Tiny Chips, Giant Stakes

Published Mar 23, 2023, 4:00 AM

Microchips are the most important driver of technological progress in the modern world, and governments are fighting over who gets to make them.

Right now, most cutting-edge chips are made in Taiwan, a country that China claims as part of its territory. The U.S. government is fighting to keep semiconductor technology out of China, and spending tens of billions of dollars to get companies to build more chip factories in the US.

Chris Miller is a professor at Tufts University and the author of a book called Chip War: The Fight for the World’s Most Critical Technology. In this episode, he talks with Jacob about the extraordinary technology and complex geopolitics of microchips.

Pushkin. This is a show about technological progress, and yet in the year or so we've been making the show, we have not had a single episode about the most important driver of technological progress in the modern world. Semiconductors, microchips, chips, as you know, are everywhere in the modern world, except on What's Your Problem. I apologize for the oversight, and we're going to fix it right now. I'm Jacob Goldstein, and this is What's Your Problem. My guest today is Chris Miller. He's a professor at Tufts University and the author of a book called Chipboard, The Fight for the World's most Critical Technology. The book explains not only how chips have become a billion times more powerful over the last several decades, but also how that extraordinary progress has made chips ubiquitous and essential and increasingly something governments are fighting over. The US government right now is fighting to keep chip making technology out of China. Also, the US government is spending tens of billions of dollars to get companies to build more chip factories in the United States. And on a related note, most of the world's cutting edge chips, almost certainly, including the chips powering your phone right now are made in Taiwan, a country that China claims as part of its territory and that China might invade or blockade or generally mess with in the coming years. Chris writes about all of this, and we talked a lot about it. But in order to really understand what's going on now and how we got here, we started out by talking about two companies, in particular, giant of companies that are doing amazing things and that are absolutely essential to the global economy. Also, almost nobody in the US ever talks about either of these companies. The first one we talked about is a Dutch company called ASML. They make this amazing machine that you have to buy from them if you want to make cutting edge chips. Nobody else in the world makes this machine. The machine uses a technique called lithography to print impossibly intricate chips. To start, I asked Chris to explain how lithography has evolved. So the way you do it is by using light as your tool, and you shine light through a mask that has the pattern you want to imprint on the chip. So the light goes through where the holes are, it doesn't go through where there aren't holes, and by using essentially an upside down microscope at the outset, you could take something that was big and make it look smaller, just like microscopes usually take something smaller, make it like when you look through the wrong end of a telescope. That's exactly right, that's exactly right. So you basically take sort of like a stencil, you stick it on a microscope, you shine light through the microscope the wrong way, and you you're able to basically print a chip that way. That's right, at least that's how it worked in the simple earliest days. So the challenge has always been to print ever smaller features on chips. And for a long time, visible light was perfectly acceptable printing device because it had a wavelength of several hundred nanimeters and that's pretty small. But a couple decades ago we got to the point where chips already had features that were so small a visible light wasn't small enough or powerful enough to do the printing. So is that is yes, a wild moment, right, So the idea that like these things on the chip are so fine, so close together, that light is weirdly too big, like they're smaller than the wavelength of visible light. So this is the next problem. How does that get solved? You need to make a jump to a very different light source with a much smaller wavelength, And that's the origins of the EUV, the extreme ultra violet lithography machines that we have today. So ultra violet light has a shorter wavelength than visible light, so therefore you can print even smaller stuff on a chip. Extreme ultra violet is presumably the short end of ultraviolet light, so you can still get smaller and smaller. So we're in this world where there's this one company, SML. Right, tell me about this company, and tell me about the EUV machine that itselfs. So the machines themselves create UV light by having balls of tin thirty microns wide, so thirty millions of the meter wide falling through a vacuum. Then, like, first of all, what does that even me? Right? You say those words, but like why it's what a ball of tin thirty microns wide falling through a vacuum, Like what's even going on? Why do you need falling balls of tin in a vacuum? Because this tin, when you strike it twice with an ultra powerful laser will explode into a plasma that is forty or fifty times hotter than the surface of the sun, and this plasma will release light with exactly the right wavelength extreme ultra violet light needed for lethar. And so then that light goes through some kind of a mask or stencil and makes the imprint on a chip. Well, the trick is that you need really unique mirrors that the flattest mirrors humans have ever made to collect that light after it's produced by the plasma. And there's around a dozen of these mirrors inside of each machine that then directs it towards the stencil and then onto the silicon wafer. So okay, so that those are the basic mechanics of the machine. How big is one of these machines? What's it look like? The size of a truck? Okay? And they look like they got wires and tubes of all sorts coming out of them from every direction. They're the most complex manufacturing tool humans have ever made. And inside of them there are hundreds of thousands of component parts. So you say it's the size of a truck, like a like a pickup truck or like a semi truck. What am I picturing like a semi truck. It takes multiple airplanes to move one of these machines. Oh interesting, okay? And how much does it cost? Around one hundred and fifty million dollars apiece, So they're the most expensive machine tool produced in human history. So okay. Because this machine is so expensive and so complex, and because the market for it is so small, right, only a few companies in the world make truly cutting edge chips. We've found it in this world where there is only this one company ASML that makes this machine and that fact is really important in terms of trade and geopolitics and lots of other big things that I want to talk about that. But before we do, I want to talk about one other big important underdiscussed in the US company that also has an acronym for a name. That company is TSMC, the biggest chip maker in the world. So tell me the story of TSMC. TSMC was founded by an executive named Morris Chang who had spent his career at Texas Instruments really building the US chip industry, and TI at the time was one of the leaders in chip technology, but he was passed over for the CEO job in the middle of the nineteen eighties, was looking for something else to do and was approached by the Taiwanese government to help build a chip industry in Taiwan. And he realized at the time that as manufacturing chips was getting more complex and as the economies of scale to manufacturer growing because you needed more complex equipment, more specialized materials, that in the future there would be a market for manufacturing services of semiconductors. And so he can see this company not to design any chips. They've never designed chips. They only manufacture, and they serve a large number of companies from Apple to AMD to Nvidia that today don't do any manufacturing. They only do chip design. And so when people talk about like Apple zone chip, you know, when Apple starts using its own chips, Apple's not making those chips, right in the same way that they're not making whatever the rest of the iPhone, right, they're designing the chip and it's actually being made in a factory in Taiwan. In fact, that's right, and I think if you if you look on the back of an iPhone, it'll it'll say designed in California, assembled in China, And that's true, but it misses a critical step because all of the key chips, not only in iPhones but in most Apple products are manufactured by one company in Taiwan, and so t SNC is now at the frontier, right, they are making the smallest, most advanced chips. No US company is any longer at that frontier. So in that TSMC story, is there like a moment that is sort of the key moment when the kind of center of gravity and the chip world shifts from California to Taiwan. Yeah, the key shifts was the smartphone. Steve Jobs actually went to Intel when he was conceiving the iPhone and asked if Intel will be interested in producing chips for this device. But at the time it seemed a little bit crazy to think that people would want computer sized processing on their phone. Intel thought it'd be a low volume product and said no thanks, and he took it to contract manufacturers in East Asia instead, first producing it at Samsung and then later turning to TSM to be the exclusive producer of the iPhones chips. And it wasn't just Apple, the entire smartphone ecosystem grew up alongside TSMC, and so today TSMC produces on eighty percent of the world's smartphone processors, and a typical smartphone will have a dozen semiconductors inside, one for the Wi Fi, one for the Bluetooth, one for the audio. And many of these ships are produced by DSMC. So your background, you're a scholar, and your background is basically in international relations, right, not in technology your innovation. And it seems like that seems like an extraordinarily interesting and useful framework for thinking about chips basically right, for thinking about sort of global the global semiconductor industry today. I mean, you have one company that's essential that's in the Netherlands. You have another company, maybe the most important company in the global economy if I wanted to be found to reach for it, TSMC as in Taiwan. Taiwan is an island right off of China that thinks it's an independent country, but that China thinks is a part of China. Semiconductors are like the most important thing in the world economy right now. Maybe oil, but you could make a good case for semiconductors, like that's a super interesting, super complex, fraught situation. How should we think about it? Well, I think you're right about how complex it is. And what is striking to me is the extent to which we think of industries like the chip industries being globalized and they're international, but actually the production is concentrated in a number of really key countries and companies. Joe Biden was in the Netherlands a few months ago in part to make sure that SML wasn't selling its fancy machines to China, right, Like, that's how important it is that the president is going to go there and be like, thank you for not selling these machines to China. Please keep not selling them to China. The Biden administration also imposed restrictions more generally on selling chips to China. Was the last year, like talk me through that, talk me through China's role in in the global semi conductor industry. Today, China produces a fair number of chips, but almost all of them are pretty low tech, okay, And when it comes to cutting edge chips, China's far behind the cutting edge of what can be produced in Taiwan or in the United States. And almost all chipmaking in China requires machine tools like lithography tools from ASML, other types of tools that are imported from abroad from the US, from Japan, from the Netherlands, and because of that, the Chinese economy is critically dependent on imported chips. It's trying to trying to catch up. Are they gonna catch up? They've been trying to catch up Since twenty fourteen, the Chinese government has made some conductors a priority, pour billions of dollars each year into the chip industry in China. But the problem is it's really hard. It's really really hard to acquire these capabilities. And it's hard because we're talking about the most complex manufacturing humans have ever undertaken. The difficulty that Chinese firms have faced are twofold. First is that the market is so consolidated that breaking into it requires enormous capital investment plus really unique technologies, and so it's just hard to break into new markets when it comes to the chip industry, which is why you've seen in many segments of the industry, firms stay in their market position for years, if not decades. And that's just an environment where new entrance is hard, Like like making becoming another SML, it's like kind of not going to impossible. Yeah, that's right. Yeah. Yeah. The other issues that the US government has been making it harder over the past around five years by cutting off China's access to certain types of tools, equipment, materials, software, and knowledge because now it's illegal for USS and to work with certain Chinese semi conductor companies and transfer knowledge to them. And the US wants to do this because it's afraid that if China develops more advanced shipmaking capabilities, it will apply these to military and intelligence systems, which of course it would, That's what all governments do, right exactly. So China wants to make advanced chips, and Taiwan makes the most advanced chips in the world, and China thinks Taiwan should be part of China. In a minute, what does Taiwan's chip industry mean for its relationship to China? In particular, what does it mean for the possibility of a Chinese invasion or annexation of Taiwan. That's the end of the ads. Now we're going back to the show. In a more general context, people talk about the possibility of China invading or blockading or annexing Taiwan. How do you think about that in the context of the chip industry. Some people argue that China would want to attack Taiwan in order to acquire the chipmaking facilities there. I think that's close to impossible. The reality is that although Taiwan has unique manufacturing capabilities, the facilities in Taiwan require importing machines from the Netherlands, from the US, from Japan, they require importing materials like silicon wafers and lots of ultraspecialized chemicals, and so they really couldn't operate without regular imports from abroad. So I think it's really really unlikely that China were to attack Taiwan with the aim of seizing that you're making facilities, because the Chinese leaders know that they be blown up in the process. It wouldn't be possible. So China certainly thinks Taiwan should be part of China, right, that's not ambiguous independent of TSMC. So in a world where China tried to make Taiwan part of China by force, what would happen, Like, what is the sort of set of probabilities you think about in terms of what would happen to the chip industry and the global economy more generally if there were attack on Taiwan or a blockade that disrupted trade and out of Taiwan. The impact for the world economy would be catastrophic. TSMC produces eighty percent of smartphone processors, it produces a third of PC processors, It produces all sorts of critical chips and data centers and telecoms infrastructure, and then it produces tons of less sophisticated chips that are critical for many other types of goods dishwashers, washing machines, coffee makers, microwaves. A new car will often have a thousand chips inside of it, and in a given car, you should assume that at least twenty percent of the chips are made in Taiwan. So when you sort of think with your foreign affairs international relations training about those implications, like, I could imagine that going different ways in a kind of game theoretical way for China right on the one hand would be like, oh, well, we don't want to blow up the world economy. On the other hand, it's like, oh, that's like a kind of leverage. Right. We could say to the world, hey, just let us make Taiwan part of China, because we all know it's part of China, and we'll let the chips keep flowing. I don't know, are those are those the ways you think about it? How do you think about it? The Taiwanese government describes the chip industry is a silicon shield, the idea of being that China one attack because it knows that the economic consequences would be disastrous for China and for the rest of the world, which it would be. And I think that dynamic is present. But I worry as well that if China tries to move on Taiwan in a way that's below the threshold of what would necessarily trigger US response, so a partial blockade, for example, which would present the US with a really difficult calculus as to what to do. In that type of scenario, US leaders would have a choice. Do you do nothing and let China pressure Taiwan while watching US credibility in Asia disappear, or do you do something and risk a disruption of the supply chains on which the world economy depends. And in that scenario, I think it's far from obvious that the chip industry helps secure Taiwan, and in fact, it could well deter the US from helping Taiwan and therefore give China leverage over the United States. So the US passed a law last year, the Chips Act, which is basically subsidizing the manufacturer of chips in the US. Right, tell me about that law. So, the Chips Act allocates around fifty two billion dollars to semiconductors. Three quarters of that goes to subsidizing chipmaking in the US, one quarter goes to funding long run R and D. And the idea behind the Act is that right now, it's it's more expensive to build ship making facilities in the US than an East Asia for a variety of reasons, government subsidies, tax policy, regulation, etc. And the US wants to reduce the cost gap and is putting government money behind that, behind that to make it more competitive to build in the US. There's real concern about what happens if China does attackable like hate Taiwan, and in that scenario, we need more chip making capacity in other geographies, not in China, not in Taiwan. And so that's why Congress put money behind the Chips Act to try to build some capacity in other geographies. I feel like there's a there's a bigger theme here that's interesting, right. An interesting theme is the the relationship of the government to the private sector and innovation more generally, and that's a theme that runs through the history of the chip industry really and I'm curious if you could just sort of talk it through, you know today, how has that played out, and how's it playing out today, and like, how do you see that sort of optimal relationship there. The government's been deeply involved in the chip industry from day one. It funded a lot of the r and D that made chips possible. It was the first buyer of chips for missile programs and for the space race, and even today it's a major funder of research and development the chip industry. But it was never the forest that let the chip industry scale. Selling to consumer markets was always more important in terms of scaling because there's a lot more consumer demand than government demand. Tim Cook, Apple CEO has a lot more influence over this and nine country supply chain than the US president because he buys a lot more expensive chips. In the end, what do you think it's the fundamental question of the book or what do you think is the question the book ends up answering. I think the key takeaway from the book is that although we don't think about I can Inductor is much at all. They're very deep in our devices. In fact, you can't understand any of the major transformations in the modern world without them. Whether it's the shape of the globalized economy, whether it's the balance of military power, whether it's the rise of big tech firms. All of them have silicon semiconductors at their core, and they've structured all these big trends and ways that until recently we were only dimly aware of. In a minute, the lightning round, including where Chris thinks the next silicon valley might be, and also what his next book might be. Now back to the show. I know you have to go relatively soon, so I want to close with a lightning round a bunch of fast questions. So you've studied the Soviet Union and Russia extensively, You've lived and worked in Moscow. What's one thing that you wish more people in the US understood about Russia. M That's a hard question to answer, right, I mean, I don't know if I don't know if you have a good answer to that question in the context of this conversation. Forget the context of this I mean, look, obviously, the context for that question is the war in Ukraine. Frankly, Yeah, Okay, well that's the context. Yeah. I think what's striking about Russia is the extent to which the Russian foreign policy elite, the people who make foreign policy in the Foreign ministry in the Kremlin, they're convinced that their country is a great power on the world stage, and they're convinced that the way to make your country great power is to assert it militarily and territorially. It seems to me like a very nineteenth century or before view. The problem is that it's here and now in the twenty first century, and we can't wish it away. Every region wants to create its own silicon valley, you know, there's like whatever, Silicon beach, Silicon mountain. Everybody has their own bad name. What place do you think has the best shot of doing it? And why India right now is putting a lot of money into its chip industry, and simultaneous to that, it's putting a lot of focus on electronics assembly that India is trying to attract assembly that's leaving China for smartphones and PCs, and so I would say that although it's starting from a very low base, India is likely to substantially grow as chip industry over the next decade. What's one thing research universities get wrong about fostering innovation. Innovation is partly about science, it's probably about engineering, but it's also probably about business models. The innovations that really transform societies are those that have a business model that allow them to proliferate. And universities aren't nearly as good at producing business model innovation in the fact that I do any of it, relative to producing engineering innovation or innovation in fundamental science. I mean, maybe the question was misguided as I hear you answer it, because like, maybe a university shouldn't be in the business of innovating business models. I guess there's a business school. But yeah, I think that's right. I think if you look at TSMC, for example, TSMC, you know, Morris Chang had no unique technological or scientific insight when he started TSMC, But by starting TSMC, he has transformed the landscape of the global chip industry. And so I would put the founding of TSMC next to many of the other key innovations of the twentieth century. But it wasn't a Nobel Prize winning innovation, even though it was areuably more important than many that have one Nobel prizes. I mean, the fundamental innovation of TSMC was we're not going to design chips. We're just going to make chips that other people design. It was just that. But that was huge, that's right, that's right, and it was informed by all sorts of technical knowledge, but the innovation was actually very simple and exclusively in the business model. What was the second most important technology of the last fifty years? Um, well, maybe this is a cop out answer, but I think the development of the software that takes advantage of chips bird most important. What's a good answer to that. I don't know I've got a smart answer to that question. I'll to think about that. Maybe genetic engineering. But we haven't seen it payoff yet, Like if I were going to it's kind of the obvious one, right, Like, I feel like maybe in the next fifty years we'll see that what has been happening in the sort of biotech world is about to pay off in a really profound way. You know, I'm I'm beginning to explore genetic engineering as a potential topic for my next Wait, that's my last question. What's your next book? Well? Yeah, so I'm looking at at genetic engineering and the intersection of biotech and aim as a potential next book. And if you if you start with the thesis that DNA's just code, then the intersection between computing and biotech seems really profound. And the ways in which biotech has developed, both in terms of the interplay between government and private companies and also the international competition around it, also seems seems very important. I think there's space to bring together some of these big themes in a fresh way. Chris Miller is the author of Chip War, The Fight for the World's most critical Technology. Today's show was produced by Edith Russolo, engineered by Amanda k Wong, and edited by Lydia Jeancott and Sarah Nis. I'm Jacob Goldstein. You can email us at problem at Pushkin dot fm, or you can find me on Twitter at Jacob Goldstein. We'll be back next week with another episode of What's Your Problem.

What's Your Problem?

Every week on What’s Your Problem, entrepreneurs and engineers talk about the future they’re trying  
Social links
Follow podcast
Recent clips
Browse 143 clip(s)