Earlier this month, the Biden administration unveiled a sweeping order aimed at limiting the export of chips and chip technology to China. It’s the latest step in building policy of constraining China’s tech and semiconductor ambitions. But just how severe are the restrictions? And what will be the effects? And why now? To learn more, we spoke with Fletcher School professor and author of the new book Chip War, Chris Miller. This transcript has been lightly edited for clarity.
Key insights from the pod:
How expected were the new chip restrictions? — 4:32
The history of US chip restrictions — 6:26
Chips vs. oil — 10:04
How Covid changed the chips conversation — 11:05
How the new chips restrictions changed the rules — 14:42
Will China retaliate? — 22:35
Can China catch up? — 25:45
How the restrictions effect foreign companies — 34:33
The goal of the CHIPS Act — 37:32
What an invasion of Taiwan would mean for Taiwan Semi — 40:02
---
Joe Weisenthal: (00:09)
Hello, and welcome to another episode of the Odd Lots podcast. I'm Joe Weisenthal.
Tracy Alloway: (00:15)
And I'm Tracy Alloway.
Joe: (00:17)
Tracy. Do you ever want to write a book one day?
Tracy: (00:18)
Yes. You know this, Joe. You know this because I try to make you write a book with me all the time.
Joe: (00:24)
I don't want to write a book. It seems like a lot of work.
Tracy: (00:26)
No, it'd be fun. Well, first of all, it is a lot of work. I wrote two book proposals last year and it nearly killed me. That's a very long story for another time, but I think it would be fun. We should do an Odd Lots book. Come on.
Joe: (00:40)
Maybe. I could see an Odd Lots book, but here's what would worry me about writing a book, which is, you, we could have a book idea today in 2022. Maybe we get a deal sometime in 2023, and then they're like, it's gonna come out in fall 2026. Congrats! It's just too long. And then, here's the thing that I would worry about, is that by the time, like, August, 2026 comes around, maybe no one at all cares about what we want to say in our book.
Tracy: (01:08)
All I'm going to say in response to that is how many GameStop books came out in 2021? Like there were at least three and they all came out within a year of it actually happening.
Joe: (01:18)
Yeah! and by then, no one cared anymore. That's my point.
Tracy: (01:21)
Okay. I take the point that it's very difficult to write a well-researched book that says something new and interesting on a topic at exactly the right time ahead of it kind of blowing up in the public consciousness. That is difficult.
Joe: (01:35)
But our guest today has done that. We are going to be speaking to a guest who has a new book out. And I think the timing like, could not have been better. I don't even know how the timing was possible.
Tracy: (01:44)
So this is actually a book on a topic that we, first of all we should say that we have had this guest on before talking about a topic that kind of became core Odd Lots content and also a core issue for the rest of the world. It truly exploded into the public's consciousness in a way that I don't think any of us weren't necessarily expecting when we first started talking about this. I think that was back in like late 2019, and then Covid happened and everyone became focused on this particular issue or thing. And then fast forward a couple years, and it also became an issue of strategic and even military importance.
Joe: (02:24)
That's right. So we've done semiconductor episodes in the past. We've done several of them, and we sort of tried to trace the semiconductor supply chain. And one of the companies at the end of the semiconductor supply chain that's very important is the Dutch advanced lithography maker, ASML, which supplies equipment. And so we're like, Okay, who should we talk to about ASML? And everyone said, ‘You gotta talk to this guy, Chris Miller, who knows all about them.’ And at the time he was writing this book and the book has just come out. It's out. And of course the timing is perfect. So Chris Miller, he is the author of “Chip War, The Fight for the World's Most Critical Technology.” And of course we just had the Biden administration announcing new restrictions on semiconductor technology export to China, which is something we briefly touched on in a recent episode with Dan Wang.
Tracy: (03:12)
It is truly the perfect guest with the perfect timing.
Joe: (03:15)
The perfect guest with the perfect timing, I'm thrilled that we have Chris back on the show to talk about these new actions and how they fit in the broader context and his research. So Chris Miller, thank you so much for coming back on the podcast.
Chris Miller: (03:27)
Thanks for having me back.
Joe: (03:29)
Congratulations. Just incredible timing because I mean, the book just came out, right? And right then we have everybody talking about the Chip War, basically because of these actions by the Biden administration.
Chris: (03:42)
Well, that's right. The timing of the regulations that were released a couple weeks ago were a bit of a surprise to me. But I think the trajectory was pretty clear from the last time we spoke about two years ago, that we're going to see more tightening of controls and more bifurcation between the US and China developing separate semiconductor ecosystems.
Tracy: (04:01)
Wait, was the trajectory clear? Because, you know, my impression was there was a lot of noise during the Trump administration about technology security and these types of issues, but no one was necessarily expecting the Biden administration to take this up to the extent that it did. So the export controls that were announced recently, well, maybe you can tell us how extreme are they in your view, but do you think it was expected? Were you expecting something along those lines?
Chris: (04:32)
I think so. I mean, if you look at the Trump administration's actions, I think you've got to differentiate a bit between sort of what the Twitter feed was showing on any individual day and the direction of travel of the national security bureaucracy underneath it. And I think the bureaucracy was trending in the same direction, actually from the late Obama administration in terms of greater skepticism of inbound investment into US semiconductor firms from China, in terms of greater concern about the difficulty of preventing chips from reaching military end users in China, greater confidence that computing power and semiconductors would be critical to the future of military power, and thus concern that China's access to US chip technologies would have a military benefit. All that has been developing for really about a decade in Washington and is less driven by the question of who's president and, and more the collective assessment of the Pentagon and the Commerce Department and the National Security Councils involved. And that's been a pretty steady intensification, even though the politicians at the top have differed.
Joe: (05:37)
I'm just looking again at the, you know, your book was released October 4th and the restrictions I think were announced October 7th or right around there, so basically the same day. But you know, before we get again to these specific restrictions, you know, I think when we talked about ASML in our earlier episode, even at that time ASML I believe was restricted from exporting its most advanced technology to Chinese companies, and maybe they were still able to export some technologies, but not the most advanced lithography. You mentioned some of the factors driving concerns about constraining Chinese semiconductors, but where did this really start, the story of the US deciding that limits on technology transfer would be part of our foreign policy with China?
Chris: (06:26)
I would actually start the story before China was the focus because there have been limitations on semiconductor technology transfer since the invention of the first chips. And if you think back to where chips came from, they emerged out of demands to miniaturize computing power to fit in guidance computers on missiles during the Cold War.
And so there's been a deep interrelationship from the earliest days between defense technology and the chip industry. And so controls have been part of the US government’s strategy since that time. So during the Cold War with the Soviet Union, there were really extensive controls placed both on the transfer of chips to the Soviet Union and also the transfer of chipmaking equipment. And so in some ways, what we're seeing today is just a continuation of that process of trying to control technology transfer that's militarily relevant.
Now it's different because the target is different this time, it's China and it's different because unlike in the earliest days of the chip industry in the late 1950s and 1960s, at that point, most chips were going towards some sort of military use, whereas today, 98 or so percent of chips are going towards civilian uses. So that's different.
But the basic concept that you want to control the ability to produce advanced chips because they have military uses is basically something we've always had. Now certainly there's been a shift in the policy towards China over the past couple of years, driven by a couple of factors. I think one of the most important is that the military balance in general between the US and China has shifted, whereas a decade or two ago it was clear the US military was far stronger. Today, that's less clear, especially when you look at potential hotspots like the Taiwan straits. And so it's easy to find analysts who say the US would have a hard time defending Taiwan, which wouldn't have been the case a decade or two ago. So that's one more concern about the Chinese military's capabilities.
Two is that China plays a bigger role than ever before in the global electronics and computing supply chain though still a circumscribed role. And we can talk about the specifics, but nevertheless, its relative weight is more important than certainly a decade or two ago. And then third, China's trying to catch up technologically in advanced semiconductors and through a number of industrial policy programs, the most famous of one, which is probably Made in China 2025, it's specifically targeting its chip industry with really substantial subsidies. And so this has raised concern not only in the US but particularly in the US that China might succeed in reaching the cutting edge in a number of key technologies, which would have both technological and economic, but also potentially military implications down the road.
Tracy: (09:01)
Right. So this is actually something we were speaking with Dan Wang about, the Gavekal tech analyst who also wrote a little blurb for your book, so it's nice to see multiple Odd Lots guests come together in that way. But can I ask just a very general step back question. I think it's become really common nowadays to say that semiconductors are the new oil in the sense that they're, you know, militarily and strategically significant. But are they really like oil in the sense that, you know, they don't necessarily seem bounded by geography. I mean, it's not like we're trying to pull out a finite resource from the ground. We're trying to build up expertise and technological capability to make additional chips available. And you know, there are some fabs in the US, not a lot, but the CHIPS Act is all about building more of these chips factories. So I guess, what are the parallels with oil and what are the key differences?
Chris: (10:04)
Yeah, I think you're right that there are obviously differences with oil. The geographic distribution is is one difference though actually I think oil is more geographically distributed than chipmaking capabilities, especially at the leading edge. And if you look at Saudi Arabia, it produces 10% or 15% of the world's oil supply. By contrast, Taiwan produces over a third of the world's processor chips every year. And 90%, if you're talking about the most advanced processor chips, ASML has got, you know, a far bigger role in the lithography production process, a hundred percent at the leading edge of UV lithography compared to what everyone else has in the oil industry. So in terms of concentration, we actually have more concentration at multiple parts of the chip supply chain than Saudi Arabia or even OPEC as a whole has. So if you're worried about the risks of supply cutoffs, there's more reason to be concerned about semiconductors, I think, than reason for concern about oil.
Tracy: (11:16)
So this is something that you touch on in your book, but when Covid happened, there were plenty of businesses that were disrupted by the pandemic, and semiconductors became a big issue around that time. And we can argue about whether or not there were semiconductor shortages because of the chips makers themselves or because the companies putting in orders massively underestimated demand and things like that. But what role did the pandemic play in, I guess, highlighting these shortage issues that you just described? Or maybe shortage isn't the right word, it's uneven distribution, I guess.
Chris: (11:55)
Yeah, no, I think that's the right way to describe it. And you know, the shortage context has to be understood in that supply increased dramatically in terms of aggregate chips produced in 2020 and 2021. It's just that demand grew even faster and that created this set of shortages in certain subsets. I think if you actually look at the track record in both of those pandemic years of the global chip industry in producing chips, you'd have to say it sailed through the crisis with comparatively few disruptions. The fact that supply increased dramatically across the board, the fact that, you know, we had major economies that played a big part in the chip supply chain being shut down for several months if not longer, across Covid, and yet the number of disruptions were actually quite limited, suggests the extent to which the supply chain actually works pretty well in, in producing the chips that are needed.
And when you look at where we've seen the most disruption in terms of shortages, which is the auto sector, I think there's a pretty good argument that what we're seeing is not a chip shortage, but an automaker inability to properly plan for the number of chips that they need, which is the problem that they faced. Automakers made N error in early 2020 of slashing their chip sales just as demand for smartphones and PCs, ETC., was booming because of the work from home trend. And as a result, when they came back to their chip suppliers a couple months later and said, ‘Actually, we're not gonna need to cut demand as much as we expected,’ they found that capacity was already reallocated. So this is really more a question of how automakers manage their procurement and their just in time procurement processes than anything to do with chips per se.
And if you look at specific disruptions to the chip making process in 2020 and 2021, you'll find that factory fires or ice storms in Texas were probably just as impactful as any sort of Covid-related disruption. But, you know, all that is to say that certainly Covid put semiconductor shortages in the headlines in a way that catalyzed politicians to take the issue more seriously and to act. And I don't think we would've seen the Chips and Science Act pass in the way that it did had it not been for the fact that Covid made this kind of visceral and seeming to be relevant to the typical person.
Joe: (14:03)
By the way, speaking of all the people who have glowingly blurbed your book I see Admiral Staviridis also is one another Odd Lots guest, so I'm glad to see the Odd Lots crew so represented. All right, let's get to the new news. And of course, as we discussed there, there were already restrictions on what kind of technology could be exported to China, but why don't you sort of summarize what was the status quo in, say, September? Where were we a month ago, and how dramatically have things changed with the new announcements from the Commerce Department and the Biden administration on exports? Like, just summarize, give us the big picture differences of then versus now.
Chris: (14:42)
So one month ago, there were, I'd say two main categories of restrictions that applied to China. First was for equipment at the most cutting edge, so UV lithography machines being the best example of this that were impossible to transfer to China. And this really only referred to a small set of tools that only restricted China's ability to move to the most advanced process nodes in producing logic chips and certain types of memory chips as well. But it was the most cutting edge set of machinery. That was the first set of restrictions. And then second, there were restrictions on specific Chinese firms that were restricted either from accessing US technology or also in some cases accessing production capacity in Taiwan at TSMC that used US technology. And so, Huawei's the best example of this, but there were also a bunch of Chinese firms that were assessed to have links to the Chinese military. And so these were the two sets of restrictions that were in place.
Today after the new set of regulations these have been broadened pretty substantially. So first off the set of equipment that's no longer possible to ship to China is expanded a lot. So it's no longer just the most cutting edge. It's also tools that are actually kind of one or two generations with three generations behind the cutting edge today. So in the logic chipmaking space, anything that's headed towards a facility that's 16 nanometers or below is no longer possible to send to China. So this is not the most cutting edge chip like you'd find in an iPhone processor, for example. But this is a much broader swath of chips. Now there's a whole lot of chips that you still can send equipment for to China, but there's a broader definition of the cutting edge, which is designed to more tightly restrict China's ability to advance.
So that's one big change. The second is that there's a new set of restrictions on the ability to send certain chips to China, and this focuses on chips called GPUs, which are used in artificial intelligence applications and data centers. And alongside this, there are restrictions on sending chips to Chinese supercomputers. So this set of the restrictions focuses on limiting China's ability to develop capabilities and artificial intelligence and advanced data centers. So this is a new focus area. And then three, there's a set of restrictions on US persons, so citizens, green card holders, and US companies on interacting with a bunch of different firms in the Chinese chip industry. And this is new. Before there weren't any restrictions on individuals in such a broad fashion working in the Chinese chip industry. And this is gonna have major effects because there are actually a lot of dual citizens who are working in Chinese chip industry, including CEOs of some of the biggest firms in China.
And so this, this is a new escalation of the controls designed again to take apart, decouple the US and the Chinese chip industry at the cutting edge. But all of this is to say that there's still no restrictions on anything that's not cutting edge for the most part. And so the goal of the regulations is to be fairly targeted in saying we're not going to allow any progress on the cutting edge, but we're still leading all of the lagging edge, mostly unrestricted. So it's still targeted, but it's pretty sweeping in the sense that it's saying no more progress towards advanced nodes.
Tracy: (17:40)
So that third point about Americans or green card holders no longer being able to do that kind of work in China. That's exactly what I wanted to ask you about, because this seems to be, so far, one of the things that is getting a lot of attention, one of the most extreme things, and we've already seen at least one semiconductor company talk about how it's having to like send people home or at least cut off their access at work. But first of all, how exactly does this restriction work in practice? And then secondly, I have seen some people talk about an expectation that maybe some of these companies are going to get some type of exemption. What's the possibility that that happens? And that some of the extreme moves that have been outlined in these export controls are actually not necessarily rolled back, but they're sort of softened through these types of exemptions.
Chris: (18:40)
Well, the Commerce Department in their announcement did clarify where they were likely to give exemptions and not likely to give exemptions. And so they were pretty clear that for foreign chip firms operating in China, and again, it depends on the specifics, but there's scope for receiving exemptions. So companies like TSMC or some of the big South Korean chip makers have facilities in China. And the goal of these restrictions is not to impact their production. So there will be exemptions for their ability to acquire certain types of machinery, things like that. But across most of the other restrictions, the Commerce Department has what's called a presumption of denial, which means that you can apply for license, but the presumption is that it will be denied. And again, there's, you know, a lot of nuance in this, but I think broadly we should treat this as something that's not a bureaucratic hiccup.
And in fact, most of the places where the Commerce Department has said there will be a presumption of denial will be a place where there's lots of licenses that are denied and probably people won't apply for licenses in the first place. I think we shouldn't assume that these are regulations that are accidentally overly broad. I think the Biden administration put two years of work into this, and I think they have a pretty clear sense of what they're trying to accomplish. And my guess is that looking at the impact in the first couple of days after the regulations were announced, they're more or less seeing results that they had anticipated.
Joe: (19:55)
So what's it going to do the sort, the first order effect. I mean, long term maybe China will redouble or accelerate its effort at building out its domestic capabilities, but what is the effect going to be tomorrow or in the short term on Chinese tech?
Chris: (20:10)
The short term effect is that China's gonna buy a lot less equipment from US suppliers, but also as a result from Japanese and Dutch suppliers. Just because there's no point in buying half a production line so that overall equipment sales will be decreasing. Now, again, it's not across the board because for lower tech production, that's still possible to acquire, but there will be an impact on equipment firms, and you've seen that reflected in their public statements. That's one.
Two, there are a number of Chinese firms that were close to the cutting edge that will be directly impacted by this. So one example is the Chinese NAND memory producer called YMTC, which was in the process process of having its chips qualified for use in Apple’s iPhones. And Apple was reportedly going to buy YMTC’ chips for a large chunk of its iPhones going forward. And that firm was targeted by these regulations. And so now according to media reports, Apple's no longer gonna be buying YMTC'S chips for its iPhones. And so for firms that are directly impacted like that, because they're right at the cusp of being at the cutting edge and therefore directly impacted by these regulations, we're going to see their ability to produce and their ability to expand directly impacted.
And then third, we're going to see disruption of of individuals’ jobs and livelihoods because of this. Tracy mentioned the restrictions on individuals. That's gonna have a pretty dramatic effect because there will be people in positions of having to choose between continuing their work or potentially, you know, renouncing citizenship or green cards. And that's a dramatic effect. Now, you know, this isn't new because there's lots of companies around the world that Americans are barred from doing business with, but I don't think many people who are impacted by this were expecting this type of regulation. And this is probably the part of the rules that have most surprised people in the industry.
Tracy: (22:08)
You know, I asked Dan Wang this same question, but I would love to hear your response as well. But what kind of reaction would you expect from China here? So we've already had some statements from various officials talking about how it's going to endanger China's economic recovery and how it's generally a very bad thing for the economy. But would you expect any concrete measures from China after these restrictions?
Chris: (22:36)
Yeah, it's an interesting question because on the one hand, you know, I think you'd hypothesize that China has to do something in retaliation. On the other hand, if you look at the track record of China's responses to prior iterations of semiconductor export controls, China hasn't done much. When the US put Huawei on the entity list and thereby restricted the transfer of a lot of US chips and chip technologies to Huawei, China issued some angry statements, but ended up doing nothing despite the fact that these restrictions hobbled one of China's biggest tech firms. Which was, I think, a pretty striking demonstration of where escalation dominance in this sphere, if you will, stands — that China felt like actually any sort of retaliation would leave it worse off than not retaliating. Now this is an order of magnitude more impactful than the Huawei restrictions, although it's in some ways a bit more opaque since it doesn't target a single firm and it’s sort of hidden in the specifics of Commerce Department regulations, which most people don't understand.
But I would think China would feel more pressure to retaliate this time. But if you ask yourself, what's the best way for China to retaliate in a way that imposes more costs on the US than it does in China, it's not obvious that there's a great option there. I mean, China's still deeply dependent on importing chips across the board, and China's economy, of course, is as weak as it's been in many years. So this isn't a time when Xi Jinping wants to probably impose further cost on the Chinese economy. And in particular, if you're a Chinese electronics firm or firm in the semiconductor space, you know, you're desperate right now to keep the connections that you do have open. Because if there's any way you're gonna get around the current regulations, it's going to be by whatever collaboration you have still functioning with leading firms in other countries. And so keeping those lines of communication and cooperation open with firms from other countries is really crucial, I think, for China's hopes of finding some way to counter these rules. So I'm not sure actually if China's going to take a dramatic retaliatory step even though I think there's obviously a political impetus in China to do something and to look like you're not just taking these restrictions lying down.
Joe: (24:40)
So, you know, it's easy enough to say, ‘Okay, well this is gonna hurt China in the short term, but in the long term they're only going to accelerate their own work on building out an advanced domestic semiconductor industry,’ which they might. But like, that seems still really hard, even if you have the entire sort of force and money of the government. And I'm just thinking back to when we talked before and one of the details that I remember from our conversation was like, ASML itself is a customer of a company that makes the smoothest substance on earth. Because that's important, I guess like for the mirrors to like shoot the light or whatever. And you know, it does not seem inevitable that China can catch up. I mean, we know the challenge that Intel has with catching up to Taiwan Semi etc., and without the sort of like best expertise of this very globally integrated supply chain, how do you assess China's chances of becoming a tech leader in the space with so many restrictions?
Chris: (25:45)
Well, I think it's gonna be really hard. Is catch up possible? I think it's certainly possible over some timeframe with some amount of expenditure, but I think the timeframe is long and the expenditure is going to be large. And so China might decide that it's not willing to to pay the price. And if you look just at the space of semiconductor manufacturing equipment, which is in in some ways the space on which all these other restrictions depend, the reality is that the number of companies in this space is very limited. Most of the key firms have been in their market position for several decades, and in some cases almost half a century. And their supply chains, as you mentioned, involves sourcing some of the most complex and heavily engineered products ever made. And so it's just going to be very difficult to do.
So in addition, we're seeing more and more the US government targeting China's firms in this space. So there's part of the restrictions that were rolled out recently limit the transfer of component parts to Chinese equipment manufacturers, which is again, deliberately designed to make their process of advancement harder. And if you think about what it takes to make a machine tool in this space, you know, it's one thing to make a machine tool that can do a specific process once or even a dozen times or a hundred times, but to make a commercially viable machine tool in the chip industry, you need a machine that not only is able to move materials at almost the atomic level, but it also needs to do so with, you know, basically perfect runtime, perfect accuracy and to be cost competitive. And that's something you can only really develop, I think when you're closely collaborating with the firms that are gonna be using your equipment.
And so if you look at how innovation in the machine tool part of the chip industry happens right now, it happens in really deep collaboration with the customers, TSMC, Intel, Samsung, etc., because the tools have to be optimized for those customers. And so if you're in China today trying to develop your own machine tools, you're working with firms that have lower volumes, especially at the cutting edge, they're actually not producing the most cutting edge chips. And that's a harder environment in which to produce machine tools that are gonna compete at all with the cutting edge you can acquire abroad. So if you've got this closed loop innovation system where everything is second or third best compared to what you can get in Taiwan or what you can get in Netherlands, it's just a really hard environment to catch up in.
Tracy: (28:03)
So this actually leads nicely into something I wanted to ask you about because of course we've been very focused on the US efforts to restrict cutting edge chip technology into China, but the other plank of US strategy is to build up its own self-sufficiency when it comes to semiconductors. And in the book you mentioned that there's growing discussion among US officials about potentially using export controls on chip design software and manufacturing equipment that comes from the US. So a lot of the chips, as we've discussed on this podcast numerous times, they're designed in the US and then those designs are shipped out to Taiwan to be actually manufactured or whatever. And so if you were to impose export controls on the designs or the software that could in effect, I guess, pressure Taiwan to roll out more technology into the US and Europe or allow the US and Europe to build up their own expertise. Do you think that's the next step in these export controls? Is that something that you see on the horizon?
Chris: (29:08)
Well, I'm not aware of it having been discussed at the government to government level. It is something that you can find written about in think tank policy papers and hear discussed to some extent in in Washington DC. I don't know if the US government is going to go to that level. And I think that would be a pretty dramatic step. You know, it's one thing to impose export controls on countries that you see as primarily competitors, something else to impose them on country countries that you see primarily as as a partner. And so I think there would be some reticence in the US to use that type of coercive strategy vis-a-vis Taiwan. But I think it is true that everyone in the industry knows that the US does have the ability to impose these controls. And indeed the US already has imposed controls on China in a way that forces Taiwanese firms to take certain steps they wouldn't have otherwise taken vis-a-vis Chinese customers.
And so the fact that the US has this threat, I think probably does influence both how US government officials and Taiwanese government officials think about the dynamic. But if you look right now at what Taiwanese firms, and really that just means TSMC, has planned in the United States in terms of opening new facilities, they're opening one facility in Arizona, it will be relatively small in terms of its capacity and most of TSMC'S capital expenditure going forward is still going to take place in Taiwan. Even if you include the Arizona facility, the new facility in Japan, potential facilities in Singapore and Europe that are being discussed, we're still talking about a company that's primarily focused in Taiwan, especially when it comes to its most advanced production nodes. So right now I don't think TSMC is planning to dramatically diversify away from Taiwan. Nor do I think it feels really substantial pressure to do so. A little bit of pressure, but not substantial enough pressure to make it change its business model.
Joe: (30:57)
By the way, when you were talking about the challenge of getting the equipment to run perfectly each time I remember our episode we did, we talked to Willie Shih at HBS and like the math is just mind boggling, like 99.99% correct execution. This is like not anywhere near sufficient for how like many steps on how right everything has to go. But I want to pivot to the CHIPS Act and you know Tracy mentioned that yes, part of strategy right now is constraining China's ambitions, but also there's been, you know, legislation that moved forward in the US, a law has passed designed to bolster US capacity. What do you see, like, how would you describe just sort of big picture the significance of that act, and how much do you think that's gonna move the dial in terms of sort of bringing, setting TSMC aside, bringing onshore to the US more capacity?
Chris: (31:49)
So I think the CHIPS Act is the biggest move in semiconductor policy since the 1980s that we've seen from the US government. In the 1980s, there was a trade war with Japan, which was at the time a big chip maker and the US imposed a bunch of, or threatens to impose a bunch of tariffs on Japanese chip imports until the Japanese agreed to limit their semiconductor exports. And that was a big focus of policy at the time. And then for about four decades, or at least three decades, US policymakers more or less ignored semiconductors as a policy issue, in part because US firms appeared to dominate them. And that's obviously changed.
The Chip and Science Act will do two main things. First, it allocates $39 billion to subsidizing the manufacturer of chips in the US, mostly focused on leading edge chips, but with some funds for lagging edge and then second devotes on $12 billion to R&D efforts in chips. And so on the manufacturing side, I think there's no doubt that the subsidy funds will result in more manufacturing facilities being built in the US. That's sort of a simple economic calculation and I think we've already seen many of the firms that would be impacted — Samsung, TSMC, Intel, Micron already in the process or in the announcement stage of opening new facilities.
So there will certainly be an impact on the amount of chip making in the US. Is it going to dramatically change the status quo? You know, it depends what you mean by that. But right now the US produces 10 or 15% of chips. We're not gonna see that go up to 50% or anything near that level anytime soon. But we will have more leading edge chip making in the US than we otherwise would've. I think that's pretty clear.
I think the more interesting part actually is the R&D funds, because my guess is that when we look back in 10 years time and ask what was the biggest impact of the CHIPS Act, it will actually not be the subsidization of manufacturing, but actually the R&D. Because the amount of money that's going into chips R&D is really substantial. There's always been some via the National Science Foundation and some via DARPA, but $12 billion is a big influx over a fairly short period of time. And so I think this actually has the chance to have a really substantial impact over the coming decade.
Tracy: (33:53)
So just listening to this and the amount of money and effort that the US is going to put into building out its own chips capacity, is the US actually aligned with some of its allies such as Taiwan, potentially Europe, when it comes to its plans for chips? Because it feels like there's a tension here if the US has to balance between saying ‘we're going to build out our own capability’ versus ‘we're going to take market share away from TSMC,’ which is a huge company in Taiwan, of course, which is one of the US' major military allies in the Asia region. That does seem like a bit of a balancing act.
Chris: (34:33)
Yeah, it is, and especially when you set that next to the export control side as well, which imposes, you know, interesting differential effects on different firms. Some negative, some positive depending on which control you're looking at. And so, I think you do need to look at the CHIPS Act funding, which applies to both US firms and for non-US firms building facilities in the US alongside the export controls as part of a unified strategy to coerce and attract the world's chip firms more into the US ecosystem and less into the Chinese ecosystem. In terms of advanced fabrication subsidies, I think there certainly is a bit of a zero sum dynamic between the US and Taiwan, and the US and Korea in the following sense that both TSMC and Samsung prefer to have most of their capacity in their home country and they prefer to have most, not all, of their R&D in their home country.
And they prefer that both for economic reasons that their businesses work quite well in the current setup and also, I think, for political reasons. And so to the extent that the US is pushing for more investment in the US, there is a bit of a zero-sum dynamic. Now the US is not just pushing, it's also subsidizing. And so that helps to compensate them to some extent. But I think you're right, Tracy, to suggest that TSMC wouldn't have been pushing for any of these changes if it were in charge. It's responding to shifts that governments are imposing on semiconductor industry more generally.
In terms of other subsidy programs from Japan, Singapore, from Europe, I think it depends who you're looking at. The Japanese, for example, have subsidized TSMC to build a facility in Japan, but it's not focusing on leading edge, it's focusing on a more of a lagging edge node that is particularly relevant for Sony, which is a big customer of TSMC’s.
And so that doesn't really conflict much at all with what the US is planning to spend money on with Europe. Europe's own chips bill is still a work in progress and I think we're gonna have to wait and see what actually Europe decides to spend its money on. I worry most actually about potential market inefficiencies created by Europe's chip subsidies since they remain quite unclear as to where they're headed. I think it's harder to imagine that Europe will succeed at leading edge manufacturing simply because it has less infrastructure even than the US, but if Europe puts its money towards lagging edge manufacturing, that's an industry that already is getting plenty of subsidies from around the world, including from China. So I think there are questions to be asked about how Europe actually plans to spend the 50 billion euros that it intends to devote to the industry.
Joe: (36:59)
So you mentioned that, okay, the US is probably not gonna dramatically jump over the next several years in its share of a global semiconductor manufacturing, but it presumably the goal is for it to increase a bit. In 10 years or however many years, what would we have to see to say the CHIPS Act was a success or what would constitute success? I mean, again, and even R&D can sometimes be ambiguous what that turns into in terms of commercialized products. So from your perspective, what would constitute, ‘Okay, this bill actually achieved its aim.’
Chris: (37:32)
I think part of the way to conceptualize this bill is as an insurance policy and it's an insurance policy that is protecting us in case of some sort of escalation around Taiwan, which could disrupt chip production, and if so, it would be dramatically costly for the world economy. And so calculating the efficacy of an insurance policy is a hard thing to do, especially because ideally you don't have to to cash it in. But I think that is the right way to look at this. And again, it's not that we're going to go from large scale dependence in Taiwan to zero, but that on the margin we're going to become less dependent on the supply of chips crossing through the Taiwan Straight and that that reduces the risk in case of some sort of war. And it also reduces China's ability to use our reliance on Taiwan as a mechanism of pressure.
So I think that's the first thing. Is that there is sort of a national security insurance mechanism here that is hard to price in economic terms. The second thing on R&D, I think actually we should be able to pretty clearly know in a decade's time whether the R&D funding worked or it didn't work. And if we can't find clear evidence of pretty direct lines between spending and commercial successes, I think we should tell ourselves that it failed. And this is where I think there's the most work actually to be done at the Commerce Department and elsewhere in the US government, it’s that it's easy to dole out funds to, to subsidize advanced manufacturing, especially when the number of plausible firms that could be involved is really limited. And if you're gonna build an advanced logic facility, there's only three firms in the world that are gonna do it and the US government will probably give money to all three of them. So that job is relatively easy. The harder job is how do you set up and build out a R&D infrastructure that will fill some of the gaps that we currently have. And that's a much harder task, but I think in some ways an even more important one.
Tracy: (39:13)
So since you mentioned the idea of the US kind of hedging disruption in Taiwan, I have to ask a question that's getting a lot of attention at the moment. You know, loads of people on Twitter talk about theoretical situations and various possibilities here, but if there was some sort of invasion of Taiwan or some sort of military action against the country, how would you expect that to impact TSMC? And you know, one of the things that comes up in these discussions is the idea that if China did a full on invasion and they took over all the fabs in Taiwan, would they be able to just keep manufacturing semiconductors like they were before the invasion, or is there something particular about the chip making business that makes that scenario more difficult?
Chris: (40:05)
Well, I think if, if we have a sort of D-day style invasion, if you will, the likelihood that any semiconductor facilities in Taiwan remain intact after the fact is pretty low. We're talking about the most precise machinery ever made. Chip fabs are full of explosive gasses. It seems, you know, perfectly designed to not survive a war. And even if by some miracle TSMC facilities did survive a war, you'd still need all the personnel in place and in place in a way that they wanted to keep facilitating the production process rather than sabotaging it. And you'd still need supplies from abroad because TSMC relies on chemicals, for example, sourced from Japan and elsewhere. So I think it's implausible. Close to a 0% probability that China could invade Taiwan and we'd have a massive war and at the end, China would have perfectly functional chip making facilities.
Now whether Chinese leaders understand that, you know, I don't know. I don't know to what extent they've dug into the chip making production process. And I do worry a bit that there is a fair amount of discussion in Chinese social media, for example, and even among Chinese government aligned analysts about taking Taiwan to seize the chip fabs. But I think that idea is a fantasy, but I also think that when you envision military escalation scenarios, there's a lot that could happen between peace where we are right now and a D-day style invasion. And in those types of coercive measures or gray zone measures, if you will, in military lingo, that's where I think you could envision down the road China applying pressure, making demands on Taiwan. And if Taiwan doesn't feel like it has the backing of the United States, or it doesn't feel like the US could actually successfully come to its aid militarily, Taiwan might be pressured to, for example, change its regulations when it came to access to TSMC'S fabs. And so that's a more plausible scenario to me, which could still have dramatic implications for the United States in economic terms also in military terms.
Joe: (42:02)
Chris Miller, it was such a treat to have you back on and sort of again, extraordinary confluence of expertise, news and new book out. Appreciate you coming back on Odd Lots.
Chris: (42:13)
Well, thanks for having me back.
Tracy: (42:14)
Thanks so much, Chris. That was great.
Joe: (42:15)
That was fantastic.
Chris: (42:16)
Yep. Thanks guys.
Joe: (42:30)
Tracy, that was great. I mean Chris really was the perfect guest, but you know, just sort of like working backwards through some of the things that he discussed, you know, the implausibility of Taiwan semiconductor capacity holding up in like a sort of like true like hot war style invasion, I think that's an interesting point to start because again, you know, chip manufacturing is so complex, relies on such a global network, relies on so much sort of embedded expertise of the people who work there. It really is sort of hard to imagine that you could have that and the chip fabs just keep running.
Tracy: (43:05)
Right, this was sort of why I asked that question about the oil analogy at the beginning because I think, you know, lots of people look at it and they go,’Oh, it's a strategic resource. People use chips for military applications.’ There's actually some great anecdotes in Chris's book about how Russia's lack of chips is basically hindering its invasion of Ukraine right now. The temptation is to draw a direct parallel between oil and chips. But on the other hand, there is a key difference. You know, you can build up your own semiconductor capacity. In many respects it's far more difficult than, you know, just installing an oil rig to pull crude oil out of the ground. It does seem to take years and vast amounts of money, but there is that possibility. It's not bounded by geography in the same way that oil is.
Joe: (43:55)
You know, another thing that's really helpful talking to Chris is the sort of distinction, and both are important, but the distinction between leading edge and lagging tech and both are pretty crucial. I mean we started talking about chips, you know, one of the big themes, it was well, what's going on with the automakers? That's not advanced tech, right? A lot of that stuff was like the least advanced chips It was stuff that displayed an image on a dashboard or something like that about as basic as it gets. But if you don't have that, then you know, it doesn't matter that is leading edge or, or sorry, lagging edge or whatever, if you don't have it, you can't make the car. And so, you know, thinking about the CHIPS Act and as he put it, you know, this sort of insurance against this disruption, like, okay, maybe the subsidies won't necessarily bring the leading edge manufacturing or expand that, but expanding access to the lagging edge is also pretty crucial.
Tracy: (44:46)
You know, the ubiquity of chips in everything from cars to refrigerators and washing machines or whatever appliance, I do find that kind of concerning — the idea that you could use chips to hold the global economy hostage for a certain strategic or military goal in the same way that, you know, to some extent Russia has used oil and gas to do the same thing. And that worries me.
Joe: (45:12)
It does. You know, the other thing about that though is like, again, because there's so much interconnection and you brought up software, you know, I'm glad you brought that up because actually at some point we should do like a design software thing. Go back to the supply chain. But like everybody depends on everyone else, right? So in a way it's like hard to imagine like this sort of standoff without like really hurting yourself because it's so cross border. You mentioned like Japanese chemicals and software that might be designed in the United States, which I think the US is still really good at designing design software. And so that sort of makes the chips diplomacy tricky and unique in the sense that it seems really hard to like hurt your enemy or hurt another entity without harming yourself.
Tracy: (46:00)
Chips diplomacy is a good way of putting it. That should be the name of our book.
Joe: (46:10)
A knockoff version of “Chip War.” Tracy Alloway and Joe Weisenthal, “Chip Diplomacy.”
Tracy: (46:15)
All right. Shall we leave it there?
Joe: (46:25)
Let's leave it there.
You can follow Chris Miller on Twitter at @crmiller1.