Trump 2.0: Semiconductor Industry Policy Dynamics — DSET Interviews Chris Miller

Introduction

In response to the escalating US-China technology competition and the potential return of the Trump administration to the White House, DSET invited Chris Miller, author of Chip War, for an in-depth discussion. This dialogue focused on the outlook for semiconductor policies under Trump 2.0, including tariffs, export controls, and the CHIPS Act, as well as the investment pressure on TSMC and related human resources challenges. Additionally, both sides explored the threats posed by China’s mature-node chips, the loopholes in US AI export controls, and how Taiwan can seize strategic opportunities from supply chain shifts while deepening cooperation with the United States.

Cabinet Pick

DSET: Regarding Trump’s cabinet pick, National Security Advisor-designate Mike Waltz currently heavily focuses on defense issues and his stance on export controls and semiconductor issues remains unknown. In addition, Commerce Secretary-designate Howard Lutnick hasn’t publicly expressed any opinions on export controls or semiconductors either. I’d like to know your take on this.

Chris: There will likely be extensive analysis of all the different picks in the next Trump administration. However, the thing more important than any individual personality is the bipartisan consensus around these policies, which has persisted from the first Trump administration through the Biden administration, and will likely continue in the next administration. These issues have support in both chambers of Congress. Although personalities may influence details, the overall framework is expected to remain consistent. This is because the first Trump administration played a major role in establishing the framework of subsidizing manufacturing in the U.S. and preventing technology transfer to China. I don’t see that shifting, regardless of who is involved.

DSET: In your book, there is a fantastic story explaining how Robert Lighthizer’s proposal and agenda back in the first Trump term were different from the then National Security Council. And, in the interview right after the election, you said we should put more attention on the Lighthizer, he might be the key figure. Have you seen his agenda personally evolved into putting more attention on export control and semiconductor industrial policies over the past years?

Chris: I think the answer is not really. I think that Lighthizer and Trump have been super consistent on this for a long time. They believe tariffs are an effective policy for reducing reliance on China and incentivizing domestic manufacturing bases. I think this is something not tech specific, it is about broad manufacturing. In Trump’s first term, their first set of tariffs were on steel and aluminum–– so not high tech, it’s the opposite in some ways.

So, I think manufacturing in general and reliance on China in general is how both Lighthizer and Trump see the world, judging by their statements. I think one difference with the Biden administration approach and the Trump administration approach has been that the Biden administration has tried to describe specific spheres where it’s focused on dependence vis-à-vis China, such as new energy technologies and EVs. I think that the Trump approach was, and I think will be, much more broadly based. Any sort of manufacturing imports from China will be seen as potentially problematic. That is a difference between Biden and what Trump will bring.

DSET: How about industrial policy-wise?

Chris: I think the policy would primarily be tariff-wise, actually. I think tariffs are going to be their preferred tool. The argument of people like Lighthizer would go something along the lines of that industrial policy creates a dynamic in which companies are lobbying to get funding and the government distributes the funding, so they would say that’s not a good approach. It’s better just to impose tariffs in China, and you’ve got a strong financial incentive as a result of that, to invest outside of China. And I think that’s their argument. You can debate whether that’s true or not, but I think that’s how they frame things.

DSET: It brings us to another two questions. First, if Lighthizer is going to be the trade czar, doesn’t that sound risky to us that tariff policies might be prioritized over these export control efforts? Especially when we see this new Commerce Secretary, who doesn’t really have much experience on relative policies. So in your opinion, would it take more time for the Trump administration to figure out the priorities between the two strategies? Second, because Lighthizer and Trump have emphasized industrial policies, do you see them pushing more on the Chips Act’s subsidies?

Chris: I think it’s likely that the new administration is gonna devote more of its attention on tariffs than on export controls, but the existing export controls will remain in place. Even if they devote no attention, you still have the existing controls remain in place. Which is still an impactful status quo. So I think it can both be true that they prioritize tariffs, and that they continue to pursue export controls because export controls are already on the books. But I think, if you ask, what will they spend more time on, tariffs is clearly the answer. In terms of industrial policy, I think the Republicans have generally been more skeptical of the cash incentives that the Chips Act had. We saw the debate in Congress, for example, there’s more skepticism about that, Trump has a couple of times said something similar. I think the Republicans have generally been very supportive of the tax credits or and actually in dollar terms, the tax credits would be more impactful than the cash incentives. The tax credits expire at the end of 2026. There’ll be a big tax piece of legislation passed next year by Congress, and so I wouldn’t be surprised to see the tax credits made permanent for example, as part of the tax legislation that comes next year. So that’s still industrial policy of some sort. It’s a different type of industrial policy, and the Commerce Department’s been executing. But I think the financial impact is the same, which is to provide financial incentives for companies to invest locally.

Chips Act and Tariffs

DSET: During the campaign, we also saw Trump, out of nowhere, saying he might impose on Taiwanese chips. Just want your observations about that.

Chris: I think that was a one-off comment. I guess I would say, Trump has threatened to impose tariffs on basically everyone. And I think one of the lessons from the first term is that one inevitably has to prioritize as president . So I take Trump at his word that his optimal endgame is to have higher tariffs across the board. But I think, if you ask yourself, what are the countries that Trump is going to prioritize imposing tariffs on? It seems to me that China is number one. I think you could envision tariff threats against Mexico as part of immigration debates, but no actual tariffs being imposed, which is exactly what he did the first term then, I think. 3rd with regard to Europe, where there’s a series of trade disagreements. And so that would be my guess as to what actually gets prioritized.

And if you ask yourself, what’s the political economy of hypothetically imposing tariffs on Taiwan, it’s true that Taiwan’s exports to the U.S. have surged in the past 2 years, it’s also true that the primary driver of that is Nvidia servers, so it seems implausible to me that the U.S. would respond to a boom of investment in AI by imposing tariffs that would hit the production process of servers that make AI possible. So I think when you start to think through what’s the likely prioritization of tariff threats, and when you start thinking of the political economy of which firms and which consumers would begin to be impacted by tariffs. I would guess that Taiwan is pretty low on the likelihood of countries that actually get tariffs imposed on it.

DSET: Trump has discussed tariffs being a potential substitute for subsidy-based industrial policy, particularly regarding TSMC. Given TSMC’s pricing power, tariffs currently affect U.S. consumers more than the company itself. However, there are key supply chain members, with capacity in both the U.S. and Mexico, not receiving U.S. support.

After Trump’s election, we have seen movements and announcements about increasing capacity among Taiwanese downstream server manufacturing firms, such as Foxconn trying to increase capacity in Wisconsin or Texas. One specific issue is whether additional punitive tariffs will be levied on Mexico to push Taiwanese capacity from Mexico, which is still a desirable location for manufacturing, into the US. Such a shift, without subsidies, could be potentially more costly. Could you comment on that?

Chris: I would say that tariff debates with Mexico are likely to revolve around two main dynamics. First, immigration. The U.S. will likely use tariff threats with Mexico as leverage to address immigration issues, as Trump did during his first term. This approach can be expected to continue.

Second, the USMCA already provides a framework to determine where tariffs should or should not apply. It’s unlikely that any of the three countries involved would take steps that seriously threaten the agreement, given its importance to businesses across all three nations. This was a key takeaway from Trump’s first term, when he pushed for a renegotiation of NAFTA, the resulting changes between NAFTA and the USMCA were more updated than punitive measures against any specific country. Many trade experts describe it as a modernization effort, and Mexico, in particular, seemed relatively satisfied with the outcome of the USMCA.

Therefore, I wouldn’t expect measures specifically targeting servers or the server supply chain in Mexico. If there were to be a scenario involving across-the-board tariffs on non-USMCA countries, the impact on the server supply chain would likely be mitigated. In such a case, companies might shift the final assembly process to Mexico while continuing to source components from Taiwan and the rest of Asia. The economic impact of this would probably be limited since the assembly process is the cheapest and simplest part of server manufacturing. While such a move might have political implications, its economic impact would likely not be substantial. That’s how I would interpret the situation.

DSET: This raises an interesting question about the technicalities of tariffs, as they target the final goods rather than intermediate inputs. For instance, with the mature chip issue, the question becomes whether we should tax just the Chinese chip itself, which isn’t directly imported by the U.S., or electronics, like TCL screen monitors, which definitely include Chinese chips. However, doing so would be more politically costly. Do you think that’s something Trump would consider? Or will he follow the pattern of Biden’s tariffs on Chinese goods, which are also practically ineffective for this reason?

Chris: There’s considerable discussion within niche circles about component-based tariffs. This conversation initially gained traction with mature semiconductors but is also relevant for displays, which share similarities with mature semiconductors. For example, 80% of global investment in displays occurs in China, with BOE, the third most heavily subsidized company in China according to official data, leading the market. It’s not difficult to draw parallels between these two industries. I believe there’s a real possibility that component-based tariffs could materialize.

The main critique of component-based tariffs is their logistical complexity. While this criticism has some truth in it, it increasingly falls on deaf ears in Washington. Manufacturers often argue that they can’t track which chips are in their systems, but this explanation is losing traction.

Another critique of tariffs during Trump’s first term was that companies simply shifted final assembly to Vietnam without significantly altering supply chains. A component-based tariff regime could be more targeted and impactful in influencing supply chain decisions.

The challenge, however, is that implementing component-based tariffs requires the government to be highly informed and agile, which historically hasn’t been a strong suit for the U.S. government. This remains a significant hurdle. Nevertheless, I believe there’s a reasonable chance we’ll see component-based tariffs applied, particularly for semiconductors and displays.

DSET: And on legacy chips, you mentioned that component-based tariffs might be where we end up. In your previous interview in March, you also mentioned there were different methods. First, just tariffs. Second one, as you proposed on the Hill in June, just banning all the Chinese ships from entering into the critical sectors. But we haven’t seen much progress on that.

Chris: I think we’re gonna see more there. I think there’s a rule called Information and Communications Technology and Services, ICTS. I think the final rule is going through a regulatory process. And I wouldn’t be surprised if they issue the final version of that rule before the Biden administration ends. And that is a rule that lets the U.S. government stop questionable electronics in critical sectors. And so I wouldn’t be surprised if, when the Trump Administration takes power and that rule is finalized, there’s some folks in the Commerce Department who would like to use that rule.

DSET: Regarding legacy chips and export controls, there are people commenting that export controls on such chips probably don’t make much sense, because China already has the ability to manufacture most of the equipment themselves. However, in the industrial information industry, they talk about manufacturing standards and keep a look at the nanometer land. The Shanghai firm’s capability to manufacture on scale is above 65nm, although it previously claimed that they could manufacture to 28nm. So one plausible tool for the U.S. is to extend the current limits to 28nm. But, the financial reports of all these companies indicate that as of now, 30 to 40% of their revenue is from China, as China desperately needs them too. So this is going to have a lot of push back on that. So, do you think that is still an option considered by the policymakers?

Chris: I think a blanket ban is highly unlikely. I think one or two ratchets upwards seems possible, but I think that the internal debate will be the China hawks saying let’s ratchet upwards, and there are certainly people on the Republican side who would say 28nm is a better line than than 7nm or 14nm. 

I think the counter argument that you’ll hear is, first, industry will say, this is too financially painful for us, and it would be to some degree. And, second, industry will point out that there’s more alternatives as you get to less sophisticated equipment, so it’s not just a dozen firms that you need to control, and not all in the U.S., Japan and the Netherlands. Necessarily you start getting some Korean firms that have comparable capabilities, so it’s a bit messier. I think lithography is still probably the place where you’ve got the most ability to, as it still is basically two firms. But I think it just starts to get messier.

Then you ask the question: if you can’t get allies to join, are you willing to go unilaterally? And I think the Trump administration will be willing to threaten that on certain issues, but I also think there will be limits to willingness as well. 

Could you imagine the administration then saying, in addition to export controls, we’ll impose financial sanctions. But that’s a costly route to go down. So I think there will certainly be people in administration who say that it sounds like a too costly route relative to the benefits. So I think that’ll be a debate. I don’t know exactly where that debate will land, but I think it’ll be it won’t land at a complete ban on equipment for legacy chips. It might land a bit ratcheting upwards, one or two notches, but not, I don’t think it’s a huge change.

DSET: During the campaign, we also saw some discussion from Trump about repealing the Chips Act, and also Mike Johnson at one point, saying that he would walk it back. How is that gonna play out?

Chris: I think the first thing to say is, what political leaders say on the campaign trail often correlates only loosely with policies that are enacted. That’s not a comment on either of the candidates this time. It’s just how politics works. I think for the Chips Act, the key things for analyzing the politics are: First, it passed both houses with bipartisan support. Second, it retains bipartisan support in both chambers of Congress. Third, there are now a bunch of very influential companies that have a very strong stake in the Chips Act continuing. Fourth, because of that, there are now a fair number of influential representatives and senators, who, regardless of their views of the merits of the policy, and many of them are supportive on the merits, they also have a political incentive to keep the Chips Act funding flowing.

So I think all that suggests that the politics around the Chips Act will be supportive of its continuation. I don’t expect any major changes to the grants that have been issued with one caveat, which is the way the grants are issued is that companies have to meet milestones: you build the first part of your factory and you get a certain share of the money, and you have to build the next part. I think for some of the projects, I don’t doubt the milestones will be met. For other projects, companies are facing public difficulties, milestones might not be met, so that could be a trigger for negotiations over a specific grant, but I don’t think there’ll be a wholesale renegotiation of all the programs.

DSET:  The rest of the Chips Act related questions are: First, people here in Taiwan are worried that TSMC is planning to produce 2 nanometer chips in the U.S., probably earlier than people expected, and we heard about Intel might be influencing JD Vance the funding and the subsidy priority. If the U.S. is really serious about ramping up advancement capacity, then we should probably expect a Chips Act 2.0. With Trump in charge, then there’s more concern about maybe access to Washington matters more in the distribution. And of course, will the Trump administration be focused on capacity or having access to the most advanced? Because then this will contradict Taiwan’s official policy of leaving the most advanced technology in Taiwan.

Chris: I think in terms of how the big companies can relate to the government, what we’ve seen over the past 5 to 7 years is that all the big companies have been spending more time with political leaders. They all had photos with President Biden. I’m sure they’re all gonna try to get photos with President Trump. That’s how it works. Yes, President Trump’s style is different from Biden’s for sure. It’s still the case you gotta be friends with your Senator. That’s still the case. I think this is just the standard businesses trying to manage the relationship with the government, and I don’t expect a major change there. 

In terms of what the U.S. government is really largely prioritizing, my sense is that, again, there’s been some consistency over time, which is that they’ve wanted to be both pretty advanced in capacity and as close to the cutting edge as possible. But I think you’ve also seen the Trump administration, the 1st TSMC investment, and the Biden administration in the Chips Act deal that has just been announced, they are comfortable with N-1. It doesn’t mean that they don’t want N, but it is a recognition that N-1 is pretty valuable. If you get N-1 and ongoing capacity increases. That’s worth something. So I don’t know exactly what TSMC will be offering in the future in terms of capacity expansion. As they’ve already promised the 3rd fab, I wouldn’t be surprised to see a 4th fab at some point. 

The trend line I see is that TSMC is gonna have 70% of its capacity in Taiwan and 30% distributed between the U.S., Japan, Germany. And the U.S., may be the primary location of additional non-Taiwan capacity expansions going forward. And that basic framework probably continues under the Trump administration.

Huawei and export control

DSET: Let’s move to Huawei and export control. We’re wondering about your understanding of the incident that Huawei was acquiring TSMC’s chips. Who do you think should be responsible for this? How do you suggest improving it? You have suggested in another interview that the financial penalties can increase as we saw in other financial sanction regions. May you elaborate on that?

Chris: The the best case interpretation is that it was a really significant failure of compliance efforts at TSMC, so I think that alone implies that TSMC has to do a lot more on the compliance front, which probably means investing more in making sure that the right number of people and the right systems are in place to detect this, and to make sure it doesn’t happen again. 

I would say, to some degree in TSMC’s defense, I think they’re not alone in under-investing in compliance. I think that’s probably something that most of the industry faces, which doesn’t mean that it’s an excuse, but it is a reality. As you were referencing, part of the reason for that is that unlike in other sectors like the financial sector, where there are really big fines that banks pay not infrequently for compliance violations, in the chip industry that hasn’t been the case. 

So my view is that governments should impose larger fines, not because we want to impose larger fines, but because if you do so, you will provide a strong incentive to put in place the types of compliance operations that will stop this from happening in the first place, and then there won’t be fines. It’s both the direction of travel, but also something we should be pushing for. It’s in the industry’s overall interest to have a more effective compliance framework. Because if you don’t have an effective compliance framework, then the worst behavior is benefitted. That’s why the entire industry ought to be pushing for tougher penalties. And pushing simultaneously for more investments across the industry and the types of tools and the types of methodologies that could prevent this from happening in the future. That is something that I wouldn’t be surprised to see in the next couple of years, some serious efforts on that front.

DSET: Also, it seems that TSMC doesn’t have the capacity or ability to really know where their clients are sending their chips to. The Minister of the Economic Affairs of Taiwan then responded to the parliament that the government might require TSMC to sign a contract that they can make sure their clients are sending to. That’s the solution that Taiwan’s  government proposed. What’s your comment on that?

Chris: If you’re a firm selling a very simple power management semiconductor, that is totally commoditized, produced by multiple other firms, and sold via a complex network of distributors to every country in the world. I understand the compliance problem. 

If you’re one or two firms in the world that sells a certain caliber of chips, and there are only a handful of countries that are buying chips like this, our expectations about compliance are justifiably higher. Regulators probably haven’t done enough to really provide differentiated compliance frameworks for different parts of the chip ecosystem, and that might be a place where we ought to be focusing more. However, I do think for really high-end production, where the number of producers is low, the number of customers is low. That’s a place where we ought to have a lot more confidence where chips are actually ending up.

DSET: It’s confirmed that TSMC has held up all the shipment of chips under 7 nanometers to China. What’s your understanding of that? You have mentioned in other interviews that you wouldn’t be surprised if there is a blanket ban afterward.

Chris: It’s widely expected that there’s going to be a new round of export control updates released in the next couple of weeks. We’ll read what those say, and then have more insight into what exactly the rules will be. And, a blanket ban is one of the types of techniques that regulators can use to guarantee there’s a bit more confidence around compliance at the high-end of the spectrum.

DSET: The potential proposal is a blanket ban on 7 nanometers and below against Chinese customers, but we also see discussions of people trying to distinguish between end use with AI, Edge AI, and automobile self-driving. 

Recently, we actually also saw BYD announce a 4 nanometer that would definitely be fabricated at TSMC, and possibly it will be in coordination with MediaTek. They work with MediaTek to do some automotive self-driving chips, that is potentially not AI, but we do see there could be more dual-use applications with the self-driving technology. So this could potentially be banned as well, but it will also be another big source of revenue. Probably will expect a lot of pullings and tug of wars on this front between the industry and the government. We were told that industry comments, in the past, had not made any effect on the actual implication of the control.

Chris: I don’t know if that’s true. I think it would not be unreasonable for governments to impose different standards on a 7 or a 4 nanometer chip that are used for different use cases, I think a smartphone chip versus an AI training chip are different. If the industry goes to government regulators and says, we have no idea what the chips we produced are being used for. I think the industry should expect that the government will say: all right, we’re gonna ban broadly.  But if industry goes to the government and says we can develop capabilities to understand which chip is which. Then I think they can expect a more targeted policy. 

My sense is that many industry players, not just TSMC but also lots of U.S. firms and other firms, have often in the past argued to the Commerce Department that there’s a whole lot of things they don’t know when it comes to export control enforcement, and I think, insofar as they keep arguing that there are many things that they don’t know, they should expect broader policies, because the government will conclude targeting is impossible. Insofar as companies are able to share more information with the government. I think they can expect policies that are more targeted because the government will know more and have more faith in targeted policies. 

I think there were some players in industry who thought it was in their interest to emphasize the gaps in their knowledge in their conversations with the government. I think the impact of  saying “we don’t know where our chips are going” is to push the government towards more broadly based restrictions, because the government concludes that targeting carefully is very difficult. 

DSET: Regarding the current Huawei incident or related to TSMC, it seems that the U.S. government is taking action unilaterally, at least in the Taiwan scenario. Would you expect that in the Trump administration, they would continue to do unilateral, try multilateral, or this would not even be the agenda of the Trump administration. Would it require the Taiwanese government to expand their export control entity list, or even any kind of other effort to make it more multilateral?

Chris: The primary challenge to multilateral action has been that many of the governments involved have found it to be the path of least resistance, to accept unilateral action rather than to take steps to change their own export control regimes. We see that in the case of Japan, and we see the case of the Netherlands, both of which have taken steps on their own. but they’ve taken steps after they’ve been able to show  to the public that they’ve been pressured by the U.S. 

When it comes to other countries like Korea, like Germany, they would much prefer, both for domestic political reasons and for reasons of relations with China, to have unilateral U.S. action force them to take steps rather than to take steps of their own volition. Because they can go to domestic constituencies and say, “it wasn’t our idea.”  And they can go to China and say, “it wasn’t our idea.” 

hy isn’t there a multilateral regime today? Many people would hypothesize that it’s the U.S. desire for unilateral action. I think actually the primary reason you don’t have a multilateral regime today is that most of the other potential members prefer unilateral action relative to taking steps on their own. I don’t think that dynamics are going to change.

“Loophole”

DSET: You might have noticed that Financial Times recently had a report about China or Chinese entities having remote access to AI servers in the U.S., and it is one of the so-called loopholes in the current small-yard-high-fence AI export control regulations of the U.S. Do you have any comments on it? Or do you think the Biden administration will have a new control on it before he steps down?

Chris: It’s certainly an issue that’s controversial in Washington, there’s a number of people who think that ought to be prohibited. One of the questions around this debate is, can you impose regulations that would make this potential loophole insignificant in terms of implication. For example, if you were to guarantee that a Chinese firm accessing U.S. based cloud computing is only accessing below a certain threshold of compute. You could guarantee that it wasn’t actually training a significant AI system and therefore might not be problematic. In addition to banning it, there are also discussions of whether you could put enough constraints in place where you felt comfortable that there weren’t gonna be any problematic implications.

One of the counter arguments is, is it really a bad thing for Chinese firms to be paying not China based cloud computing providers, but U.S. based cloud computing providers. I think there’s something to that logic as well, and that’s one of the reasons why what’s often described as a loophole was deliberately left open. I think that will structure the debate in the next administration.

DSET: Chinese subsidiaries in the U.S. also need to be dealt with, such as TikTok and ByteDance, and one of them is an American company. So it’s just a lot of loopholes. They’re also talking about building up data centers in the Middle East, using their capacity. A lot of Chinese firms have been doing this in Singapore and Malaysia. I guess the end goal is actually to stop or to increase the gap between the American leading edge in use of AI models versus Chinese that maybe regulating chips. It’s not enough, maybe from some regulators perspective, it also has already imposed a big cost, if you look at Huawei’s recent progress with native chips. 

Chris: I think the Middle East actually is not that hard to control, so long as China’s an importer of AI chips from Nvidia, which it still is. They’re still buying lots of H-20 chips, China will not be an exporter of AI hardware to the Middle East. In 5 to 10 years, who knows, but right now, China is a net AI infrastructure importer, so when Middle Eastern countries say, we’re going to buy AI infrastructure from China, I don’t think that’s a plausible threat right now, given that all the key Chinese firms are importing infrastructure from Nvidia.

We’ve already seen the Commerce Department discuss rules to impose a compute threshold  for accessing U.S. cloud computing services above certain limits. So you were to ask how many cloud computing providers could provide computing power above that threshold, it’s basically going to be the Big 3 U.S. firms, and then maybe Ali Cloud in Southeast Asia at some point. I don’t know the answer to that question, but it’s a small number of providers. And then you will ask, what will the Chinese cloud providers do in Southeast Asia? What will their access to computing power look like in 3 years time? I think the U.S. is very much hoping that it’s constrained because China will be an importer of lower end Nvidia Chips, and it’ll be harder for those firms to buy cutting edge chips. I think that the aspiration of the Commerce Department is that Chinese cloud providers, and even in Southeast Asia, end up being compute constrained relative to Google, Microsoft, Amazon Cloud. I don’t know if anyone’s done a great analysis there, but that seems like a plausible outcome in a couple of years.

Investment screening

DSET: Do you see any improvement or even enhancement coming in inbound and outbound investment screening under trump 2.0?

Chris: On the one hand, CFIUS (Committee on Foreign Investment in the United States), which is for the inbound screening, was updated in 2018 via legislation, so that’s pretty recent. On the other hand, outbound investment screening rules were just finalized a couple of weeks ago. I think the perception is that both of those have now been updated, and probably don’t need updating in the short run. I would not expect major action on either of those fronts.

Human resource and immigration

DSET: Of course, we don’t know what Trump will do, but apparently, he may be more open towards “high-skilled immigration,” or foreign graduates. Everyone thinks it’s a cheap talk, but we’ll see. A practical issue is that part of the friction TSMC has had with the U.S. is workers. And TSMC definitely prefers to let more  Taiwanese graduates from U.S. universities be employed, let’s say. On the other hand, a lot of Americans are being recruited into TSMC as well. This is part of the reason TSMC believes it should expand to America, even without subsidies, because they think they need more talent, and the U.S. remains the best in that regard. So, do you think there might be any friction in this area, or would Trump be very strict about requiring companies to hire Americans only?

Chris: We’ve heard at various points in the campaign, including on podcasts with Silicon Valley venture capitalists, that Trump embraced the idea of increasing H-1B visas. We have regularly Republicans embraced that. But the reality is, you need the Congress with that, and passing immigration reform through Congress is very hard, which is why it hasn’t happened in 40 years. The last serious immigration reform bill was under Reagan. 

Given that precedent, I would guess that Congress likely won’t take any major steps on immigration. If Congress doesn’t act, there probably won’t be an increase in pathways for H-1B or other high-skilled immigration. Will Trump make statements saying he wants jobs for Americans? No doubt. Is it the reality that the chip industry has a super internationalized workforce? Yes, it is. Would chip companies like more scope to import labor without restrictions? Absolutely. They’d like it next year, they wish they had it last year, and they wish they had it 10 years ago.

It will likely be continuity with a system that is tolerable but not ideal for companies, and a system that won’t be improved by Congress because Congress won’t be able to agree on what should be changed. That will be my guess.

DSET: We would like to know about your opinion on the TSMC and their Arizona fab facility.

Chris: TSMC said publicly that yields in Arizona are the same as they are in Taiwan, which isn’t surprising because TSMC is a very capable company. That runs counter to much of the media narrative about delays. Yields are very important, and costs are also critical. Costs are still meaningfully higher in Arizona than in Taiwan. While the cost curve may come down somewhat, Arizona will remain more expensive than Taiwan for the foreseeable future, and everyone recognizes that.

The key inputs to the question of how much more capacity TSMC brings online will include the cost dynamic—when you build more capacity leads to economies of scale and reduced costs, customer demand, and government incentives. Customer demand for localized production isn’t likely to increase dramatically because customers have expressed a desire for a more diversified manufacturing base but haven’t shown much willingness to pay more for it. I don’t think that’s gonna change.

So the role of government incentives and pressure becomes crucial. The U.S. government has been focused on this issue for over half a decade and is likely to remain focused on it for the foreseeable future.

Advice 

DSET: If you were suggesting the Taiwan government do some preparation for Trump 2.0. in response to new tariffs, and probably more pressure, or the U.S. government’s efforts to track TSMC to invest more, what preparation would you suggest to the Taiwan government? 

Chris: I think, obviously, anyone who has  tariffs threatened against them is not going to like it. But the interesting thing to think through here is the electronic supply chain shift, what is the U.S. trying to accomplish and what is Taiwan trying to accomplish, where is their overlap, and I think there’s a lot of overlap. 

I think one of the interesting charts to look at is the export of servers by country by dollar value in the last 3 years. When you look at that chart, you’ll find:  3 years ago, China was by far the number one exporter of servers plus Mexico and Taiwan. Today, Taiwan is rocketing up towards number one all thanks to Nvidia and TSMC. That’s a really interesting shift. It’s worth the U.S. and Taiwan thinking collectively about what the smartphone supply chain will look like in 5 years. We’ve seen Apple make really big moves in India, and Google is now assembling phones in India as well. It  has huge implications for Taiwan, and not necessarily bad implications. Mixed depending which company you’re looking at, but implications that are worth kind of thinking through, because I think there are a fair amount of places of alignment between the U.S. and Taiwan on those types of issues. If there are more U.S. tariffs on China, that will encourage more shifts in the electronic supply chain. 

I think getting as much alignment as possible between like minded countries that are all dealing with China and the challenges it poses in different ways. There are opportunities there as well as challenges. I think the headlines are likely to focus 90% on the challenges and 10% on the opportunities. I’m interested in what are the opportunities that can emerge from this? Because it’s gonna happen whether we want it or not, there will be tariffs. The question is, can we find ways to make sure that the results of the shifts that follow are serving a kind of a strategic result rather than just kind of companies changing their assembly locations without much thought given to it. That’s actually a place for really interesting analysis for you and for me and others to do. I think for the U.S. and Taiwan to talk seriously about and think through where do we want things to be in 5 years, and how do we get there?