. This means that he will negotiate on two or more unrelated topics, whereas the Biden team and the traditional diplomats will look for common values, interests, and principles. 

Eventually, any export control decision-making will be a function of consensus among the four departments, as led by the White House and the National Security Council.  So, we really will not know what the Trump administration’s export control policy vision will be until after the administration begins and we learn who the people will be confirmed for the various Assistant and Under Secretary positions in the export control and related agencies.   In particular, I have no idea what a Trump administration’s view regarding export controls should be to address non-China specific development of AI-related capabilities outside the United States. The Biden administration is reportedly working on a rule to impose worldwide controls (minus a few close allies) over the inputs for advanced AI capabilities. I am assuming some portion of that vision will be published before January 20th.  If so, it will be interesting to see how much of it survives during the Trump administration.

U.S. Unilateral Export Controls and Their Considerations

DSET : With growing U.S. concerns over China’s AI chip development, recent restrictions target Korean and Japanese firms selling high-bandwidth memory (HBM) products and request TSMC to halt AI chip supplies to Chinese clients. Are these measures necessary, and should controls extend to legacy chips? Are there gaps in the current semiconductor export controls that need addressing before Biden’s term ends? Does the U.S. primarily engage with governments or industries on unilateral export controls? If Trump returns, how might this approach to ally coordination change?

Wolf: The reasons for unilateral controls are twofold. One is the need to act quickly because the issue is particularly serious or to prevent stockpiling. But the second reason is that most allies don’t agree with the broader national security objective regarding China. They largely feel that export controls should be limited to the more classical views – i.e., to control the hardware, software, and technology more directly related to the development, production, or use of conventional weapons or weapons of mass destruction. And, particularly with the European and, to some extent, Japanese and Korean companies, it is very hard to economically give up in each of these key sectors, and Japan and Korea are also anxious about Chinese retaliation on critical minerals and other items. 

The Commerce Department and the State Department have been working to get new controls by the Japanese, Dutch, and Korean, and we won’t see that until sometime next year because I suspect that the US will move first and use the leverage it has, given the fact that all HBM and GPUs, and all semiconductors that are made in part with equipment that was made from US technologies. They are using the choke points of the US companies, such as the US semiconductor production equipment and EDA software companies to create the jurisdictional hooks to assert extra-territorial controls. 

In addition, all GPUs are produced with US equipment or US technology. Advanced GPUs are critical choke points for a data center to run an LLM. And so, the national security objective of the US is to deprive China of the computing power for data centers, cutting the flow of GPUs and, as of December 2^nd,  HBM directly to China and through countries that are a diversion risk. In November 2023, export controls of GPUs to the Middle East and a few other countries were imposed. Any export of a GPU to any company that was headquartered in China or owned by a company that was headquartered in China required a license, and that was the way in which to cut off Chinese companies from setting up data centers outside the US to run models in order to create the algorithms for AI applications.

So, they don’t really need too many unilateral or multilateral controls for the GPUs or HBM because there are only a very small number of companies that make those. And they all use US technology, and they all use US equipment. With respect to SME, here are only two major non-US companies, Tokyo Electron, and ASML, so it’s been the Biden administration’s hope that Japan and the Netherlands would agree to a common security threat and impose their own controls, and we won’t exactly know how successful or not they were until probably deep into 2025. My suspicion is that neither the Dutch nor the Japanese were particularly convinced about the arguments of the Biden administration, and what that would do was create a structural advantage for the Japanese and the Dutch companies over the US companies.

Unilateral controls are all government-to-government discussions, and it’s somewhat hard for industry to find out exactly what the governments are discussing. The difference with the Trump administration, I suspect, will be far less considerate of allied considerations. They might, for example, be willing to sanction an allied country company that continues to sell semiconductors or semiconductor production equipment legally into China or to Chinese companies of concern. I could easily see a Trump administration taking a more transactional approach to trying to get the allies to impose their own controls, to increase the effectiveness of the controls by threatening to remove troops or threatening to impose sanctions, or, you know, removing spare parts that are needed for the F-35 to function in those countries.

DSET: Do you know if industry comments have in the past played a big role in shaping how the Bidens administration actually implemented their policies?

Wolf: Very limited. These rules were published in final form, and industry input was largely ignored. There have been other examples where public comments did make a difference, and one example was with the quantum computing controls that were imposed a few months ago by the Commerce Department. They didn’t want to harm the development of quantum computers in the US, and they were sympathetic to industry concerns. The primary call from the US industry has been: do whatever you want on controls. The sentiment is that we’re not denying what you’re saying about the threat with respect to China, but we need to fund our research and development to stay ahead of our foreign competition; please work harder trying to get the allies to agree to the same controls to make the controls more effective and less counter-productive for US industry.

DSET: Will Trump be more susceptive of industry concerns?

Wolf: Trump, as an individual, took a more “transactional” approach toward policy, including with respect to export control issues.  A classic example of that is ZTE Apparently without coordinating with the export control agencies, President Trump, according to his own tweet, decided that controls on ZTE should be removed; after a phone call with Xi, He was trying to get a deal on more investment from China, so he compromised on the export control issue, it seems, based on his tweets.

We will have to see who gets nominated for his career and political staff, but it’s highly unlikely that they will agree to any sort of reduction of controls. It is possible that they’ll be more sympathetic to concerns that the controls are not being imposed equally against their non-US competitors, and I could easily see that US companies lobbying being more successful at convincing the Trump administration to impose sanctions against their foreign competitors until the allies agree to abide by the same controls the US companies are subject to.

DSET: Given the current goal of expanding security and defense cooperation with allies, how impactful would Trump’s transactional approach be?

Wolf: Under the Biden Administration, it is all government-to-government, and I seriously doubt that unrelated topics would ever come up in the export control discussions. And, that’s a principle in and of itself, agnostic of any agreement or disagreement on any other area. The Trump Administration has been pretty hostile toward NATO and other allies and more friendly toward Russia. Trump himself has said several times that Taiwan has stolen all of our IP for making semiconductors and is asking for protection money. I doubt that Senator Rubio or Mike Waltz would have those same views, but I suspect that it is possible that the Trump Administration will say that, unless you agree to controls or defense spending, we will sanction your companies or stop selling arms. We will have to wait and see who is in charge when they come in. We really don’t know. 

DSET: Would export controls expand to legacy chips?

Wolf: Probably, but we will need to see who is in charge. Several people within the Trump Administration have been overt in saying that controls should be imposed on the tools needed to make legacy- node semiconductors. A Trump Administration view could be that, given the oversupply by YMTC, CXMT, or SMIC into the ecosystem of legacy chips, controls should be imposed on tools, and tariffs should be imposed not only on chips coming into the US but also on equipment and electronics that contain them. And that  a combination of export controls and tariffs should be used to achieve a broader supply chain concern. 

One person you might want to interview is Attorney Nazak Nikakhtar. She is very vocal about these arguments and was an acting Under Secretary of the Trump Administration.  Her views and speculation would be far more relevant than mine. 

DSET: How useful would controls on legacy chips be? Some commentators have argued that China has already had the ability to make equipment to make legacy chips.

Wolf: I agree with Chris Miller that it would be ineffective because Chinese companies would already be able to provide support for legacy-node production. It would also be inflationary and disruptive. But, if you talk to Nazak, you can see her views on whether effectiveness is a relevant consideration. There is another school of thought that sees this as a matter of principle, and we should not authorize the export of equipment that would be used to make semiconductors that would swamp the US ecosystem – regardless of whether the controls would be effective.

Global Multilateral Controls and Taiwan

DSET: Under U.S. extraterritorial controls, foreign firms may create supply chains independent of American technology to serve China, highlighting the need for multilateral export control regimes. Have tech democracies made progress toward this, and what are the main obstacles? How might a Trump administration shape these efforts? The CHIPS Act, with TSMC receiving $6.6 billion in grants and $5 billion in loans, strengthens U.S. industrial ties but encourages reshoring, countering globalized supply chains. How could this impact the effectiveness of U.S. unilateral controls and efforts to establish multilateral regimes?

Wolf: It is quite normal for US companies to find ways to produce items outside of the US that don’t rely on US technology or equipment in order to ensure compliance with the regulations and to avoid even the appearance of violating the regulations. It is easier in some sectors than others, however. In the semiconductor equipment sector, they cannot completely make foreign-made items without US technology. However, for other electronics and equipment, companies can de-Americanize because of economic incentives and the wide availability of comparable foreign-origin technology.  In “de-Americanizing,” however, US companies need to be careful not to engage in prohibited evasions of US controls. 

I agree on the importance of multilateral regimes, but what has happened in the last couple of years is that we have fallen into four ad-hoc regimes. First is AUKUS, the defense trade and dual-use arrangement. Second is the Dutch, Japanese, and the US tool controls, and maybe Korean, in the ad hoc plurilateral arrangement. Third is the 38 countries’ cooperation on controls against Russia. Fourth is the Wassenaar-minus-one control, which is the leftover of the Wassenaar countries realizing that they cannot reach a consensus because Russia is blocking any progress. That is how the quantum computer controls were imposed because a fair number of Wassenaar countries informally agreed to impose controls together but not under the umbrella of the Wassenaar arrangement.

I cannot imagine that the multilateral regimes succeed under the Trump Administration since he will be quite hostile toward the allies. With the threat of withdrawing from NATO, imposing tariffs against allied countries, and withdrawing support for Ukraine, I cannot imagine any allied country willing to cooperate with the Trump Administration. It is likely that the next many years will be somewhat chaotic with respect to which unilateral or multilateral controls are used in the foreign direct product rule for extraterritorial controls.  But I could be wrong. We’ll see. 

DSET: Have the recently announced export control measures effectively addressed existing loopholes? What is your vision for an ideal multilateral control regime?

Wolf:  I don’t like how some people see that as blocking “loopholes.” Sometimes, people refer to policy objectives they would like but that the government deliberately didn’t take as a “loophole.”  But, yes, sometimes the government misses things in its controls.  So, I see it more as the government’s fine-tuning its controls based on having studied how the previous year’s controls worked and after learning more about the technology ecosystem. Remember, the export control agencies were built and staffed to address non-proliferation objectives. Although there are very smart people in government, there are few who understand deeply the technology and the supply chains behind the development and production of advanced node semiconductors, AI applications, semiconductor production equipment, and supercomputers, which now include quantum computers.  

Since we last spoke, Commerce published on December 2nd an additional update. This rule, like the others, is extraordinarily and unusually complicated. Even for export control experts, they are hard to understand and ensure compliance.  The complexity is a function of several things. First, clearly, the rules reflect informal understandings about what would be acceptable to close allies Japan and the Netherlands. Second, the technologies involved are unusually complicated relative to many of the other types of items the EAR regulates.  Third, the Biden administration has tried not to create rules that result in a broad “decoupling” with the Chinese economy.  Fourth, for the rules to be more effective, they are extraterritorial in novel ways. That is, the regulations impose controls over foreign-made items outside the United States that do not contain any US-origin content or US person involvement if the items are either produced directly for US technology or produced with equipment that was produced from US technology.  These are novel and complex jurisdictional hooks over foreign-made items produced in countries that do not control the same items in their systems and that are not clearly directly related to the production of weapons.  

But, again, although the new rules are complicated, they all have a very simple objective, though the Biden team has not explicitly described it this way, which is to cut off all the inputs, directly or indirectly, US or foreign, for the indigenous development and production in China of (i) advanced node semiconductors, (ii) the compute side of AI-related applications; (iii) semiconductor production equipment, and (iv) supercomputers. The first objective focuses on the production of logic, NAND, and DRAM in China or by Chinese companies. (Although I think this may expand to include controls on the export of logic for data centers before January 20th.)  The AI-related objectives first focused on GPUs (almost all produced in Taiwan) needed to run large language models. The December 2nd rule added controls on High-Bandwidth Memory (HBM), which are needed to work with the GPUs.  The HBM controls are a chokepoint technology because there are only three companies that produce HBM, and none of them are in China.  (Two are in South Korea and one is in the United States.)  

The biggest change in the December 2^nd rules is that they added about 140 Chinese companies to the Entity List, which includes companies that make semiconductor manufacturing equipment (e.g., Naura,) and EDA companies (Empyrean) that make the software used to design ICs. This completes the four-part policy objective I mentioned with respect to the policy for adding entities to the Entity List.  In previous rules, Commerce added to the list the companies involved in advanced node integrated circuits, GPU and AI-related development, and supercomputer development. This rule adds to the list the companies in China that produce semiconductor production equipment. 

Some critics saw the new rules as not very effective since their goal was not to cut off all inputs for making any semiconductors in China. The Biden team, on the other hand, is careful to do two things that the Trump team might not be as careful about.  (But, again, we really do not know what the Trump administration’s export control policy will be.)  

First, the Biden Administration has not wanted to affect the production of legacy node semiconductors so as not to create COVID era-like supply chain shocks to the global system. Thus,  apparently for this reason, it did not list SMOC, SMTC, and other fabs that produce only legacy node chips. Second, the Biden Administration wants to be respectful of the allies. There were no extraterritorial controls imposed against exports of these tools from allied countries’ companies or most of the A:5 countries* (other than Korea and India). BIS excluded, for example, exports to China from Japan, Germany, and the Netherlands from many of the controls, but they didn’t exclude Korea or India.  

This  suggests to me that Japan and the Netherlands will be imposing their own controls at some point, but not Korea. That, I speculate, is why Japan and the Netherlands got special treatment. In terms of effectiveness, it will depend on how far the Japanese and the Dutch are willing to go. If they do not impose similar controls, the new rules will not be very effective over the long term.

Eventually, the core theme that I have been arguing is that multilateral controls are more effective. This doesn’t mean that unilateral controls are illegal or should never be used.  I’m only saying that a basic rule of all technology development is that, over time, multilateral (or plurilateral) controls are always more effective. How much time that is depends upon the technology at issue.   Some types of items can immediately be produced by companies in allied countries or China that are subject to controls.  Others will take many years or decades to create substitutes for what is no longer able to be controlled. The issue is not simple.

Historically, China has not responded much directly in retaliation. This time, their response was to cut off the supply of critical minerals. However, there are two more important things. First, such threats of retaliation will have a bigger impact on countries like Taiwan, Japan, and Korea, who are all much more exposed to retaliation than the United States. There might be forces within these countries that may be reluctant to align with the US for fear of critical mineral dependence unless the US can arrange for better supply chain security. Second, no matter the leadership style, the US Government’s objective is for US and allied companies to decrease their dependency on China – as a market and as a source of raw material or other inputs.

The Biden team has been pretty aggressive with US companies to achieve this broader national security objective – i.e., that  such dependencies on China will not only eventually harm themselves but also the US industrial base. For example, in the new rules, there are temporary general licenses, extended by one year for production, development, packaging, repair, and other activities using Chinese companies to make components for semiconductor manufacturing equipment for end uses outside of China. The Biden team aims to give US companies one more year to wean from dependency on China. When the Trump team starts, this will further justify them to find alternative suppliers. But, if China keeps imposing its own controls in retaliation to US controls, it will do more to accelerate decoupling than the Commerce Department and US export controls ever will.

DSET: What is your view on the Taiwanese government, given the recent incident of TSMC and Huawei and your exchanges with the MOE and others?

Wolf: Taiwan’s position is hard. Given its close proximity with China, sanctioning Huawei might be very difficult. And it’s better just to rely upon US extraterritorial controls where it gets the benefit of the same controls against Huawei, but without having to actually impose them domestically.

DSET: In your testimony on Capitol Hill, you emphasized the importance of establishing the multilateral export control regime. We want to hear more about your vision of the regime, particularly the role of US allies. Secondly, you mentioned that Taiwan might be better off relying on the US extraterritorial controls. However, we have seen that there were Taiwanese companies helping Huawei in building its chipmaking capabilities. Given this incident, how do you assess the necessity for Taiwan to strengthen its export control regulations, and should the US leverage diplomatic tools to achieve the multilateral export control goal effectively?

Wolf:  I have been an advocate for a regime of a smaller group of allies to address both (i) traditional proliferation-related issues  that cannot be addressed by the legacy regimes because of Russia’s membership in the regimes and (ii) the non-traditional common security and human rights issues that are not within the mandates of the legacy four regimes. In particular, the four legacy regimes are country-agnostic and are not designed to address non-traditional national security concerns, particularly those specific to China and Russia. Also, to address non-traditional national security issues in emerging technologies, the focus of a new regime cannot be only those types of items that have a direct relationship to weapons.  In addition, a new regime needs to focus on common human rights issues, particularly with respect to mass surveillance. 

We need a new regime of a smaller group of allies that are producers of those technologies – and that are also willing to impose end-use and end-user controls, as item-based controls will not be effective. That is, there are three types of export controls – those based on an item’s technical parameters (“item-based” controls), those based on how unlisted items could be used (“end-use” controls), and those based on specific entities, regardless of the item and its end use (“end-user” controls).  All three need to work together for an effective system.  Now, however, the legal authority of the allies to impose end-use and end-user controls is limited to situations involving the development or production of weapons of mass destruction.  In my view, all allies should have significantly broader legal authorities to impose controls on (i) items that are not identified in the multilateral regime lists; (ii) end uses, even if not related to WMD; and (iii) end users that are supporting activities contrary to broader common security interests, particularly in China and Russia. 

One reason I think the allies have resisted changing their laws to give themselves such authorities is that they don’t want to create the perception of ganging up on China. They want to stick to the Wassenaar arrangement and maintain the same image from the past of regulating technology of concern. We can do the informally named “Wassenaar-Minus-One” approach because of Russia, but they do not want to take actions that are specific to China.  In proposing my new plurilateral regime ideas, I undervalued the anxieties of the allies on this issue. This is why they have preferred the the cover of the Wassenaar process, even if it is far less effective

This leads to the second reason, which is the disagreement within the agencies on whether the approach is. Some in the US and allied country governments believe that my ideas would never work because the allies were not going to accept the idea of a formal new regime. My approach would have been to find a way to address the allies’ concerns. There is also the issue of manpower. The export control and related agencies – in the US and in the allied countries – are already thinly staffed, and they have to deal with regular things plus the time-consuming Russia-specific controls. The fourth related issue is that other than me and a few think tank commentators, there was not really any coherent vision of the idea of a new regime being articulated.

What has evolved in the last few years is four ad hoc plurilateral regimes, which is somewhat chaotic. First, there is the informally named “Wassenaar Minus One,” group which is the core group of Wassenaar members that have agreed to unilaterally impose controls over what would have normally been agreed to at Wassenaar in previous years Second, there is the AUKUS arrangement, which is more straightforward. Third, there is a group of 38 countries, including Taiwan, imposing controls against Russia. Lastly, there are the Japanese, Dutch, and US control over semiconductor production equipment that are not controlled by the regimes We are going to be limping along these four ad hoc regimes for slightly different objectives unless, though unlikely, some allies can get behind the Trump Administration to create a new regime founded on a common, coherent vision of common, contemporary national security issues to the allies.

The issue with Taiwan is that there are always some countries’ laws that would not permit them to participate in an organization where Taiwan is a member. This goes back to the earlier reason for China’s retaliation and the fear of provoking China. To answer your question, in my personal view, Taiwan must absolutely be included because this is where advanced node semiconductors are produced. TSMC, MediaTek, and all the core technology companies and experts in Taiwan should have a seat at this very important table given that so many of the emerging technology items at issue in the discussion are produced in Taiwan. 

There are ways to reflect Taiwan’s interests without violating the allies’ limitations involving  Taiwan.  One would be sort of the IPEF model, such as having different meetings with multiple countries and bilateral meetings with Taiwan. This could work to address both concerns of Taiwan’s inputs and being respectful of the legal and diplomatic impediments of Japan, Korea, and the other allies with similar limitations.

DSET: We also would like to hear about your experience with Taiwanese officials and your key takeaway from the discussion with them.

Wolf:  First, you will need to talk to the Taiwan government officials to get an answer to this question. They would generally ask me more questions than tell me about what they were thinking.  They don’t discuss their own policy views with me much.  My sense is that there is not a grand vision for new export controls, and that they were worried about retaliation. Without a coherent vision, compliance and effectiveness are more difficult to achieve, though I understand that not only Taiwan but any allied country will be concerned about political and economic reasons. Sometimes, allied countries would even be grateful for US extraterritorial jurisdiction since they can have the controls imposed against China but not directly from them.

However, this only goes so far as the US can assert control over non-US-produced items that are made with US technologies or equipment. In the Russian case, it was critical for Taiwan to impose controls on the export of items for the Russian industrial base in the domestic law. Companies must acknowledge that violating these rules will draw significant enforcement consequences brought by that country’s government. Besides, local enforcement is also much more doable since the US has limited manpower.

My personal goal is to inspire allied countries to enforce their own controls, similar to the group of countries that came together to impose sanctions against Russia. They should gradually move toward more plurilateral regimes or arrangements, similar to the Wassenaar-minus-one, ensuring that they have adequate controls. Taiwan is not a member of any of the regimes, as in the case of Singapore or Israel, which are allies with the US but not members. Taiwan is referred to as Wassenaar adherent, which is adopting the EU common list and implementing it in its own system. So, the indirect benefit would be that if the EU adopts all the non-Wassenaar controls into the EU list, those would automatically flow into the control systems of countries like Taiwan, Singapore, Israel, and other Wassenaar adherence.

Regarding Huawei, it is particularly frustrating and not unique to Taiwan at all. No country other than the US has the legal authority to impose end-user controls, which are controls over unlisted items specific to particular companies. There are sanctions against companies or sectors, but no legal authorities anywhere close to the US entity list. During my tour to Taiwan, Japan, and Korea, I was advocating for allied countries to come together and, at a minimum, agree to give their country’s export control agencies the legal authority to impose end-user controls or at least create the legal authorities for it. This is part of the basic principles I have about export controls: the old way of targeting items of particular technical characteristics is not effective for certain types of controls, particularly human rights ones. It is necessary to have end-user controls prohibiting certain activities and companies engaging in bad acts, and the US should leverage its diplomatic tools to achieve the goal. 

I hope that the Trump Administration will agree that it is in the US and allies’ common interests to have the same authorities, end-user controls, and standards. However, we will have to see who is in charge, and it is possible that the Trump team would prefer to stick to carrots, punishing allied countries’ companies if they export items to China while their US competitors cannot. I could see that a Trump team would say, “If you keep doing that, we’re going to add your company to the entity list or your product to the control list. We’re going to impose this tariff on your products. We’re going to stop cooperating with this military training exercise in your country,” instead of trying to convince through appeals.

DSET: DSET’s working project on comparing entity lists of the US and Taiwan finds that it is problematic for Taiwan since it arbitrarily follows the UN and the EU lists without clear transparency. We are working on visualizing it and are interested in hearing your thoughts on comparing them.

Wolf: The UN list (UN Resolution 1540) is specific to Non-Proliferation issues, and it is true that different countries have private lists of different licensing policies with respect to controlled items. But that’s different from imposing actual controls on an actual exporter. What I am advocating is more aggressive and transparent, which serves as an actual legal prohibition.

Regarding the last question, when I meet with government officials, I try to have them tell me what they’re thinking. Everybody in the US loves to talk about Taiwan, but I don’t hear anybody actually saying, “What do Taiwan people want? What does the Taiwan government want?” My view is that Taiwan, Japan, Korea, and other countries have more common interests with the US than with China. There is a core commitment to core democratic values and a commitment to human rights among these countries, which obviously doesn’t exist in China.

Regarding Taiwan’s limited international participation, I have mentioned that there are workarounds that can reflect Taiwan’s equity while being respectful of the legal and diplomatic concerns of other partner countries. One point I have been very clear about in my ideas about multilateralism is that, unlike Wassenaar, where you’re not consulted as part of the development of the controls, Taiwan should absolutely be reflected in the type of items and end-user controls, and perhaps it should be done so in a separate bilateral structure or a small group of nations. I suspect this kind of talk will be held even with the Trump team. Trump, as an individual, has made some outrageous speeches, treating the US-Taiwan relations as a protection racket. However, Senator Rubio, who might be Secretary of State, Trump’s national security advisor, and pretty much the rest of the Republican national security establishment, will be quite committed to the security of Taiwan, consistent with the broader strategic ambiguity policies that have been consistent throughout administrations. But I suspect they’ll be quite forward-leaning at needing to ensure through military sales and diplomatic arrangements. But that’ll remain to be seen.

I cannot presume to tell Taiwan what is in Taiwan’s interests, so I would repeat the things that I said above. It would, however, be in the common interest of all allies, which should include Taiwan, to have several changes in the export control regime: greater enforcement, new authorities for end-user controls, an articulation of standards as to non-classical items, a principle regarding human rights issues, and more specific focuses on emerging technologies and countries of China and Russia. This should be equally applied to the democratic allies and producers of technologies, including Taiwan, Korea, Japan, Australia, the Netherlands, and others. It might not matter which nation leads the new initiative, and as China will be decoupling, these nations might need to find alternatives for the impacts of retaliation among themselves.

We need to come up with better arrangements for sharing and allocating critical minerals and joint venture opportunities and reducing the burdens on trade between those countries. That is where Trump’s obsession with tariffs is going to get in the way because one way of enhancing collaboration is by removing barriers to trade. The US can remove tariffs among allies in exchange for their cooperation with China, and in the meantime, it can increase access to US government procurement and expand the number of visas for scientists and others. Additionally, the US can reduce the regulatory burdens for direct investment in the US and find alternative markets and synergies as a payoff for agreeing to the same controls against China, as well as the benefit of increased collaboration. Eventually, from a more Taiwanese perspective, all these approaches can be done, but they have to depend upon the interests of the Taiwanese.

DSET: Do you think that Taiwan and other allies should be prepared for Trump’s hostility and punishment against them?

Wolf: I listed it as one possibility. To prepare the slides I sent, I went through all the testimonies and speeches and reflected on things I’ve heard at conferences by people who might go into a Trump administration or Republicans that might influence a Trump administration, including the Project 2025 report. That idea about retaliation against allies was in a Republican-led appropriation bill from a few months ago that said any allied country company that exports US-prohibited items to China should be added to an unverified list, which is similar to an entity list.

That idea made its way into an appropriation bill from the Republican-led House Foreign Affairs Committee. So, if that idea was there once, it could be there again. I have also heard Republicans in conferences say that the Biden administration tried to be too nice to allies on imposing controls against China, and the Republicans are not going to be nice. So, that is why I put it as a possibility. Eventually, we will have to see who is in charge. 

Trump, as an individual, can be offensive to long-standing arrangements and allies, such as NATO and Taiwan, but the person in charge might come in and say it differently. Also, Trump is transactional, meaning that he will negotiate two unrelated topics, whereas the Biden team and the traditional diplomats will look for common values, interests, and principles. However, other countries can use that as part of a deal to get to Trump, with some kind of flattery. This was very evident during the first term, and we might get used to it in the second term, as country leaders like Justin Trudeau and French President Macron invited him to grand national events.

How Effective Are Export Control Measures?

DSET: Other than decoupling, many commentators recently have suspected a bifurcation might happen in the Trump second term, where companies have expressed more interest in setting up subsidiaries and separate entities to maintain separate global and Chinese markets. Between bifurcation and decoupling, which outlook do you think is more likely to happen?

Wolf: Yeah, it clearly will happen. As you put it, bifurcation is opposed to decoupling. De-Americanization has been happening for the last several years. One thing I noticed as a lawyer is that when companies ask for legal advice, they say to me and every other export control attorney in town, what do I need to do to set up my operations so that I’m never subject to US export controls and can keep making money selling to China so that I don’t inadvertently violate current or future US export controls? It is pretty simple: if you don’t have US technology, US persons, equipment made from US technology, or anything of US origin in your products, American law cannot touch you.   (This is why plurilateral controls are so important to enhance their effectiveness and reduce counter-product impacts.). Instead, companies are more into using technologies, buying equipment, and hiring engineers from outside of the US so that they can continue to sell products to China and make sure that they do not inadvertently violate US export controls Companies might even think that they are doing the right thing by creating a separate market and bifurcation, but the US government needs to call out that this is completely missing the point. US government officials have to be clear that the goal is not to sell advanced technologies to China and not to help them make advanced-node semiconductors and support the indigenous capabilities of China. I have not heard of a US official who gets frustrated with an ally for bifurcation, but the US needs to repeat that bifurcation misses the policy objective completely. So, my short answer is that it will inevitably happen since it is a basic instinct of compliance and business operations, but it will be completely missing the point of the US export control policies to create two separate ecosystems.

DSET: This seems to be more on firm-level decisions. I wonder if you have more insights on the state-level policies.

Wolf: This has been for several years and is the inevitable result of the US extraterritorial controls since companies do not want to be subjected to the US rules. It is also my passion to get allied countries to see the common interests, impose controls under their domestic laws, and eliminate de-Americanization. This would also align with America’s interests if allies agree to the same controls, as multilateralism levels the playing field for the US industry as well.

DSET: How do you assess the effectiveness of export controls? I think the metric should include how much it delays China’s technology development.

Wolf: When I was in charge, we deliberately didn’t look at economic impacts or effectiveness. My mantra at the time was that export controls don’t pick winners and losers. If something could be diverted for use in producing or developing a weapon, then we control it. We did not use export controls to give advantages to US companies or to harm their foreign competitors. Export controls were based on core principles that allies could agree on, which was regulating things that could assist the development, production, or use of conventional weapons or weapons of mass destruction.

My view is simplistic and not as complicated as all the economic security discussions because I believe that mere mortals (including myself) are unable to fully understand supply chains and technologies. Russia learned during the Soviet Union that trying to manage economies is impossible, let alone a complex advanced technology economy. Trying to manipulate all that for economic reasons will always fail, and regulations will always move slower than technology and business.

By the way, I’m not opposed to the policy objectives of the Biden team. I’m a big advocate for them and have been impressed by how the Biden team has thought more broadly about the emerging or the force multiplying technologies, such as AI, advanced chips, and tools, that are needed for China to modernize its military. We are not in the ‘90s, and  when I was in government, we weren’t quick enough to see the future that the Biden administration sees.

You have to make controls based upon principles at a root cause. If you didn’t impose a control because it wouldn’t be completely effective, then you would never impose any controls. The history of Cold War export controls has shown that they were always imperfect. The Soviet Union still got washing machines from the West. There will always be escapes, and it’s all about degrading and following through on a principle and just trying to make it more effective rather than seeking it to be perfect.

Regarding the actual effectiveness of degrading, I don’t think anybody has all the data, and unfortunately, sometimes commentary is influenced by ideology. Those strong China hawks will come out and say that these controls are very effective and they’ve shut down the Chinese semiconductor ecosystem and the military-industrial complex. And then the pro-business types will say that controls are crap and counterproductive, and it incentivized China to produce and deprived us of our income. I don’t know of any objective data and commentary that are not influenced by their positions.

My personal view is that we have to go by technology and then factor in effectiveness over time as opposed to making it perfect. For example, the cut-off of the EUV and the DUV tools from the Netherlands and the multi-patterning tools from the US and Japan have been very effective because companies like SMIC are only able to, at best, make seven-nanometer chips at very low yield with a lot of error.

This delays China’s ability to make advanced-node chips at scale for multiple years. That wouldn’t have happened if the controls had not been imposed. But then, if we look at all the other tools that are more node-agnostic, all the controls have been done by handing the market over to Tokyo Electron and ASML, and thus those controls are ineffective. There are also the back-end tools, the testers, which the Japanese companies have pretty much swapped out overnight since the US tool will taint items from exporting to China. And thus, those have been not only ineffective but counterproductive. 

Extreme cases like the A100s are very effective because the Ascend chip is not anywhere close to it, and the scale of SMIC production cannot come up to the same output as TSMC. That will change over time as non-US companies design their own GPUs because they don’t want to pay the price or be dependent upon Nvidia. It will be delayed a few years, but it will get there. The controls on manufacturing equipment are quite disruptive, but we won’t know the true degree of disruption because the 120 Chinese companies were just listed on Monday. We will see in a year or so whether they can design these things domestically or start buying from German or Japanese companies. Those controls might be very effective for a few months and then eventually become ineffective and counterproductive. 

So that’s an example of my argument. First, controls are based upon principles. Second, effectiveness is hard to measure because of data limitations and ideologies. The data is really hard to gather because all these reports do not project changes in relative market share that these controls will have on allied country companies versus US companies. There are also variations in non-public data about why certain tools were sold. Third, the definition of effectiveness and successful control is not clear. Is delaying China by two years a success? I think most people in the national security community would realize that it is never going to be perfect.

DSET: Following up on the last question, a lot of commentary has said that the US has done everything it can do regarding EDA. Do you agree with that?

Wolf: The ChinaTalk podcast with Dylan Patel and Gregory Allen is pretty critical, and I suspect Senator Rubio will also be critical. These controls are an example of two precise lines: one is going after specific companies, and another is concerning technologies with specific capabilities. And it’s clear that in the new rules, the Commerce Department addressed diplomatic issues with Japan and the Netherlands. I think a Trump team will be less nuanced and impose controls of tools across the board.   Interestingly, a less nuanced set of controls will be easier to understand, comply with, and enforce. 

A lot more could be done depending on the definition of effectiveness, but complexities get in the way of policy. It is very difficult for normal companies to figure out how to comply. Even for prosecutors, it is almost impossible to explain these rules to a jury. So, we are at the point where compliance is harmful because people don’t know what counts as violations. You can hurt normal trade, and you can hurt compliance. Unfortunately, that is my one criticism of the Biden team: they have written rules that mere mortals cannot understand, and that hurts their policy objectives.

So, to answer your question, I think there’s more that can be done. By the way, even if Harris had won, more would have been done because, in the last three years, Democrats have also come up with new ideas to address the gaps. However, the virtue of the Biden team is to cut off the input that is needed for the indigenous development of production in China of advanced semiconductors, supercomputers, and production equipment. They look for new ways to impose controls or get allies to impose them. And the best part is that they had a grand vision.

I hope that the Trump team comes in and it either continues that grand vision or adopts its own. In Trump’s first term, there were actions but no grand vision, and it was impossible for anybody to know what the goals were and impossible for any career official to know how to determine effectiveness other than being tough. The virtue of the September 2022 Jake Sullivan speech, the articulation of that in the October 2022 rules, and then in the outbound investment controls and the executive order, the vision has been consistent throughout the Biden team. The Trump team will need to adopt a vision.

DSET: In the ChinaTalk podcast, they specifically mentioned the two approaches to the issue of enforcement. One is the legal approach, which is very much related to your argument about the complexity of compliance. Another approach they mentioned is counterintelligence, which had been crucial in the Cold War era, and the CIA and other organizations had been assisting the export control regime. Moreover, if we think about Taiwan, this also implies more security cooperation between countries, which might be more difficult in the Trump administration. Do you see this as a viable path?

Wolf: I’m a lawyer and think of things through legal terms, but in the ChinaTalk podcast, they made a really good point about Secretary Raimondo’s China visit. She was surprised by the announcement of the Huawei phone during her visit, and it looked like a failure of the intelligence agencies. The degree to which the intelligence agencies provide support is not obviously something that has generally been published. It was done through the expertise of intelligence analysis in our enforcement group with staff, who have the clearance to go to various US intelligence databases for information and review individual license applications against information available to both the agencies and the intelligence community. Every year, they have gotten more resources to do more. And in terms of policy evaluation, I’m sure they’ve tapped into the intelligence community. Again, I have no idea because I’m not there. 

The intelligence community is weak in the sense that it’s able to gather raw data, but what is needed are technical experts, economists, and supply chain experts. And so, my suspicion is that there is an enormous amount more the intelligence community, which has trillions of dollars of budget, could do. One is to support the US agencies more in making both policy and licensing decisions. Secondly, Taiwan is not part of the FVEY community, where truly exquisite intelligence is shared among certain allies. However, declassified intelligence sharing as part of both enforcement and policy about particular end-users of concern has been fairly traditional, even when I was in government. The reason for not sharing intelligence is to not reveal the sources, and there are ways in which you can share declassified information without revealing sources and methods. It is clear that more should be done. However, as a governmental checklist, countries should have intelligence community support in addition to legal support.

DSET: Following up on the last question, do you imagine how to strengthen the intelligence support for the export control regime?

Wolf: More resources to the intelligence services and connecting those to the export control decision-making and enforcement so that information can be gathered from other parts of the government is obviously a virtue. The US has an amazing network, as does China, Russia, and other countries, of both signals intelligence and human intelligence, particularly signals intelligence of gathering information through the internet and other sources, which smaller countries usually don’t have because of limited resources. That’s where intelligence-sharing arrangements often come in, where the governments of Taiwan and other countries can benefit from that infrastructure. So yeah, it’s a long way of saying a great idea was a good point that they made on the podcast. I have no idea of the scope, but in terms of a recommendation for making the system better, that should absolutely be in the queue.