2025 / 1 / 10

Trump 2.0: U.S.-China Semiconductor Competition Policy — DSET Interviews Matthew Turpin

Author: Fanny Chao

Table of contents

Introduction

During President Trump’s first term, U.S.-China semiconductor competition escalated rapidly, driven by tariff wars, Huawei sanctions, and later evolving under the Biden administration into export control regulations and the “small yard, high fence” strategy. Matthew Turpin, who served in the National Security Council under Trump and shaped the U.S.-China semiconductor competition strategy, is now a visiting scholar at Stanford’s Hoover Institution, a senior advisor at Palantir, and the co-author of the book Silicon Triangle: The United States, Taiwan, China, and Global Semiconductor Security. Turpin remains a key figure in the U.S.-China semiconductor rivalry.

DSET interviewed Turpin to assess the past four years of Biden administration policies, China’s advancements in advanced chips, the risks of overcapacity in legacy chips, and U.S.-Taiwan cooperation. The interview, conducted on October 22, took place shortly before the U.S. presidential election. With the Republican Party poised to assume full control of government, Turpin’s insights on competition strategies and policy tools are thought-provoking for Taiwanese readers.

Development of U.S.-China Competition Policy

DSET: As one of the main planners of the U.S.-China semiconductor competition policy during the first Trump administration, could you walk us through the discussions that took place? How did you, the National Security Council, and the Department of Commerce collaborate to design the regulatory policies? Additionally, it seems that the U.S. government is in the process of identifying the best methods for regulation. For instance, the Department of Justice and the Office of the U.S. Trade Representative have each made regulatory efforts. How do you evaluate these initiatives?

Turpin: Certainly. At the end of the Obama administration, there was deep concern about semiconductors, the activities that obviously the big fund had already been stood up by the PRC. We were starting to see money move in those directions. We had witnessed either talent poaching or cyber espionage against various semiconductor companies. So then Secretary of Commerce Penny Pritzker, who gave a speech on U.S. semiconductor leadership and the things that the Chinese government was doing to undermine the semiconductor field. She gave that speech on November 2nd in 2016, at CSIS. Seeing that speech, you’d think that’s what somebody would be saying today. So, going all the way back eight years, we can find someone laying out the arguments of what’s going on, and that reflected a set of policy discussions that were happening across the Biden administration or the Obama administration on this. To look at what tools were available, how do we work with export controls, how do we think about various other actions that either the Department of Defense, State Department or the Treasury could take? I was involved in that effort, and learned a few things from that. And when I went into the Trump administration, I brought that knowledge and the connections with the people that have been working on that, and we started it up again.

There were three different issues mixed together: the issues around semiconductors, the issues around the transition of the types of telecom equipment that we would have for networks, which is from hardware-defined-networks to software-defined-networks, and the issues around violation of U.S. export control laws, by two big Chinese telecommunications equipment manufacturers, ZTE and Huawei. These are three separate issues, and they all happen to be coinciding at the same time, and in many ways, the common denominator across all of those is advanced semiconductors. 

Like, the thing that makes 5G base stations and various other things, radio, net access, network boxes and other things, advanced is that they have, whether they’re FPGAs, the restricted chips that only the United States makes. That’s what you’re actually talking about as being the advanced equipment, not the plastic box. So, those things all came together, but they all touched on separate authorities. 

Our criminal cases against ZTE and Huawei for violating U.S. export control laws had predated all of this. The investigation into ZTE started in around 2012, and it was about finding a set of documents that a ZTE executive had had on him as he entered the United States, in which they described how they would violate U.S. export control laws and sell to Iran. And in those documents, they laid out how one of their competitor companies, a company code named F7, and it was Huawei, how they were already doing this. And they described that this is why they had to copy what Huawei was or F7 was doing because, like, they were losing market share because Huawei was violating U.S. export control laws, so they should do it too. And so that case against ZTE led to another case, in the summer of 2016 against Huawei. Those prosecutions were working their way through a legal system. ZTE was put on the Entity List in March of 2016. It then has a denial order put against them in  April of 2018,  the denial order is different from any listing. And Huawei isn’t put on the Entity List until May of 2019. So these things are all sort of happening on their own timelines, and they’re really based on a legal right. They’re really based on a legal case and a prosecution against Huawei and ZTE, not necessarily on the other things that are going on. 

At the same time, we’re looking at semiconductors, and what the Chinese are doing around semiconductors, and what we’re looking at in terms of the 5G transition. So, all those things came together at the same time, but they’re different activities.

Then, the foreign direct product rule experiment developed over time. Obviously our actions don’t stop when we put Huawei on the Entity List; in December of 2018, it’s the arrest of the Huawei CFO that then prompts the Chinese government to take two Canadians hostage in order to force the release. It isn’t until May of 2019 that Huawei has actually been put on the Entity List. And that coincides with a breakdown in the trade negotiations. That’s the time. 

In that intervening time, there’s work going on to think about the cost and benefits of placing Huawei on the Entity List, largely because Huawei had been buying $2 billion worth of U.S. goods over the preceding years. So, what would hit U.S. companies if they weren’t able to export to Huawei anymore? Of course, an entity listing does not mean that automatically all companies have to cease exports to that company; it means that they have to receive licenses on those things that are covered, and that those licenses are judged with a presumption of denial. In many ways, as we saw in the summer of 2019, a lot continued to get exported to Huawei, largely because what they were buying from U.S. companies were things they could buy from a variety of other companies. So it didn’t make much sense to deny, for instance, Micron-made memory chips because they could easily shift over to someone else. Huawei bought a lot of Micron’s memory chips, I think they were buying $200-300 millions dollars a year worth of those chips. But restricting Micron from selling chips that are not particularly advanced didn’t make much sense. That wasn’t what we were going after. We were going after those FPGAs and those other things that it was really hard for them to replace.

The case I would look at is the Fujian Jinhua case, which manifests at the end of 2018. We put Fujian Jinhua on the Entity List in December of 2018. That shows an effort to use export control regulations in a way to take action against a company that has abused, from our perspective, stolen, Microns technology. And it used the Taiwanese firm to do that. We prosecuted the Chinese firm and the individuals involved in that, and the Taiwanese government cooperated with that. But the ultimate beneficiaries of that stolen IP, a state owned firm, Fujian Jinhua, which had just stood up to essentially produce memory chips, copies of what Micron was doing. Our action against them was largely because we defined the Entity List as an action against people who violate the foreign policy, national security interests in the United States. Receiving that stolen IP, and then putting it into production violate what we would consider our national security and foreign policy interests. So placing them there is really kind of that first step to use those export control regulations to shape the outcomes of that. And I think that actually leads to it more than the entity listings of Huawei and ZTE are related to court cases, which are the punishments for which companies that refuse to comply with U.S. export control laws knowingly. Fujian Jinhua is, in a conspiracy, to get stolen IP and then make those chips themselves.

Evaluation on Biden’s Policy

DSET: Four years have passed under the Biden administration, which has continuously updated and strengthened export controls on semiconductors to China. How do you evaluate the Biden administration’s semiconductor policies towards China? Are there areas you think need improvement?

Turpin: My complaint is on speed at closing loopholes, and a willingness to be more aggressive and more expansive. I think there’s a desire on their side to compete, but they also want to persuade China that they have a mutually beneficial economic relationship. And those are not compatible objectives. So, the challenge is where you’re really going to go after. If you’re going to start this, then make sure you actually do it. Don’t half-ass it. That’s my main concern. 

On October 7th, 2022, export restrictions were put into place, within two or three months, I think there was a pretty good understanding of the loopholes the Chinese were going to use and were using to get around those restrictions. And it took until October of 2023 for them to begin to close those loopholes. And in that interim, Beijing made a whole bunch of purchases of tools and various other things to build out their capability. Then, within two or three months, we knew what the loopholes were in the October 2023 tightening, and they still have not yet closed those loopholes. So, you can choose to be slow and deliberate, and maybe there are some benefits of doing that, but not in this case. I think there are very few benefits in being slow and deliberate in this case. 

In terms of outbound investment restrictions, the way I view it is that they spent two years internally arguing whether or not restrictions should be put into place. The Treasury Department was the one arguing against putting them in place, and the negotiated settlement was, let’s give the Treasury Department a year to figure it out. To me, we had been discussing outbound investment restrictions for years before that— they were a part of the whole FIRRMA and ECRA reform. They had been a part of the entire reform around export controls and CFIUS. They got pulled out because certain constituencies did not want to be restricted in how they invested in China. And that just continued on. So, we’ve made announcements that we’re going to do things, but then we don’t actually follow through on them. And that, to me, is like the worst of both worlds. If you’re going to do something, do it, don’t half-ass it. That’s my main complaint. 

There’s really effective lobbying on behalf of the companies that want to sell. Those companies have done enormously well since October of 2022, they have done massive business in China because Beijing could see that the window might close, and so they are rushing through as many sales as possible. That is just absolutely mind boggling that we allowed that to happen. They didn’t have either the political will or the consensus internally to make the decision to do it. They had no thesis of what they were doing. Because, again, they had convinced themselves that what they’re going to do is this very narrow action. The reality is that once you start to try and restrict AI chips, there’s nothing narrow about AI chips. AI is going to presumably touch every industry, so there’s nothing narrow about that. So I think they had a set of rhetoric that was meant to appeal to both Beijing and to their partners and to the business community: hey, we’re being responsible and narrow and very scoped on this just national security. But they picked out a thing that is the most dual use. In order to stop it, you’re going to touch every industry. 

And so instead of making the decision to actually do that, they pick the middle ground which is the least optimal for both of the outcomes. If you don’t want to make China mad, and you want to convince China that you have a mutually beneficial economic relationship, then don’t do those actions. But if what you want to do is prevent them from becoming advanced in those areas like artificial intelligence, then actually prevent them from doing so and take the action. And if that means that China gets angry, so be it, right? I mean, they’re building a military to destroy you, if they’re angry, so be it. Pick your policy. Don’t try, what essentially you got is a negotiated policy, not a not a bold policy. You got a negotiated, watered down policy. 

DSET:  The U.S. government is often criticized for having insufficient manpower when enforcing semiconductor technology regulations, especially regarding the execution capabilities of the Department of Commerce Bureau of Industry and Security (BIS). How do you view such criticism? Besides the BIS, are there other departments involved in implementing export controls? How do these departments collaborate?

Turpin: In all fairness, there’s a whole team that sits over at DTSA, the Department of Defense’ Defense Technology Security Administration, that works on all these things. And there’s an entire team in the State Department’s ISN, Bureau of International Security and Nonproliferation, that handles export controls for both dual use and the International Traffic in Arms Regulations (ITAR), and they have a whole set of folks there. I think there indeed may be a limited number at BIS, but the running of those export controls actually spans a number of departments. And if we look at where the resources are and the resources they can pull on across the rest of the U.S. government, it’s more than three.

Evaluation on China’s Capability of Advanced Chip

DSET:  In the face of such export controls, China’s technology continues to advance. Huawei seems to have already produced 5-nanometer chips, and reports indicate that Xiaomi has successfully trial-produced 3-nanometer chips. Moreover, according to a recent on-site report by CommonWealth Magazine on Chinese AI companies, they believe that the actual impact of U.S. export controls is limited. On one hand, China can develop its own chip alternatives; on the other hand, the chip smuggling market allows these companies to still acquire relevant products. How do you assess the progress of China’s technology? Do you think China still has the capability to compete with the United States in advanced technology fields?

Turpin: My sense is that at the scale they need the number of advanced chips, smuggling is actually quite difficult. To get them at the number that you need, you could get hundreds or thousands, but it is much more difficult to get tens of thousands or hundreds of thousands. So, could we go down to a night market or whatever in Shenzhen, and find H-100 chips? Sure, I’m sure we could. Can we find 10,000 or 100,000 of them? No. But that’s the scale needed. So, they’re absolutely right, there are some available, and smuggling does work, but that is not a long term industrial level solution. 

I think in many ways, what we’re seeing is a desperate effort by Beijing and some of its corporate allies to portray export controls as ineffective and something that the U.S. government shouldn’t even bother with. They don’t work, so the U.S. just shouldn’t even do it. There’s an old saying in the U.S. Air Force that the flak is heaviest when over the target. So when the Chinese government complains the most about bringing up export controls over and over again, on the one hand saying that this is terrible and it must be relieved, and on the other hand saying it’s not effective, you shouldn’t even bother doing this, because we’re getting around it and releasing a three nanometer chip. You should be a little suspicious of the time and energy they put into releasing all of that. Doing all that might be because you’re actually making a difference, right? Because if they were actually being quite effective, they might not want to talk about it. So one of the reasons for talking about it is to feed a narrative that export controls don’t work. 

And that’s a strong narrative in Washington right now. There is a resourced effort across multiple think tanks to portray export controls as ineffective, and therefore we shouldn’t even bother. And when you scratch the surface a little bit, there’s a handful of companies that spend a lot of time funding research and reports. So we should understand the dynamics of what’s happening. Here’s a company that’s complaining about all the harm that the export controls do. I think their growth since 2019 or 2020 has been around 400 or 500%, so my uneducated understanding of the commercial dynamics is that the October-7th-2022 export controls have not done anything to harm this company. In fact, it has done incredibly well since that time. Just go look at their stock price. Wall Street thinks they’re doing well. So either they’re lying to their investors, which I don’t think they’re doing, or they are making a lobbying case to the U.S. government that export controls are too harmful and you need to stop them. I think that the latter is far more likely, and that Wall Street actually does assess that they are making a lot of sales, mostly to the Chinese.

Evaluation on Legacy Chips

DSET: In the field of legacy chips, you warned in the book Silicon Triangle: The United States, Taiwan, China, and Global Semiconductor Security about China’s strategy of  massive over-capacity of legacy chips and its harm to the overall market. Could you elaborate on this strategy and its impact?

Turpin: In all fairness to the Biden administration, they have made a policy decision to go after a narrow set of advanced chips. The tools, the materials and the know-how that goes to that, and they have specifically decided not to do more legacy chips. That’s maybe not the decision I would make, but that’s the decision they made. So we should judge their policies based upon that decision. So we shouldn’t criticize their export controls for letting that go through when, in fact, they weren’t actually designed to stop those things from going through. 

The criticism I tried to lay out in the chapter in Silicon Triangle was that: this is a competitive process between us and the PRC. We laid out the logical step, like the most obvious step that Beijing is going to do when it’s restricted at the high end, is to do over-capacity in legacy chips, which will then rob the revenues of all of those chip companies, who need those revenues to do further innovation. If the ecosystem of companies are no longer commercially viable, all the R&D money you would use to go to the next generation isn’t there. And the CHIPS act isn’t enough to fill that gap, nor is the money in Japan, nor is the money in Europe, nor is money in Taiwan. The Chinese will be able to essentially halt innovation in that field. While it’s right that we’re going to likely hold them off, we won’t continue to advance. 

So, what we expected is that taking this action on China at advanced nodes is going to be followed by the massive over-capacity of legacy chips. That means that Media Tech, UMC, Micron, SK hynix, and presumably Samsung and Intel are going to come under enormous pressure, as China ramps up its production of legacy chips, which are the actual revenue makers for the semiconductor industry. And, it’s those legacy chips that primarily go into advance in the military systems. They aren’t advanced chips. The chips in this iPhone are not the chips in an F-35 right? The F-35 has older chips in it. 

So again, this goes into if you’re going to start a fight, be prepared to go all the way through. Don’t just pretend that you can just take this narrow step and then the other side is going to just give up. You need to actually follow through with it. It seemed that they didn’t think through, or they thought it through, but they couldn’t get the consensus, because, again, they couldn’t decide what their policy was. Are we actually competing with China to win? Or do we want a mutually beneficial economic relationship, as Janet Yellen has talked about in her China policy speeches. So which policy are we following? Pick one. The reality is that we have tried to do both simultaneously, and that is a terrible outcome.

DSET: The U.S. has a tool box to address legacy chips. In Silicon Triangle, you mentioned anti-dumping and countervailing measures (AV/CVD). During Trump’s term, Section 301 investigations were common, and former Congressman Mike Gallagher called for tariffs on products with China’s legacy chips. Could you assess the applicability and impact of these tools?

Turpin: The process of an AV/CVD needs a company that’s been harmed to bring a case. So let’s imagine that China starts dumping DRAM chips to the United States, that means an American company would have to come to the U.S. government and accuse China for dumping in the U.S. market. Right now, that’s like Micron. Yet, Micron is unlikely to do so because they fear retaliation in the Chinese market. 

Section 301 investigation is also an option,  it enables a broader set of actions. It’s completely up to the executive branch of how it uses it. The President would have wide authority to be able to do things. To impose Section 301, you have to do an investigation that is opened up in public comment for 6 months. So, let’s say the administration starts on January 21, a few months before they even could start the investigation, the process takes at the fastest six to eight months. You could take an existing Section 301. There’s a chance that we don’t know all the investigations are ongoing now, so there may be some that will finish up and would be available to a new administration. ​​(Note: According to media reports, the Biden administration plans to launch an investigation into China’s legacy chips under Section 301 of the Trade Act before leaving office.)

For other options, you could change export control regulations relatively quickly, and impose those changes so you should block the export to certain components or software or spare parts. Also, if you find a connection with forced labor that could fall under the Uyghur Forced Labor Prevention Act, the customs department can block it at the border. There’s a variety of options. All those things you’d have to investigate more thoroughly. Some of them are much quicker than others. Some of the burden of proof is relatively low.

My fear is, if you get to a point in time where it’s a little bit too late, and then you need to think about it, okay, so what do you do afterwards? It could be that there are still things that can be done. With tools like the executive order of Office of Information and Communications Technology and Services (OICTS), you could prohibit the import of change manufactured legacy chips. Which would force a company like Apple or any other ones that wanted to sell a job to the United States not to put those chips in and only rely on chips that are either from Japan, the U.S., Taiwan or Korea. You could do that. It might not be easy to do, but you could do it, which would then keep that market open. 

Future of U.S.-Taiwan Discussion

DSET: What differences do you anticipate in the next Trump administration’s policy on semiconductors? For readers in Taiwan who want to observe the future policy directions, which discussions should they pay particular attention to?

Turpin: It’s hard to predict. Because fundamentally, we could see now, from the Trump administration, who also had a hard time picking out what its policy was, to be perfectly honest, and a Biden administration has had a hard time picking out what its policy is. I suspect that we should be pretty confident that it’s going to be hard for whatever administration comes next, and they’re going to be competing voices calling for actions that are circumscribed, rather than following this way through. 

Now, I think what we should be looking at is what we think Beijing’s intentions are. And I don’t think Beijing is trying to manage competition, trying to figure out a negotiated middle ground. I think they’re trying to win. To steal a sports analogy, if you’re not going to try and prevent the other side from scoring points, and you think it’s only important for you to score just one or two points. That’s a really good way to lose the game. 

DSET:  It is said that in early 2020, the U.S. government began discussions with Taiwan on export control regulations, and in May of the same year, TSMC announced plans to build a plant in Arizona. How do you assess the significance of TSMC’s decision to build a plant in the U.S.? And the future of this investment?

Turpin: I was not involved in that since I had left near August of 2019. As that all that started to pick up, my impression is that: while certainly the White House had some interactions with that was mostly led by then Under Secretary of State for Economic Affairs Keith Krach and a team in the State Department. 

As I look at it now, there’s a variety of reasons for why it would be in the U.S. interest and in TSMC’s interest to expand the locations where they produce advanced chips. The first is that there’s a saturation point in Taiwan of the number of fabs that you can have, and the talent that’s available, the power that’s available. All those things, you’re reaching a saturation point where, let’s say the demand for chips is going to increase by 50% over the next 10 years, it’s very difficult to see how they could be 50% more fabs in Taiwan, right? So at some point in time, TSMC needs to start to think through, what does it look like for them to expand their production significantly, and where would you do that? I think the reality is that it just isn’t all going to fit on the west coast of Taiwan. Just isn’t the space or the power or the people physically available to do all of that. And so I think it took a little while for them to work their way through at which point in time and where. I think there was probably some thought early on that they could do that in China, and then they found that there would be quite a few restrictions that would be placed on them in terms of tools, precursor materials and other things. If they chose to do that in that direction, the U.S. government and others would take action to prevent that from happening. But if they did that in Japan, in the United States, or in Europe, they could get benefits from doing that, there could be incentives that they could gain from making that happen. 

And I think that’s fundamentally how the negotiations worked out. If I’m TSMC, I would hold out a bit and hope for a pretty sweet deal to be able to make that happen. And so I think that’s kind of what happened. But I mean, fundamentally, TSMC had to think about expansion, so it’s kind of a no brainer, right? Certainly, energy is going to be cheaper in the United States, right? So, you start to think through some of the things that you can do, and it looks like this is a deal. 

And of course, none of that undermines the absolute necessity of all those fabs on Taiwan. Because all those fabs are going to still be needed. You still need all of them to churn out chips. We’re talking about additional capacity to churn out advanced semiconductors like 14 nanometer and below. I think TSMC’s calculation is the world’s going to need a lot more of those chips, and if the U.S. and others are eventually going to prevent the PRC from being able to move to that direction, which we ultimately did with the October-7th-2022 export controls, then it’s TSMC who is likely to be the one that does that expansion. So, it seems like a fairly good deal for both sides.

DSET: How do you assess the concept of “silicon shield”, and its impact on the next administration’s semiconductor cooperation policy with Taiwan? For instance, as supply chain security gains attention, might TSMC face more U.S. demands to expand facilities or to deepen ties with American suppliers?

Turpin: I think we should be very clear that Beijing doesn’t wish to take Taiwan because of semiconductors. Semiconductors are not the reason why either Beijing covets Taiwan or is deterred from taking Taiwan. The Chinese Communist Party desires to take Taiwan because it is a threat to the legitimacy of the party. It demonstrates that the Taiwanese people can have a prosperous democracy without a Leninist Vanguard party leading them. The party is terrified of mainland Chinese learning that, in fact, you actually don’t need a Leninist Vanguard Party that has a monopoly hold on power, you could actually have a multi-party democracy, and you could be prosperous. That’s the reason why the party wants Taiwan. We should be very clear, semiconductors are not the reason. So the idea that there’s a silicon shield that guards Taiwan is a myth. So, what do I think of the U.S. Taiwan relationship is likely going to continue to be the relationship we’ve seen over the past few decades. The U.S. has its own interests for why it wants to have a strong relationship with Taiwan, which is independent of the competition with China. We have our own interests in doing so, and that’ll continue.

For TSMC’s part, I would go back to the logic of what we discussed: if you’re TSMC, and you’re looking at the total available market for the kinds of chips you produce. And you do some back of the envelope planning on, like, the number of fabs you need. It’s really hard to understand how you would put that number in Taiwan, right? Given that your existing ones are all going to still run and you’re going to upgrade them, they’re going to continue to churn out advanced chips. The number of chips that the market will buy is going to exceed what you can produce in Taiwan. And so I would suspect to see TSMC say: once I have a footprint in Japan, Arizona and Europe that is functioning and able to grow, I’m going to grow that. Because, in fact, it’s in my interest to be able to expand, to be able to service the market that’s under demand. And if I don’t do that, I’m going to get a competitor that will grow to be able to fulfill that need, and then I’m going to be under pressure, right? So I think from a commercial perspective of TSMC, it makes perfect sense to do this. 

Share

Related Articles

Policy Commentaries
Artificial Intelligence

Winning the Marathon, Not the Sprint: An Interview with Technology and the Rise of Great Powers Author Jeffrey Ding

Author:You-Hao Lai

2025 / 1 / 10

Policy Commentaries
Cybersecurity

Taiwan’s TikTok Liberal Paradox

Author: Kai-Shen Huang

2025 / 1 / 9

Policy Commentaries
Semiconductors

Trump 2.0: Semiconductor Industry Policy Dynamics — DSET Interviews Jimmy Goodrich

作者:Cosette Wu , Chen-an Wei, Min-yen Chiang

2025 / 1 / 6