Leslie Johns 0:03

Hi everyone, welcome today's webinar on Chip War: The Fight for the World's Most Critical Technology. Before we get started a few quick announcements. First, remember that as always, that this talk is webinar is being recorded both video and audio. However, we're only recording the speakers, which would be me as well as our speaker, Chris Miller. Those of you in the audience are not being recorded, we're respecting your privacy. The recordings will be available as both audio and video recordings through various podcast platforms. As you attend today's talk, and you if you have any questions about the book or for the author, please submit your questions via the Q&A portal. You can do that by pushing the button at the bottom of your screen called Q&A. I'll do my best as the moderator to get to as many of your Q&A questions as possible. But first, I want to go ahead and introduce today's author, Chris Miller. Chris Miller is an associate professor of history at Tufts University at The Fletcher School. He was trained, has received his PhD from Yale University. And he has a background in technology, geopolitics, economics, international affairs, and Russia. He's written three prior books on the history of politics and economics in Russia. So today's topic is a bit of an intellectual transition for him. Because a lot of today's book actually involves Asia, which is why one of our co sponsors today is the Center for Chinese Studies, I know a lot of our audience came from the Center for Chinese Studies. He's had a broad intellectual history, a lot of affiliations with organizations like the Carnegie Moscow Center, the German Marshall Fund, the Brookings Institutions, a lot of diverse institutions. But for now, I'm gonna go ahead and turn things over to Chris Miller. So he can go ahead and share with you his really interesting book, ChWar, which has been getting a lot of attention. So with that, I'll go ahead and throw things over to Chris.

Chris Miller 2:35

Great. Well, thank you very much, Leslie, for the introduction. And thanks to the entire team at UCLA for the invitation to join you all today. What I'd like to suggest to you in the next 20 or so minutes, is that you cannot understand the shape of the modern world, whether the military balance, the structure of globalization, or the development of technology, without putting the semiconductor at the center. And although most of us don't think very often about semiconductors, and I certainly didn't, before I started writing this book, the reality is that we don't know how to live without the thinking that semiconductors do for us at all. And the reality is that the typical person will touch hundreds, if not 1000s of semiconductors over the course of their daily lives even though we most of us don't ever see a semi conductor at all unless you pick apart the devices that you use. But today, all sorts of devices only work with the help of semiconductors or computer chips. They're more colloquially known. It's not just smartphones, or PCs, the types of devices that you think of as requiring computing power. Today, all sorts of household appliances or automobiles can't function without chips. When you wake up in the morning, your alarm clock has a semiconductor inside. You look at your phone, there are many semiconductors inside of a typical smartphone. If you get in that car, it will have dozens or many cases hundreds of semiconductors inside. Over the course of your daily life, you'll touch dozens of devices all of which have chips inside. And that's even before you start using the internet, which they function that thanks to data centers, which are nothing more than vast warehouses full of silicon chips. The modern world wouldn't work without chip. And yet it's only 60 years ago that the first chip was invented. Now in Texas and then California in the late 1950s. The first commercially available chip had four transistors on it and transistor the electric circuit that can turn on or off when it's been on. It's a critical one or off grade zero and these produce all of the ones and zeros that undergird all software and all digital computing from that first chip that had four transistors on it that the proportionally valuable chip today you go to an Apple store and buy a new iPhone, you'll acquire a phone with a primary processor chip that has 15 billion transistors on it. That's billion with a B, and the transition from four transistors to 15 billion, explained why today we take computing power almost for granted that even though it was only 60 years ago that building power was very expensive to acquire and to employ. The process of making chips today with 15 billion transistors on them is the most complex manufacturing process that humans have ever undertaken. It's only possible to produce a chip with 15 billion transistors on it because the transistors are extraordinarily small. The first chip that was commercially available heatron that you could see with your with a naked eye, whereas the day on a new smartphone, for example, the transistors will each be the size of a Coronavirus, measuring just several nanometers in dimension. Then, the process of manufacturing these requires a complex supply chain that stretches from Europe, United States to Asia, and requires a set of tools and software that has been honed for 60 years of precisely this purpose in mind, producing the most advanced chip by manufacturing miniature transistors by the millions and billions, and doing so with almost perfect accuracy. And what I'd like to do today is explain how it is that we have come to take for granted a world that wouldn't work without the billions of transistors that we all rely on, how these transistors are produced, and the international supply chain that produced them. And then why today like in the earliest days of the semiconductor industry, the process of producing chips is not simply a question of how do we get the most advanced gadget but also a question of the balance of military power because like in the early states and the chip industry today to semiconductors are critical for military systems, as well as for PCs and for smartphones. But first off, what does it take to create an advanced chip today? The earliest chip companies those that created the first chips and practice in Silicon Valley in the late 50s and early 1960s produced almost all of the tools and materials they needed to make chips in house. Texas Instruments, for example, which was one of the early leaders in chip making, produced their own silicon wafers. They refined their own chemicals in the house, they devise their own machine needed to manufacture chips. But today because chips require patterning, etching depositing materials with extraordinary precision, it's not possible for any company to do it alone. And indeed, it's not possible for any country to do it alone today either. If you want to make an advanced chip today of the type that you would find inside of a new smartphone or PC, you can only do it by accessing a single supply chain that runs across three continents. First off, you need computer software that is specially designed to allow the design of vast chips, which is produced predominantly by three countries, three companies, all based in United States. You take this software and you hire a team of very talented chip designers who lay out the map of where transistors should fall on each chip. And you need a lot of chip designers to do this. Well an advanced chip like the one on a smartphone or PC can cost half a billion dollars just to design it before you get into the manufacturing stuff. But the design itself is an extraordinarily difficult process. Once you've got a chip design using this ultra specialized software, you need to acquire the chemicals, the materials, and also the machine tools needed to produce a chip. The chemicals in chipmaking need to be ultra ultra purified because there's basically no margin for error and chipmaking when you're patterning shapes at the nanometer level and for the purification of chemicals and the refining of the materials that that are needed for chip makers of chip making is an extraordinarily difficult path. Only a small number of companies of which the manufacturers have managed to reach the level of purity are required - largely in Japan and also companies in the United States and Europe produce the the necessary materials, chemical. Once you've got the materials, you need to use a machine to actually undertake the manufacturing, and I think this is the most important part to realize about the chip making process. We think of the digital world as being somewhere ethereal in the cloud without really recognizing, reckoning with the fact that all of the devices we use only are able to compute because of manufacturing of silicon chips. You actually need to change the shape of the silicon by depositing materials or fetching materials off of it that produce the circuits that create all the ones and zeros undergirding your computing. And the machine tool needed to undertake this manufacturing process are simply put the most precise and complex machine tools ever made in human history. If you want to make an advanced chip, you need tools that are capable of depositing ultra thin films and material on a silicon chip, of etching canyons in the silicon that measured with a couple of nanometers wide, and of patterning the on the silicon, which creates the shapes that eventually become the transistors using a technique called lithography. And they carved efficiently accurate transistors which again are so small, they're far smaller than possible for human to see. You need to use a special type of light called extreme ultraviolet light with a wavelength of 13.5 nanometers because digital light is too large. The pattern with the specificity and precision needed for advanced chip visual light has a wavelength of several 100 nanometers, which is two orders of - at least one order of magnitude larger than is needed for advanced chip making. And to acquire the machine capable of producing extreme ultraviolet light in sufficient quantities for chip making requires tapping into a production process that has taken over three decades to hone. Each new extreme ultraviolet lithography tool costs over $100 million each. They take multiple 747 to transport and they're produced only by a single company based in the Netherlands which has a 100% market share in their production. So the machine tools needed to produce chips are extraordinarily complexto produce, which is why there there's an oligopoly of companies that produce them. There's basically five companies, three in California, one in the Netherlands, and one in Japan, which have a dominant role in the production of the machine tools needed to make chips. They've been in their market position for several decades, in some cases, almost half a century. And you simply cannot today make an advanced ship without these tools. Once you've got the amount of tools though, as well as the chemicals I've discussed in the design that are quite specialized themselves, you actually need to make chips and here too there is extraordinary specialization in the capability of chip makers. Today, the world's most advanced chip maker is a company called TSMC, that Taiwan semiconductor manufacturing company which is based in Taiwan. And on that small island, they produce 90% of the world's most advanced processor chips and putting the types of chips that go in smartphones, many PCs and also run advanced AI application in data centers. Without TSMC, which produces over 1/3 of the new computing power the world add each year, entire industries would simply grind to a halt because it's not only PCs or smartphones, the types of good do you think of it being high tech that require TSMC chips, it's also a whole array of lower tech goods like dishwashers and microwaves that also stores the semiconductor through TSMC. So it's arguably the most important company in the world because hardly any other company can operate well without the chip that TSMC produces. Now, the fact that yes, it is a company whose output is almost entirely on the island of Taiwan illustrates the extent to its geography really matters in the chip industry and although the past several decades has been lots of loose talk about globalization, the reality is that the chip industry shows just how inaccurate the term globalization is for describing many advanced industries because although the chip industry has a very international supply chain stretching from Europe where lithography machines are created, United States where many advanced chip designs are, are created, to Taiwan where many of the most advanced chips are produced, the reality is that the international supply chain is decidedly not a global supply chain. Most countries play almost no role in the production of not only advanced chip but on a chip whatsoever. And there are only single countries that can produce the necessary goods needed for producing advanced chip. So far from being globalized, we've got the exact opposite in the chip industry. The production of advanced logic chips has not globalized, it has Taiwan. The production of advanced lithography machine has not globalized, it is concentrated entirely In the Netherlands, and this fact creates interesting economic ramification companies that on these choke points have extraordinary pricing power, which makes them good businesses. But it also produces geopolitical ramification because I think it will come as no surprise to anyone in the audience today that when countries have positions of power in supply chain, they face a strong incentive to use them. And what we've seen over the past five years is that every country with a position of influence and the semiconductor supply chain and even several countries without position of a major influence in the semiconductor supply chain have tried to use the capability to achieve political goals. The most interesting example of it, although one of the least effective was earlier this year, when Russia, which used to be one of the major producers of neon gas, which is one of the many gases needed in filmmaking, tried to punish the West for sanctions over the war with Ukraine by cutting off supplies of neon gas exports to world markets and hoping that this would drive up the price of chipmaking. globally and in in, in the end, this didn't actually have much effect on chip makers. There's been a little bit of difficulty in sourcing enough gas, but it hasn't had an appreciable effect. And so this effort failed. But the fact that even Russia with the tiny chip industry, and plays a marginal role into the supply chain, has been weaponizing their semiconductor relevant industry, shows just how widespread the trend is of trying to use positions of power in the semiconductor supply chain for geopolitical purposes. But in fact, we should find nothing shocking in this at all, because the origins of computer chips is in defense demand. There the relationship between the desire to have more advanced weapons systems, and the development of the computing power that we today take for granted in our civilian lives. The first major order for semiconductors in the early 1960s was for the guidance computer in the Apollo spacecraft that took astronauts to the moon. The second major order for semiconductors around that same time was for the guidance computer in the Minutemen who intercontinental ballistic missile which was intended to deliver nuclear warheads from the United States to the Soviet Union. And in some ways that no surprise that guidance computers on missiles were a key driver of the origin of their chips. They had a huge reliance on computing power to guide them accurately and needed computing that was small to fit in the cone of a missile, for example, a big change from the computers of the 1940s and 50s, which were often the size of an entire room. That would work for certain use cases. But it certainly wouldn't work for a missile and since that point, there's been a deep relationship between the demands the defense industry largely in the United States, and advances in computing power, further miniaturization of transistors which have allowed the industry to put more transistors and more computing power on each chip, and further precision in the software tools and the machine tools that are needed to produce chips. If you look at how the US government has funded chip research, now what you'll find is that DARPA, the Defense Advanced Research Projects Agency, which is the Pentagon's R&D arm, has played a tremendous role in driving forward advances in chip making technology from the 1960s all the way up to the present. And today, in the United States, as well as in other countries, defense use cases are the driver of advanced chip technology. And even though today 98% or so of chips end up in 2 billion uses for phones and PCs, and household appliances, it's still the case that defense users are pushing the boundaries for what technologically possible in advanced semiconductors.

Now they're doing that today because now more than ever before, there's a huge interrelationship between computing power and military power. And this has been the case since the 1960s. This is why the US wanted to apply computing to military systems in the first place. But the trends that began at that point have only accelerated communications capability, sensing capabilities, signals processing capability, and autonomy and military system. All rely on advanced semiconductors and all of these are going to be even more important in next generation defense systems than they are today or wwew several decades ago. One way to conceptualize the role that semiconductors and computing have played in defense systems is to ask yourself what differentiates a fighter jet today from a fighter jet 50 years ago or a tank today from a tank 50 years ago and certainly there are differences in their design and their manufacturing but it is not hard at all to recognize a 50 year old plane as being fundamentally pretty similar from the way planes are designed today. It's not like they fly 10 times as fast or carry 10 times the payload. The key differentiating factor in how they operate on a flight is that they've got a lot more information coming onto the plane and a lot more ability to process the information once it arrived at the ability to connect different, different systems on a battlefield to network them together to provide them the information that they need, and then to help them reach their target, which has transformed the character of war over the past half century. And one of the I think most interesting observations of the Russia Ukraine war is just how reliant both countries are on advanced micro electronics that either power or failed to power their military systems. It's been striking for example to me the extent to which the precision guided IMR systems that Ukraine began receiving this summer really amounted to a turning point in the Ukrainians ability to push back into the Russians in the war with a combination of precision guided home HIMARS munitions, plus, all of the semiconductor enabled intelligence that the US has been transmitting to Ukraine collected by satellite by electronic surveillance, and then transmitted Ukrainian in terms of target coordinates, has been transformative for Ukraine stability, to launch counter offensive against the Russians and degrade Russia's logistical capabilities by striking deep behind Russian lines. Then whether you look at the United States or China or Russia, all of the world's major military that they think about next generation systems are envisioning military systems that are even more dependent than today on advanced computing capabilities. Systems that are semi autonomous and their operation systems that have even more data flowing in and out of a given platform systems that are more interconnected than ever before, and therefore rely on more advanced and capable communication systems. And all of this depends on semiconductors. The analogy of the in the civilian world is automobile. A half century ago, it was possible to have an automobile that didn't have any chips inside. Whereas today, the newest automobiles, and especially the most advanced automobiles will have hundreds in some cases, 1000s of chips inside because the features that they want, autonomous driving system, more advanced sensors, more advanced communications, and entertainment all require more chips, and these are the same types of systems that advanced mili,tary systems require too so there's no doubt there's going to be an increase in the number of chips per military system, and an increasing reliance on chips to provide the performance increases that advanced military systems require. Given that, given that militaries around the world think that chips will be even more important in their systems in 10, or 20 years time than they are today, it's no surprise that governments are trying to increase the control they have over the semiconductor supply chain. But the fact that the semiconductor supply chain is a single supply chain that stretches from Europe to the US to Asia, and has focal points in the United States and the Netherlands, Japan and Taiwan means that the control over the supply chain will impact the military balance in the future. Today, China finds itself in a very vulnerable position. On the one hand, it's been building up its military in a very dramatic fashion over the past two decades, focusing in particular on the types of capabilities that would be needed to seize Taiwan. But the Chinese military has been hugely dependent on its access to imported semiconductors. And although, although China has a domestic chip making capability, which is not very advanced, but nevertheless, the this, all of the world's militaries, we now know, rely quite fundamentally on access to dual use chips that are relevant both for civilian and military purposes produced in countries like Taiwan, and South Korea. And if you take apart Russian weapons systems that have been captured in Ukraine and several recent open source studies have done, what you'll find is that they're full of US, South Korean, Taiwanese produced chips and the Chinese military is probably not all that different, and it's reliant on access to foreign chips for military purposes. China is a whole spends as much money importing chips each year as importing oil. Semiconductors are China's greatest vulnerability when it comes to gaming out potential blockade scenarios around a crisis in the Taiwan Strait, which is why since 2014, the Chinese government has been pouring 10s of billions of dollars a year in an effort to domesticate chip making technology and chip design technologies, so that it's no longer so reliant on the international chip supply chain and can produce more of the technology itself. But because of the criticality of semiconductors and the future of military power, China's access to advanced and conductor's is now under threat by new restrictions rolled out by the Biden administration. In early October, the US government looking at the military balance in Asia, sees a military environment that has shifted dramatically in China's favor over the past two decades. China's tremendous build up in its military forces and especially its advanced missile system put at risk America's ability to defend Taiwan in a crisis or at least dramatically raised the cost of such a defense to the point where it's unclear whether the US would actually be willing to intervene on Taiwan's behalf. And the effect that is to reduce America's ability to deter Taiwan. And so long as China continues with its quantitative build out building more chips, more planes, more metal, the military balance is likely to continue shifting in its favor over at least in a couple of years. And so the US military around almost a decade ago decided that it's only hope to stem the shift in the military balance was to bet even more heavily on micro electronics to match China's quantitative edge with a qualitative edge. And the hope in the US is that the semiconductor advantage that the US still has, by producing some of the key software and machine tool input needed to produce that chip anywhere in the world can be used to preserve the US advantage in producing computing power, and thereby gives you an access to a more advanced military system. And over the past several years, but especially over the past few months, the US has been cutting off China's access to the types of tools and software needed to produce advanced chips. And so what we're beginning to see is a bifurcation of supply chain into a US and allies focused supply chain on the one hand and a China domestic focus supply chain on the other, where China lags behind pretty substantially today, technologically, and it faces a tremendous challenge in trying to catch up. But it has every incentive to try to do so though it will be brutally expensive of an effort because if it doesn't do so, the future of the military balance, may well begin to shift back in United States' favor that if you believe as the Chinese military clearly does, that the future military power depends in no small part on applying AI to military systems. And you got to have access to the most advanced semiconductors, to to actually put intelligence into your military systems in a way that you're planning to do so. And that requires China to invest even more into trying to domesticate the types of technology. And I'll conclude by noting that this is going to be an extraordinary challenge for China to do because although China is the world's second largest economy today, in the chip industry it is still a very small player, especially when you're talking about at the cutting edge of the chip industry. And although it's made some real, real drive in the past decade, especially on the thought of chip design, when it comes to the machine tools needed to produce ship, China remains quite far behind. And even the Chinese companies that produce somewhat analogous tools to what you can buy in Japan or the US Netherlands tend to rely on component parts sourced from the US, Japan and Europe, which are also quite vulnerable to control. And so what we're finding today is the future of the US and a military balance being being decided not solely based on what happens in defense ministries or what happens in the battlefield, but also what happens in the control of the component technologies that will provide computing power and all future military systems. And unlike all previous eras of technology, today, there's one supply chain that can produce vastly better examples of that component technology than any other one. Never in history, has there been a supply chain so prone to monopolization at so many critical parts of the production process. And I think that will make this supply chain particularly difficult for China to try to replicate using domestic components. It still won't be impossible, but it won't be an easy task either. And so I suspect that for the next decade, China will find itself in an uncomfortable position of trying to reduce its reliance on US technology and semiconductors while still remaining quite reliant, even as the US tries to use what remains of its semiconductor advantage to bolster qualitative military advantage and preserve the existing balance of military power in Asia. So I'll stop there, Leslie, and turn it back to you and look forward to any questions from the audience.

Leslie Johns 29:36

Okay, thanks so much, Chris. You know, obviously, it's a really exciting project that you presented. You know, for our audience members, there's, there's so much going on in the book, really. Chris has told us about how the book relates to some of the ongoing foreign policy issues that are confronting the Biden administration. I'll just chime in and say that there's a lot more to the book, sort of tracing out the intellectual histories of how these technologies developed. I think if the Burkle Center were to issue like a Christmas gift guide, this would definitely be a book that would be a great gift to any technology lovers or history lovers who you might have in your family. You know, one thing I wanted to sort of bring up in our discussion is to circle back to sort of the issues relating to the Russia and Ukraine conflict that Chris briefly alluded to. You know, one thing that made me really excited to invite Chris is an op-ed that he published a couple months ago. I can't remember if it was the Washington Post, or The Wall Street Journal that I saw. And you know, I'm a great lover of international law, is my intellectual interest. And one thing that I thought Chris did a really interesting job in the op-ed of talking about was sort of the relationship between chip technology, and the way that the war was actually unfolding in the in the ground in Ukraine, and particularly, how we think about, you know, the rules of humanitarian and law fair and how the war was being conducted regarding civilians. And so Chris, I was just wondering if you could sort of like, you know, take us through what you outlined in that op-ed, because I thought it was really insightful about how technology affects the ways we fight wars and its implications for international law.

Chris Miller 31:36

Yeah, yeah. Well, thank you, Leslie. I think the the Russia Ukraine war has taught us a lot about technology and also about supply chain. And as I mentioned, the guidance computers have Russian missiles, for example, that are guided by by largely Western or Japanese or South Korean technology. Russia couldn't achieve the precision and can achieve without those chips, so there is a direct relationship between foreign technology and Russian military capabilities. But what was also striking is that precision in warfare can cut in different directions. On the one hand, precision allows you to avoid targeting civilians if you'd like. Yeah on the other hand, precision also allows you to target specific types of civilian infrastructure, if you'd like to that as well. And one of the trends that we've seen in Russia's war over the past several weeks, is that Russia is using what remains of its precision missile stockpile, to specifically target by Ukrainian energy, infrastructure, electricity infrastructure. And so one, you can imagine someone hypothesizing that precision warfare will be friendlier to civilians, because we'll have only targeting military systems. But the track record, I think of the past nine months is that actually, that's far from guaranteed to be the case, because you can also more precision target civilian relevant infrastructure, as well. And that's exactly what Russia is doing right now.

Leslie Johns 33:00

Okay. Okay. So it sort of can cut both ways. Right. Yeah. Okay, as opposed to sort of. I also sort of want to go back to you mentioned sort of, very briefly, the Biden administration's big announcement. I can't remember was it September, October, regarding his policy towards domestic chip production? And we've had a couple people who are guests in the audience today write in about this. One of our guests wrote, I heard engineers in Silicon Valley say we should bring semiconductor manufacturing equipment back to the US from China. Is this an option for the US? What are other options that the US has to protect its national security? I was wondering if you had any insights in regards to that?

Chris Miller 33:53

Well, I think first off, I would say, I think it's not necessarily the case that manufacturing chips on shore and national security are necessarily aligned. I think it's very easy to imagine placeswhere you wouldn't want that manufacturer chip for national security purposes, but manufacturing them in Japan or in Israel, or in Ireland, as many chips are manufactured doesn't seem to me to be an obvious security risk. So I think we should push again the thesis that onshore equals secure and offshore equals non secure. And the reality is that even if you onshore the manufacturing profit, you still haven't onshored the production of the chemicals, the production of certain machine tools, the production of the silicon wafers themselves. The supply chain is so internationalized, that onshore and offshore is not the right way to think because something will always be offshore. It's impossible to produce an advanced chip using all onshore technology, simply impossible, and will never be. So that's that's not the right way to think about it. But I do think that there's a frightening degree of concentration in the world chip production in the Taiwan Strait. Between, especially in Taiwan, where the risk of war I think it's undeniably grown over the past several years, and without Taiwan's chip making capacity the world would careen into a depression with a severity of which we have not seen, since the 1930s. Simple as that. Manufacturing globally would grind to a halt without the chips that are produced in Taiwan. And I think that's something that's dangerous. And so I, I would focus less on what percentage of chips we manufacture on shore and more on how do we increase our assurance that A, we're not growing increasingly reliant on chips produced on both sides of the Taiwan Strait and B, what can we do to ensure that we don't have a war induced disruption to Taiwan's chip supply?

Leslie Johns 35:48

Okay, so one of our audience members wrote, for example, about the recent announcement that TSMC, the Taiwanese chip manufacturing company, announced that they're opening a new plant in Arizona. So do you view that as weakening the Taiwanese monopoly on manufacturing, or is that just sort of them diversifying where they're locating? I mean, it's still a Taiwanese company, right, even though they're just producing in a different location?

Chris Miller 36:21

Yeah, that's right. And I think the key question with the TSMC Arizona facility is what scale it will operate at. The initial announcement was a pretty small scale facility. There's been some discussion that they might ramp up the size of that over time. But right now, it looks like it won't be a huge facility. And unless that changes, it's not going to make a dramatic difference in terms of the world's reliance on Taiwan as the center for shipping.

Leslie Johns 36:48

Okay, okay. So you view it more as sort of a pilot program rather than a major change in policy?

Chris Miller 36:56

I think that's right. Yeah.

Leslie Johns 36:57

Okay. Okay. Here is a comment. While there are many specialized machines that are necessary to produce chips, you focus on five companies and produce presumably five types of machines. Are those companies producing the most specialized machines? And what are the five companies?

Chris Miller 37:20

Yes, in this space of companies that produce manufacturing equipment, the semiconductor industry, there are five companies that are far bigger than the others. In the Netherlands, ASML, in Japan, Tokyo Electron, and the US supply of materials, KLA and Lam research, which have been in their market positions for a long time, and you can't produce an advanced chip without buying their equipment. As simple as that. And they do have an extraordinary role in the industry. Because anyone who wants to make advanced or even not that advanced chip needs to buy tools from at all.

Leslie Johns 37:56

Okay. One of our guests points out that one thing that's really important in order to manufacture chips is the importance of having workers with the necessary skills to work in these factories that usually require advanced degrees in STEM disciplines. Given our immigration system, doesn't it make it very difficult for the US to attract these skilled workers? Is it really possible for the US to develop this domestic productions like the Biden administration wants without significant immigration changes?

Chris Miller 38:43

Well, I think there's no doubt that facilitating high skilled immigration would make things easier for chipmaking. Yeah, that's, that's pretty clear. I can also clear that the reason chip manufacturing has developed more offshore than onshore over the past several decades is primarily due to differential tax policy. If you look at the cost of operating a new chip making, facility labor costs are relatively small, because they're actually not that many people. And it should make facility relative to the extraordinary cost machines involved. And if you look at salaries in Taiwan versus California, they're lower, but they're not, it's not a 10 to one differential. The key differences are actually in taxation, and the types of tax credits that are provided to chip makers because they have such extraordinary capital investment needed to build a facility and stock it with the machies involved that countries that have more generous investment tax credit get more chip making investments. That's just a fact of life. And still if you ask what explains the shifting of chip production from the US and Europe to East and Southeast Asia over the past several decades, you can't understand that process, not looking at different tax policy. I think next to that there's a critical part of the explanation is permitting and regulatory approval. It's just a lot more difficult to get all the permits, you need to open a chip making facility in the US or in Europe than it is in Taiwan or South Korea. And if it takes an additional six months, and the type of chip is only cutting edge for two years, if you're six months behind, you're a long way behind. And so the US over the past several months, as you mentioned, luckily has taken some steps on the subsidization side to reduce the the subsidy gap between East Asia and the US. I think it still hasn't done nearly enough on the permitting side. And the reality is that the slow regulatory process in the US at federal, state, local level, is still a real burden relative to East Asia in terms of opening up new facility.

Leslie Johns 40:53

Okay, and so that sort of further reinforces your argument that doesn't necessarily have to be about offshoring. It can be about sort of diversifying and building production facilities and other countries with whom we have alliances.

Chris Miller 41:09

Or to argue in favor of doing permitting factor, I, I suspect that most of the permitting delay in the US relative what you'd see in Taiwan, or South Korea is not due to good reasons. It's due to your bureaucratic inertia.

Leslie Johns 41:21

But that's probably not going to change. Right, you know?

Chris Miller 41:24

Well, I think, I think there's an open question about it. I mean, the reason why permitting is faster in Taiwan, and everyone realizes that chip making is important and for those reasons that that you shouldn't cause problems for chip makers. So I do think that there's the role here for political leadership and putting pressure on permanent properties.

Leslie Johns 41:42

Yeah, you're an optimist. I work in a political science department, I realized that like, it's not going to get better. Maybe at The Fletcher School, you'll raise the next generation of leaders to solve these problems, I don't know. But so one of the guests in our audience comes from a background in mining, and points out that that a big part of the story in terms of making chips is getting access to the minerals necessary to produce them, and that China has a lot of the ability to extract these minerals. And so isn't there isn't a part of the story that they essentially have. I don't know if they use the word charcoals, but that they have a lot of market power over access to the minerals necessary to produce these chips. So don't they kind of have a market power as well? Isn't that part of the story?

Chris Miller 42:41

So I think that's not really true. I think there is, the way you should think about market power in commodities or in industrial products is, how hard is it to find alternatives? And how long will it take and how costly will be in the interim, and to find an alternative to the US dominated microchip supply chain will cost easily half a trillion dollars and take many, many years a decade. To find alternatives to our reliance on, for example, many types of rare earths from China, it will cost a lot less and will take less time. So is there some market power there? There is some. Did it work well when China tried to cut were worth exports to Japan in 2011? Didn't work that well. Does that mean we should need to tolerate a situation in which China is such a dominant supplier where the world? And no, I think we ought to be spending more money now to reduce that dependency. But I think the power of our dependency on China's position in rare earth is far less than the power of kind of dependency on our chip making tools because of the difficulty in getting around them.

Leslie Johns 43:54

Okay. So one of my colleagues here in the Political Science department has a question. Given the high concentration of high quality chips in Taiwan, does the Biden administration's attempt to sort of cut China off from chips, might that actually provoke China to be more likely to invade Taiwan? Is that going to be interpreted as a provocative move that might lead towards faster invasion?

Chris Miller 44:24

Yeah, it's certainly possible. I think that dynamic is balanced more than the other dynamic. So I think, first of all, the question of does Taiwan that is the the major chipmaker deter China, an attack from China? I think this is sort of like the mutually assured destruction theory and in nuclear deterrence, we have this sort of theory of mutually assured economic destruction, visa vie, China and Taiwan. And, you know, I hope that's true. We're going to find out if it's true. I will note that 2022 has been a bad year for theorists of the mutually assured economic destruction because that was exactly the theory that undergirded Angela Merkel's energy policy, and it hasn't turned out very well this year. And I think there's an alternative dynamics that could be present as well that could actually stabilize dynamics. And, and the argument would be the following that the Taiwanese argue that they've got a quote unquote silicon shield visa vie Chinese attacj, that if China tries to begin to be an invasion, it will disrupt your supply chain, and then China won't do it. But I think it's implausible that a crisis in Taiwan would start with a large scale amphibious invasion. Both because history suggests that amphibious invasions are very, very, very hard to pull off, that most independent military analysis suggests that China would struggle with a amphibious invasion even today. It might get easier in a couple years time as their, their sealift capabilities grow as they are forecast to do. But it also seems like if you were looking at international politics in the last decade, you would include that Russia, full scale invasion of Ukraine, starting in February hadn't worked very well. The rest of the annexation of Crimea in 2014 worked brilliantly. And would be the strategy you want to replicate. And so I worry a lot about the scenario in which China take something small in the Taiwan Strait like one of the uninhabited islands, which you could plausibly do so bloodlessly, and then look at Taiwan and the US, and say it's your move. And then the US president gets a briefing from their national security adviser that would say you have two choices, Mr. President. You could either do nothing and accept that like Crimea, or you could militarily escalate in which you'd be firing the first shot. You would risk World War Three escalation dynamics, and you'd be guaranteed to cause $100 billion or more disruption to the global economy by gumming up the semiconductor supply chain. It seems quite plausible in that scenario that Taiwan's critical role in producing semiconductors could deter us rather than hurt China. Well, I think that silicon shield argument is probably wrong, and that the in most likely military scenario, we ended up getting more deterred by Taiwan's semiconductor industry than China does. So I'm skeptical of the silicon shield argument for that reason. I'm on the second question of do the latest controls incentivize a Chinese attack? No, possibly. I think it depends on what you think China thinks about the relative trajectory of its capabilities versus American capabilities when it comes to military power. And here, I think that China is getting overly optimistic advice or Chinese leaders are getting overly optimistic advice about their likely future technological trajectory. I think that they're being told by their chip industry, that they're going to succeed in domesticating relevant technology in a relatively short period of time, even though the chip industry in China knows that not possible. They're telling Chinese leaders what they want to hear. And therefore, I think Chinese leaders think that time is on their side in the tech struggle, even though I think that's wrong. So if my analysis is correct, and there's uncertainty there, but if my analysis is correct, then it's most likely not that the control will, will will encourage China to strike first. I acknowledge that there is uncertainty around that and that the dynamic pointed out will play some sort of role.

Leslie Johns 48:28

Okay, okay. I should start for those of us, for those of you in the audience, who are super intrigued by this question. I should point out that that we have lined up in February 15, I believe it is Susan Shirk will be coming to discuss her new book on China called Overreach, which I am currently studying up on in preparation for our conversation. So I will try to remember to ask her the same question. And maybe we'll get opposing biewpoints going on this. So another question from one of our guests is, given the geopolitical importance of semiconductors, have there been any discussion by national governments about trying to break up some of these oligopolies? And perhaps, you know, treating them as you know, monopolistic powers, perhaps trying to incorporate more competition, and or perhaps trying to bring some of the production processes under government control? From reading your book I don't recall any discussion of that having occurred. Can you recall any examples of that happening? It seems like even though there was a lot of government involvement in purchasing chips, there was never really any serious contemplation about the government actually producing the chips or was there?

Chris Miller 49:51

Not in the United States. There has been in other countries but I think the track record has not been great.

Leslie Johns 49:56

I can't imagine that the US government would be good at that. Yeah.

Chris Miller 50:00

Yeah, I think there's a there certainly are monopolistic tendencies in the chip industry. But unlike, you know, electric utilities, for example, yeah, where there's, you know, basically no productivity improvements year on year, the chip industry must produce the doubling in computing power every two years to stick with Moore's law. And that the rate of improvement that I don't think governments have a great track record of matching.

Leslie Johns 50:21

Okay, I do have one technical question that came in from one of our Burkle interns. Do you buy the narrative that Moore's law is slowing significantly? And if so, what ramifications do you predict that this will have on manufacturing and supply chains of semiconductors?

Chris Miller 50:39

So the Moore's law is the prediction made in 1965, that the growth of the growth rate in computing power on chips will grow at an exponential rate. So it's been doubling annually, roughly from a couple of decades. It's weakened over the past decade. Unlike previously, we've no longer got a declining cost per transistor. In other words, we now have to pay for additional computing power, and we didn't have to for most of the past years. However, it's still been the case that we're getting more computing power for year two. We're just having to pay more for it. I think we've got a pretty clear pathway between now and 2030, to keep shrinking transistors at roughly the same rate. Beyond then, it's hard to say, but it's always been hard to say what technology will do 10 years out. So I'm a relative optimist, when it comes to the question of Moore's law, even though it's certainly true that at some point, we're going physical limits when we just get down to the atomic level, and you just can't make transitors anymore.

Leslie Johns 51:38

Okay. And perhaps a related question, this will probably be the last question. I know, we've put you through the wringer a lot here. From your perspective, where are we in the life of chips, specifically as an emerging disruptive technology? Are we still waiting to see more development? Or are we content? Are our contemporary minds capable of mitigating the threat of chips, allowing for lethal autonomous weapons system resulting in imbalance? I'm sorry, it's kind of a long question. But but I think I think the question here is like, you know, are we fundamentally sort of at the beginning of explosive change? Or are we kind of at a more steady state? Or are we kind of at the end of this disruption, where we're kind of more at a steady state?

Chris Miller 52:30

Well, I think what differentiates chips from all other parts of the economy, is that in most of the economy, productivity grows by 2% a year or something like that. In chips, you get a doubling of performance every two years. It says this, you know, imagine if planes could fly twice as fast every two years, or houses were built twice as large every two years, same price, nothing else, the economy works this way. And to say that we're in a steady state, I think dramatically misinterprets the transformative power of chips on our lives, because nothing else we know, works at that exponential growth rate, but semiconductors have and that's why they've been transformative for the economy, for technology for military power, as well, because they changed far more rapidly improved far more rapidly than anything else that we have access to.

Leslie Johns 53:16

I see. Okay. Well, thank you so much to our guests today. Chris Miller, I really encourage everyone in the audience, if you haven't bought the book already to go out and buy it, even though it's a highly technical subject. It really is a compelling read. It's really like a story book. It's a story about people, places. It's also a really great story about migrants, about people traveling all over the world, interacting with each other moving across borders. It's kind of an inspirational story about people traveling across the world to pursue their dreams. That's certainly what I took away from it. Right, the story about like people just going places, you know, pursuing their education and discovering things. I thought it was a really exciting book. And, you know, I really congratulate you on your success. It's wonderful to see your name reappearing on you know, top 10 lists all over different newspapers. Financial Times, I think it was yesterday name this top 10 book and technology for the year. So congratulations, and thanks so much for making the time to be with us today.

Chris Miller 54:29

Thank you for having me. Bye, everyone.

Leslie Johns 54:31

Have a have a wonderful winter break and I hope to see you all in the new year for a wonderful new series of book talks.

Transcribed by https://otter.ai