La guerre des puces : la lutte pour la technologie la plus critique au monde, avec Chris Miller

26 avril 2023 - 56 minutes d'écoute

Les micropuces sont le nouveau pétrole, la ressource rare dont dépend le monde moderne. Jusqu'à récemment, les États-Unis étaient la première superpuissance, mais leur avantage s'amenuise en raison de la concurrence de Taïwan, de la Corée, de l'Europe et, surtout, de la Chine. En La guerre des pucesl'historien de l'économie Chris Miller explique comment l'avantage des États-Unis sur le marché des puces a conduit à leur supériorité économique et militaire, et ce que cela pourrait signifier si la Chine rattrapait son retard.

Lors de cette discussion virtuelle, Miller et Doorstep Tatiana Serafin et Nikolas Gvosdev discutent de l'état actuel de la politique, de l'économie et de la technologie, ainsi que du rôle vital joué par les puces.

Chip War Doorstep Podcast Lien Spotify Chip War Doorstep Apple Podcast link

NIKOLAS GVOSDEV: Welcome, everyone, to this edition of The Doorstep Book Talks. I am your co-host, senior fellow here at te Carnegie Council Nick Gvosdev.

TATIANA SERAFIN: And I am Tatiana Serafin, also a senior fellow here at Carnegie, Council so excited to show you my dog-eared copy of Chris Miller's Chip War: The Fight for the World's Most Critical Technology. Chris teaches international history at Tufts University's Fletcher School. He also serves as the Jeane Kirkpatrick Visiting Fellow at the American Enterprise Institute and is really the premier person to talk to about the most important thing we can possibly talk about in April of 2023—chips, why they are important, and why we need to pay more attention.

This book I think does so many things, Chris. Let me first congratulate you on the book and second say that I learned so much, and I encourage our audience not only to listen to our talk today but pick up this book and forward it to all of your friends because this makes you feel smart. I say this because this subject is very complex—I am not an engineer; most of our listeners today probably are not if I may venture to guess—but this really nails down why this industry is so impactful and is so important.

I think you do it in a variety of ways. One of the most important to me as a journalist is all of the scene setting you do and the look at people because when I say "chip" or when you say "semiconductor" to the average person their eyes might glaze over, or they might feel terrified because some media outlet has hyped the crisis in the supply chain, which turns out is not so much a crisis, but we will get to that. This narrows it down to the people who made the difference, and that is why I really appreciate this book.

If I can, I am going to start out with the scene that you ended with because I love this scene in December of 1958, 65 years ago, when you had Morris Chang, Pat Haggerty, Weldon Word, Jay Lathrop, and Jack Kilby from Texas Instruments in a wintry Washington, DC, talking to Gordon Moore and Bob Noyce, who were out for beers, and they started this great industry that has transformed our lives exponentially in 65 years. I want to talk about those people and their impact not only on the United States but globally, how you found them, and what your process was to start this book because it is an interesting way to get into something that people normally view in the purview of engineering or military. How did you start that and end that with these people?

CHRIS MILLER: Thanks, Tatiana and Nick, as well for the introduction and the invitation to join you both today. It is interesting, Tatiana, you started with a question about the people because the people were what actually drew me into this topic. Some of the people that you mentioned in that scene from 1958 are obvious characters in this story, like Jack Kilby or Bob Noyce, who were the two individuals who get the credit for inventing the first chips, and there have been books written about them on the development of the technology, but the people who stood out to me alongside those inventors were also the people who built the industry itself.

I was expecting to end up writing a story that was more about inventors, and I have great respect for the inventors, but I ended up concluding that next to the inventors we also need to take just as seriously the people who made the inventions economically, financially, and logistically viable. There were next to those inventors a number of people like Morris Chang, who you mentioned, who was not an inventor per se, but he was the person who more than anyone else played a fundamental role in making it possible to mass manufacture devices.

I think that gets to a key challenge that we all face when we think of technology, because we often think of technology as being about producing a small number of very large things, but what I realized over the course of this research is that it is often easier to produce a small number of things than it is to produce many, many, many things, and what has made chips revolutionary is that now we cannot live our lives without them because they are embedded in all aspects of economy and society.

TATIANA SERAFIN: Let's just read your sentence: "Last year"—which I think was 2021—"the chip industry produced more transistors than the combined quantity of all goods produced by all other countries in all other industries in all human history." All human history. I think that is what really started to coalesce, that it is not just this phone or this computer that we are on, it is all products, everything that touches our daily lives—and this is so important at the doorstep—and we need to understand where this is coming from.

I think the biggest sentence that I say to people from this book that they do not believe but let's delve right in: "Taiwan Semiconductor Manufacturing Company (TSMC) builds almost all the world's most advanced processor chips." We read about Taiwan in the news and potential war, and people do not connect the fact that TSMC is such a big player today. They really don't. Maybe if we can tie our people talk—because there were very important people who made this happen from the U.S. side and from the Taiwanese side that make what happened 60 years ago very relevant and important today.

CHRIS MILLER: That is actually a perfect segue with the character that I mentioned, Morris Chang, who I think is one of the most underestimated entrepreneurs of the last hundred years.

Just a bit on Morris Chang's background, which will then end up with the Taiwanese company that you mentioned: He was born in China before the Revolution, fled because his father was an official with the nationalist government, so he had to flee the communists when they took power, and enrolled at Harvard planning to study English literature. His uncle told him to study something serious and get a real job, so he moved to the Massachusetts Institute of Technology and studied physics and electrical engineering. He was eventually hired by Texas Instruments and helped build up Texas Instruments as a real critical player in the chip industry, and then in 1987 he founded TSMC in Taiwan around a unique business model which was about only manufacturing chips, not doing any design, just doing the manufacturing.

Since that point, over the past 35 years, he has turned TSMC into the world's biggest chip maker and also the world's most advanced chip maker, so today almost all smartphones, many personal computers (PCs), our telecom infrastructure, and our data centers rely on the most advanced chips, which TSMC products 90 percent of.

TATIANA SERAFIN: I think the point here with Taiwan is that most people did not realize until—and I think you called it in your book—the Trump administration, the strategic importance of that, not just the business importance of what was happening in global supply chains but the strategic importance. Before we talk about other players—because we have a big meeting with one of them, namely South Korea, today—let's talk about what happened in terms of U.S. policy and understanding where we have left maybe things lying on the table that we were not paying attention to and how what happened under Trump has led us to the CHIPS and Science Act (CHIPS Act).

CHRIS MILLER: I think to tell this story completely you have to rewind the clock about ten years because it was ten years ago—or in 2014, so nine years ago—when China, which is a small player in the chip industry but has big ambitions began identifying semiconductors as a key focus of its industrial policy and passed a number of important industrial policy programs, of which the most famous is Made in China 2025, but there are a bunch of others, and tried to pour money into its chip industry and win a bigger market share for itself.

That was happening partly for economic reasons but also partly for strategic reasons because the Chinese government was concerned that today they are the world's biggest importer of chips and today China spends as much money each year importing chips as it spends importing oil, so it is a real vulnerability for Chinese leaders, and they import chips almost exclusively from the United States and U.S. allies—Taiwan, South Korea, Japan, etc.

So a decade ago China started pouring money into the chip industry, and that coincided somewhat coincidentally with China's growing military power on average. China has been building up its military over the past several decades, but in the past decade it has really begun to shift the military balance in Asia, which has worried a number of governments—Taiwan, Japan, and the United States. So there was both an economic and a security aspect to this.

The implications of this were actually first recognized in places like Taiwan and Japan, which are closer to China and are more embedded in electronic supply chains, so they realized the implications of what was potentially happening first. It was not until 2016, 2017, and 2018 that in Washington alarm bells started to go off, not only about the economic implications if U.S. firms were to be edged out by these government programs that were supporting new Chinese firms but also about the potential military implications.

I think that is important to recognize because although today most chips that are produced end up in civilian applications—smartphones and PCs—the origin of the industry is of course in missile systems, missile guidance computers, and today, like in the early Cold War when defense ministries think about semiconductors they see a component that is absolutely critical to most of their advanced technology.

NIKOLAS GVOSDEV: Chris, this is a fascinating conversation about supply chains and dependencies. I cannot help but for pop culture references think that if they remade the James Bond A View to a Kill movie, which was destroying Silicon Valley in the 1980s in order to cripple the chip industry. Today it would be Taiwan. That is of course the worry, that something happens to Taiwan and chip manufacturing is impacted.

You then see a lot of people saying: "Well, okay, so just open factories up somewhere else. Go to Arizona, go to India, go somewhere else." What is it about the production of semiconductor chips and how we set up the production, why is it so concentrated in Taiwan, and how easy or how difficult is it to just say, "Fine, close the factory in Taiwan, box it up, move it to Arizona, and problem solved?"

Why is it that we have this physical concentration of the manufacturing? First it was Silicon Valley, then Taiwan seems to have surged ahead to become the place, which as you say about Morris Chang, that his thing was: "I am going to make things, I am not going to design them, they will do that somewhere else, but I am going to actually make the physical chips."

As you said, China is dependent, and that is a fascinating statistic, the idea that as much as it has to import oil it has to import chips, and then you think of the implications in so many areas, not only for China but obviously Russia now getting chips from China in order to try to—as you said, back to the original purpose—maintain missile systems. We know that Chinese components are finding their way into Russian weaponry.

Why Taiwan? Why did that become the choke point, and how easy or how difficult is it to say, "Fine, we will just move manufacturing somewhere else?"

CHRIS MILLER: It will be very difficult to begin to shift the base for manufacturing. We are trying to do so slowly via the CHIPS Act, but it is going to take a long time. There are two reasons why it is difficult. The first is cost and the second is complexity.

First off, on cost, if you want to build an advanced chip-making facility, it will cost around $20 billion. These are the most expensive factories in human history, and these are facilities that are cutting edge for a couple of years before there is a new type of chip-making technology that comes along with exponentially better capabilities, at which point your $20 billion facility is second-rate. Just to put that into context, if you want to buy a cutting-edge aircraft carrier, it will cost you $10 billion, and it will be in service, the U.S. Navy hopes, for half a century, whereas a chip facility costs twice as much and is cutting edge for a couple of years.

The cost is truly enormous, which is why, although institutions like the Pentagon, which has an $850 billion budget, used to produce chips in-house decided over a decade ago that it had to outsource. We have specialized facilities to make aircraft carriers but not to make semiconductors. In every advanced fighter jet there are chips that can only be made in Taiwan, and the Pentagon decided that is the most efficient way given the cost of it.

The cost is actually the easier part to solve than the complexity because making an advanced semiconductor is the most complex manufacturing humans have ever undertaken, full stop. If you go to the Apple Store and buy a new iPhone and open up the iPhone, you will find dozens of semiconductors inside, but just the most important one will have 15 billion transistors carved into it. A transistor is just a switch that turns a circuit on and off—when it is on it is a 1, off it is a 0—and these transistors produce all the 1s and 0s undergirding digital computing.

Your smartphone has 15 billion or so of these transistors carved into just a primary chip, and the chip is roughly the size of your fingernail. In order to make 15 billion of them on a chip the size of your fingernail means that each one of them has to be roughly the size of a virus, and they are produced with basically perfect accuracy in every single chip that is made.

Nothing comes close to this level of manufacturing complexity. We measure the size of these transistors in nanometers, which is billionths of a meter, and in order to produce these things we have had to create a whole new set of manufacturing tools, chemicals, and software design processes that previously didn't exist. It has worked marvelously, but it has been horrifically complex and expensive, which is why today there are only three companies in the entire world that are anywhere close to producing cutting-edge processor chips, and Taiwan has 90 percent of the market.

TATIANA SERAFIN: Just to follow up on that question, there is a great map in your book—again, get the book; it is amazing—and glossary. I love the glossary, I love the people glossary, I love the tech glossary, but I love, love, love the map of where semiconductors are produced. If you can tell online [holds book to camera] that looks like East Asia. That is not the United States, that is not the Middle East, that is not Latin America, that is not Africa. This is a really big deal in terms of connecting the dots on why U.S. administrations talk about a "pivot to Asia." I think your book succeeds in connecting diplomacy, business, the past, and the future in ways that I think are important for us to understand where we are going in the next generation of products and technology.

I want to go back to how this map was created, why it happened, and maybe talk a little bit and go back to why Taiwan. It really was a business decision initially, right? Even though the U.S. government, in Vietnam you have great examples of how they were the first to put these guided missiles into action, it really was an opportunity—I think you had the little sign of the engineers: "I want to get rich." That is what they wanted to do with these products.

The decision to outsource to Taiwan, to Korea, and to Japan was based on cost and labor. If we can look at why that still makes sense today and what our competitive advantage is in the United States for our audience in order to understand why that system is not going to change on a dime. Yes, it is because of cost and complexity. Maybe it is not globalization. I am going to use the term that I found in your book, "weaponized interdependence."

CHRIS MILLER: It is a great question, and it does force us to rewind the clock again half a century because the chip industry began offshoring some assembly and then production since basically the earliest days of the industry. The first chip was invented in 1958, and the first offshore assembly facility in Hong Kong was set up in 1963, so five years after the industry set up.

This process of internationalization or offshoring is in the industry's DNA. It was important at the outset because as you say the labor cost differentials were huge, and from the earliest days chip making involved both early engineering and expensive capital equipment but also for certain parts of the production process lots of labor.

In the earliest days of the industry there was gluing of wires that had to happen by hand, for example, and it is still the case that there is a fair amount of labor-intensive work at the end of the chip-making process. All of the nanometer-scale manufacturing today happens by tools that are highly automated, but when you are taking a chip and putting it into a smartphone or a PC some labor is needed there, and that is part of the process that basically has to happen today in countries with lower wages than the United States.

What has changed in the past several decades is that it is not only the labor-intensive parts that are not happening in the United States. The assembly is happening offshore, but today also a lot of the fabrication, the actual manufacturing stuff, is also happening offshore, and this is not solely due to labor differentials. If you look at the cost of an advanced chip-making facility, what you will find is that over the lifetime of the facility around 70 percent of the cost is the machines inside of it, these ultra-precise complex machines that are capable of doing this minute manufacturing, and those machines cost the same regardless of which country you install them in, so the capital costs are roughly the same.

There are labor cost differentials between the United States and Taiwan or the United States and Korea, but these are not low-wage countries, and so the key reason why Korea and especially Taiwan have become centers of chip manufacturing has more to do with the companies in particular that have been set up there, the fact that the huge capital investment has been treated very favorably by those countries' tax systems, and by the fact that those companies have been willing to pursue very capital-intensive investment strategies, which is expensive and is often not the most profitable way to produce.

What we have seen in the last several decades is that U.S. firms have focused on designing chips, and chip design does not require a lot of upfront investment; you just hire designers, you buy a couple of software tools, and then you get chips that are designed. Chip manufacturing requires huge upfront investment, billions and billions of dollars to build these facilities, and they are both roughly equally profitable, which means that you are much better off in the chip design space from a purely financial perspective, which is why there are U.S. firms like Nvidia, Advanced Micro Devices (AMD), and Apple, which most people do not think of as a chip design firm, but actually it designs semiconductors, and Apple has done very, very well financially by just focusing on design and outsourcing the expensive part, the manufacturing, to a couple of companies in Asia.

TATIANA SERAFIN: You mention those companies and then you spend a lot of time talking about Intel here and how Intel lost some of its competitive advantage. Is that a reflection of the United States losing sight of the prize? Is Intel's struggle to get it back our struggle as well?

CHRIS MILLER: I think there were some structural factors across the United States that impacted everyone. For example, when we conceptualize advanced technology we think of social media and search engines rather than advanced manufacturing, even though it is the ability to manufacture chips that makes social media possible. I think that has been an intellectual error on our side. We have underplayed the importance, the significance, and the fascination behind mass manufacturing.

I think the fact that our software firms are still profitable has let them pay very, very high wages, which is great in some ways, but it has also meant that if you are an undergraduate deciding between a degree in computer science or material science, the labor market is telling you to go to computer science, and that has made it harder for firms in the chip industry to attract as many engineers as they have needed because there have been so many good jobs in the software space. That is another structural factor that has impacted the entire industry.

I think on top of that the government in the United States has been less supportive of the chip industry than other countries, and I do not mean primarily in terms of financially supportive, although there is some truth to that, but it is also just a question of making life easy for chip making. Chip making first off requires a lot of toxic chemicals, so there are all sort of environmental permits that we understandably require chip makers to take out before they build a facility, but we have been a lot slower at approving those permits than Taiwan, Korea, Japan, or even Europe actually. It takes longer to build a chip-making facility in the United States than in Europe. We have been bad at permitting, and therefore have slowed down manufacturing. Those are the structural factors that have existed across the industry.

Intel, which was until recently the world's biggest chip maker, for a long time one of the most successful chip makers, faced a number of unique circumstances about a decade ago that caused it to miss some opportunities. I think the primary driver of Intel's challenges is that it was too successful for too long.

Intel chips were at the center of the PC industry when PCs were a growing business—most PCs have an Intel chip inside—then they were at the center of the early days of the data center and cloud computing business, so they had these almost monopolies in these two critical markets. What it meant was that the company under the previous CEOs struggled to take advantage of new opportunities.

There is a famous moment where Steve Jobs came to Intel in the 2000s and said, "I've got this idea for a phone that will work like a computer. Would you like to make chips for it?" and Intel's old CEO said, "You know, it sounds like a low-volume product with mediocre margins." Boy, that was the wrong bet.

I think now Intel, because it has faced a couple of difficult years over the past several, is now in a very different position. It realizes that it has to innovate in terms of product and in terms of technology to regain its top position. I think we have seen a very different strategy but also a very different level of focus on innovation over the past couple of years precisely because unlike in the past where it had an almost monopolistic position now it has a real fight on its hands with the other leaders in the chip industry.

NIKOLAS GVOSDEV: When you are telling that story about Jobs going to Intel, it is so interesting how many times in Steve Jobs' lifetime big companies passed on opportunities. I am just thinking of the time that he—well, actually it was more Steve Wozniak—goes to Hewlett-Packard (HP) with what was the design for the Apple, and HP passes on it by saying the same thing as with smartphones: "PCs? Why would people want computers? It is only the big institutions." It is interesting that that seems to have been a pattern in Apple's history of pioneering products that established firms decided were not worth pursuing.

As you are discussing all of this, as you are discussing the real difficulties in standing up these kinds of facilities, that it is not a question of calling up Taiwan, boxing up a factory, sending it over, and reassembling it in Arizona—which is I think what some people popularly had in mind—or that we see coming out of India, that India is saying, "Look, we can become the new Taiwan; just make the investments here," the complexity, the cost, and then you have to be willing to deal with the tradeoffs including, as you know, the environmental ones, which suggests that there is a dependency that we will have to learn to live with, that this is not something that is just going to be solved in a matter of a year and a half or two years, and it comes back to this weaponized interdependence question.

I will maybe ask you to speculate. I am struck by—sorry, this may be somewhat of an obscure historical reference—we know from the ancient world in the Middle East that Babylonians, Assyrians, and Egyptians, when they would fight one of the things they would agree not to touch were the olive groves because olive groves were so important, the oil was necessary, and olive groves were difficult to grow and maintain, so you might sack and destroy the city but you left the olive groves alone.

China is dependent and we are dependent. With weaponized interdependence is there going to be a move to where we say, "Hey, we are not sure about the final status of Taiwan, but we all need these chips and we need the exports to continue," or can countries survive, can their industries, can their products survive, if there is this interruption of Taiwanese chips?

Of course, what happens if Mother Nature decides to send another tsunami or massive earthquake that damages Taiwan? Do we have the capacity to survive a potential interruption in Taiwanese supply, or do we all have now this shared interest, that China and the United States both have to agree that no matter what their disagreements the chips must flow in order for the global economy to function?

It is just fascinating, this notion of coming to grips with weaponized interdependence with something that is so vital to the global economy and to our day-to-day lives with 90 percent of the supply coming from Taiwan. That would suggest that we all need to make sure that nothing happens to Taiwan over the next number of years.

CHRIS MILLER: If something were to go wrong in the Taiwan Straits and we were to have a disruption or destruction of Taiwan's chip-making capabilities, the results would be no less than catastrophic for the world economy. There is no alternative source of supply, we do not have spare capacity we could easily bring online, and it would take years to rebuild that capacity in other countries.

In the interim it is not just that we would struggle to produce smartphones or PCs. It is autos—a new car will have a thousand chips inside on average, it is dishwashers, it is microwaves, it is airplanes. Good luck producing an airplane anywhere in the world. So again the costs would be horrific. It would be a Great Depression-level disruption to manufacturing. I think the optimistic view is that that does provide a very strong incentive for everyone to keep the peace around Taiwan.

I think that optimistic view might be right, but I am not super-confident that it is right, and I think we can find lots of instances historically where interdependence has not guaranteed peace. All of the disastrous wars that have emerged throughout history were not started because they were expected to be disastrous wars. They developed into disastrous wars after a series of calculated risks taken by leaders of countries, some of which were based on reasonable bets that went wrong and others that were based on just bad bets in the first place. We have one example of this from last year of that going horribly wrong.

There are a million differences between the situation in Europe and the situation in the Taiwan Straits, but I am struck by the extent to which the "don't cut down the olive groves" logic could have also applied to the natural gas infrastructure in Europe, and it was a casualty very, very early on. I worry about putting too much faith into this thesis, not because I think it is not true but because I am not sure that it is true, and that uncertainty makes me nervous.

TATIANA SERAFIN: Thank you for the question from the audience.

We have a question here that I think is important because there are other players in this industry. I stress Taiwan because it has been the thing that I have been talking to people about that they seem not to understand as the number one thing that is not understood, but there are many other things we can talk about. Here is a question from Peter Novitzky, asking about other chip makers. What about AMD, Arm, or Advanced Semiconductor Materials Lithography (ASML)? Where do all of these other businesses fit? The entire book talks about that, so that is a little bit much, but if you can maybe give us a little bit of a road map so that we understand where we are.

CHRIS MILLER: That is actually a great question, and I think it probably helps to lay out a bit of the details as to where different companies fit in. If you talk about the chip industry, you are actually talking about firms that do a whole number of different things because the production process requires companies that specialize in different parts of the supply chain.

What does that mean? First off, there are companies that design the software tools that are used to design chips. Second, there are companies that actually design chips themselves. This is where AMD, for example, fits in. Arm as well fits in roughly in this level. They produce chip designs, but they do not do any manufacturing themselves. AMD manufactures zero chips, it only designs them, then AMD will take its designs to Taiwan, to TSMC, which will then do the manufacturing.

TSMC is only capable of manufacturing these chips by acquiring materials, ultra-specialized chemicals, for example, and also machine tools from companies like ASML, which Peter mentioned. ASML produces one of the types of tools needed to make chips, called a lithography tool, which shines rays of light to create patterns on chips. That is a very, very, very oversimplified description of what it does, but that is what it does. ASML as well produces no chips, designs no chips, and does almost nothing but produce these lithography tools, yet their tools are used in most chip-making facilities around the world and in every single advanced chip-making facility around the world.

Within the chip industry writ large there are actually a lot of firms that do not produce any chips because they either design them, produce chemicals or software, or they produce the tools that are used to manufacture chips, so TSMC has these extraordinary capabilities, which we have discussed, but it can only do so by drawing on the expertise and the products of the rest of the supply chain.

That is what makes this industry I think so interesting. There is not a single company nor is there a single country that can produce advanced chips on their own. It requires cooperation between firms in Europe, Japan, the United States, Taiwan, and Korea, so this international supply chain is core to how the industry operates.

NIKOLAS GVOSDEV: That is a fascinating point because we say that Taiwan produces 90 percent, but then Taiwan's capacity to do this is impacted then by these other suppliers. Again, we are back to these questions of interdependent supply chains and can they be weaponized and so on.

You talked a bit earlier about a lot of cost and complexity. In advance of your appearance with us today, in advance of the Book Talk, Mohammed Soliman tweeted out this morning that, yes, the United States and Europe are going to devote X billions of dollars to building facilities but that China pretty much has tripled their investment, in essence portraying this as a race to see who can and who will build alternate facilities.

As you look at the CHIPS Act, as you look at the proposals in Europe, obviously the People's Republic of China (PRC) is very good at making announcements—sometimes delivery or fulfilment may not always match the initial rhetoric—and then of course over the last year a lot of both governmental and private industry figures have been saying that India is poised to duplicate what Taiwan has done, do you have a sense as you are looking forward over the next five to seven years whether we are going to see the 90 percent figure for Taiwan drop and then it will be 60 percent comes from Taiwan?

Where would the rest come from? Will it come from the United States, from Europe, from India, from the PRC from the mainland, or are there other players in this space that we are not paying attention to that we might be surprised by?

One of our earlier Doorstep guests, Ambassador Charles Ray, said that we overlook Africa's potential as a manufacturer. Could we be talking about what Taiwan did in the 1960s an African state could do in the 2030s? What is your sense as to how the industry itself, and how much do governments put their thumb on the scale to influence those outcomes?

CHRIS MILLER: I think a lot depends on which types of chips you are talking about, also if you look further down the electronic supply chain which types of electronics assembly you are talking about.

If you envision a smartphone, from the production of the chips all the way to the completion of a smartphone there are lots of different companies involved. There is the main processor chip in your smartphone, which is probably not produced at TSMC in most cases, but then there are lots of other chips in your smartphone, some of which require the most advanced production processes and others that do not. In your phone there is a chip for the Bluetooth, there is a chip for the WiFi, there is a chip for the camera, there is a chip for the audio, and not all of those have to have the most cutting-edge production processes.

I think we are likely going to see some reduction in Taiwan's market share at the cutting edge. You mentioned the facility in Arizona that TSMC is building. Intel is also trying to expand its capacity to produce these ultra-cutting-edge chips, but we are not going to see a collapse in Taiwan's role. Today it is 90; maybe it will go down to 70 over some time horizon. That seems plausible to me, but it is not going to go down to 30, 20, or 10. Taiwan will stay very, very important at the cutting edge.

For other types of chips you have different dynamics. In memory chips, which are also very important, Korean firms are actually the largest player. Today Korean firms produce memory chips both in Korea and in China, but actually they are in the process of removing their production from China and refocusing it on Korea. So we could have for certain types of advanced memory more concentration of production in Korea than we do today.

For less-advanced types of chips that require less sophisticated manufacturing processes, we have a huge build-out around the world with India as mentioned focusing on these less-advanced chips, Japan focusing on more foundational chips, and then most of China's investment dollars going to building out these types of foundational chips, which are not the most sophisticated in terms of their production but are in everything—in your dishwasher, in your car, in your smartphone, and in your PC. China is building out huge capacity.

Because everyone is building out capacity in these foundational chips there is a risk that there is overcapacity at the global level, which could set off a new round of trade disputes over the coming years. There is also uncertainty as to which companies and which countries will win market share in the foundational chips.

The next step is to ask, well, who is going to be doing the assembly of all these different chips into the plastic case of your phone or your PC? The assembly process traditionally has taken place mostly in China. Today China assembles around 90 percent of all smartphones and a slightly smaller share of PCs and servers but a large share. That is where you are seeing I think a rapid shift right now because the companies that do the production of phones and PCs are trying to get more diversification, and Vietnam, India, elsewhere in Southeast Asia, and Mexico are going to play a much bigger role in the assembly of all the chips into phones and PCs.

NIKOLAS GVOSDEV: Chris, since you mentioned dishwashers—and I think it is important to mention this as well—we do not think of dishwashers as being particularly that advanced but, yes, they have chips. As we have seen from the Russian invasion of Ukraine people laughed up the idea of, "Oh, they're stealing dishwashers and bringing them back," but if they have chips that can be used, it may not be that you are taking a dishwasher because you expect to wash your dishes, it is because you want to get the chips inside. We do not think that so many of our daily products contain these high-tech products.

TATIANA SERAFIN: We have a question from Deen Chatterjee: "Isn't weaponized interdependence the underlying reason that dictates global geopolitics?" We can get into his further question, "What does that mean for democracy and sovereignty," but I do think that it means that governments are super-involved. As you have mentioned and alluded to, the Taiwanese government was very involved in funding its semiconductor industry in Taiwan. Similarly in China, the Chinese government is making huge investments.

I am going to attack this interdependence question by an example you had in the book of how an IBM or an Apple wants to get the Chinese market, but in order to do that the government gets involved with a tariff, and then that impacts the business of a large company, and there is this dance then that happens: How much can the government get involved in regulating trade, how much should it be involved in regulating trade, or should it let companies just do their own thing? When does it become a strategic security issue? That I think needs to be more understood and talked about. Maybe we can get at Deen's question from that angle?

CHRIS MILLER: It is an interesting, complex, and hard-to-answer question. On the one hand there is a school of thinking dating back to Kant or perhaps earlier that economic interdependence produces peace, and there are plenty of reasons to think that that dynamic is sometimes true.

But we have also seen—and there is some great work by the political scientists who coined the phase "weaponized interdependence;" I borrow it from them, and it is not my own phrase—is it always true that interdependence produces peace, or is sometimes interdependence a sphere where competition can move to? What we have seen in the electronics industry over the past decade is that we have had competition between countries moving to this sphere of electronic supply chains.

This creates I think interesting policy dilemmas because on the one hand it is very clear that businesses have invested a lot of money in their businesses and so lobby their governments to make sure that politics does not get in the way of the business. We see this time and time again. Insofar as that is true business lobbying is actually a stabilizing force because businesses will tell their governments, "Don't mess this up; our business model depends on it." That dynamic is present to some degree.

On the other hand, we also see governments looking at their vulnerabilities, their reliance on other countries, and saying, "We have got to find ways to mitigate these vulnerabilities," which usually means reducing interdependence. So what is happening in the chip industry right now is an effort by the United States and simultaneously by China to reduce their interdependence on each other, and both countries are doing so because they believe they will be in a better competitive position if they are less dependent. It also probably though creates a dynamic in which you have fewer interests lobbying for stability in the relationship, so it creates a more unstable bilateral dynamic, even though both sides are taking steps that they believe will make themselves more secure.

It makes me think about the security dilemma in international security where both sides build up their arms to feel more secure, but the dynamic is an arms race that makes everyone less secure. I think we have similar sets of dynamics in the economics sphere today, where both sides are taking steps to make themselves more economically secure, and there is an open question as to whether the end result is a situation in which we are more economically secure in aggregate after everyone takes these competitive steps.

I do not know that the answer to that question is that we end up being less secure. I think it is a sphere where we are still learning in real time what the impacts are, but I think we certainly should be cognizant of the fact that security dilemma dynamics are present in the economic sphere as well as in the traditional military and security sphere, where that concept was first set out.

Next to that, there is a question of escalation dominance, which is again a concept that was developed in the military sphere but is also present in the economic sphere because we often see in economic relations that you have tit-for-tat dynamics where one country will punish and the other country will respond to the punishing, but then we often see situations where you do not get that and where one country—because it has a larger economy or because of its own political and economic dynamics—feels like it is more capable to mete out costs against its adversaries and not be punished in response.

In the chip industry that has been the U.S. dynamic. The United States has cut off a series of Chinese firms from access to advanced technology, Huawei being the most prominent example, without really suffering much from China in retaliation. That is an interesting dynamic as well. It has encouraged the United States to do more of that, and I think we are going to see whether China continues to basically accept these moves by the United States or decides to push back with its own costs and measures.

TATIANA SERAFIN: Because you said that and because we recently had our TikTok hearings here, do you think that there are enough subject matter experts in the government? In your book if you look at the timeline where there was government and industry working together, they seemed to be in the same conversation, and then there was a dip in knowledge and expertise at the government level here in the United States.

Do you think that we are rebuilding that? Do you think that our congressmen and senators understand what they are talking about? As we are looking ahead to a redo of the 2020 election do these two players understand what is going on? Should we be confident here at the doorstep that our lawmakers understand?

CHRIS MILLER: I think one could always want for more expertise in government on all sorts of tactical issues, but I am struck by the vast improvement in the government's expertise over the last five years. Five years ago it was hard to find people at the upper echelons of economic policymaking and foreign policymaking who had much of an expertise in this industry at all. Now I think what you find is that the people who work on these issues have a ton of expertise, and the typical policymaker knows a lot more than they would have five years ago, so I think we have actually done an okay job at getting up to speed on this issue, although I am sure there are many different ways in which our knowledge could be even better.

TATIANA SERAFIN: Just because both Nick and I went to the School of Foreign Service, I am going to plug the Netflix show The Diplomat, because it is really good, but I am also going to say that on a diplomatic level, because a lot of things signaled at the diplomatic level—you mentioned, how will China respond? They haven't yet. What is going on at the diplomatic level because there are some great stories you have in the book about spying and how Russia spied on the United States in the early days when they tried to get some technology, which by the way did not happen. What is the spying situation? What is going on with that and the diplomatic situation because I do think in this industry it matters, even though it might seem just like a Netflix show. It is very real.

CHRIS MILLER: It is interesting. On the one hand, there has been historically a lot of spying in the industry, which I address in the book. On the other hand, I think it is hard to find examples of where spying really turned the tables. One of the reasons is that unlike a lot of other industries the chip industry improves at exponential rates every two years. That is what Moore's law is all about, a doubling of computing power every two years from new generations of chips, and what that means is that if you are ten years behind you are five Moore's laws behind, which means that you are in the Stone Age. In an industry that improves so rapidly spying can only get you so far.

It is different from many other spheres. I think of aviation, which is a really complex type of manufacturing; they do some extraordinary things in aviation, but I like to imagine, What if planes flew twice as fast every two years? It would be a totally different world. We would be halfway to Mars by the time we have been on this call. In that industry because the rate of improvement is slower spying is more impactful because you have time to integrate what you have learned into your system.

It is still hard to spy in aviation, and aviation engines are still one of the things that it is very hard to replicate, but I think it is more impactful there than in the chip industry, where in the time it takes to spy and to apply whatever you have learned the industry has already jumped forward, doubling its capacity relative to two years ago. It is not a great strategy.

NIKOLAS GVOSDEV: Also as you said, "Great, I have stolen the plans for your latest chip," but if I do not have the capacity to replicate it, that is, that I can go to my factories and say, "Here is the design; now make this"—as you were pointing out the cost and complexity issues are real barriers there.

As we are beginning to wrap up, it is interesting that you titled your book Chip War, that there is this sense of conflict and of competition. Wars can carry with them also the prospect for loss. We talk in the abstract about, "Well, we use chips for everything," and as you said we could end up in a Great Depression situation where a lot of things that we take for granted would not be manufactured—new planes, new cars, and not just an iPhone 20 version or something like that.

This kind of ties back to the expertise question that Tatiana raised: Do you think that realization of how ubiquitous chips are in everyday products—I have always used this example, that MasterCard ad from a few years ago with Camila Cabello essentially with her smartphone buying all of these goods and services while she is walking down a Brooklyn street. People just sort of think these things appear, that goods are made, I go to the store, I need a dishwater, I go to Home Depot and it is there. Do you get a sense from the people you have been talking with that there is this understanding of the level of economic disruption or is it still largely abstract?

You have been on book tours, you have been around the country, and you have been taking questions from people; do you get a sense that the seriousness of the consequences is there or is it still largely an abstract, like a meteor: "Yeah, a meteor could hit Earth and it would be really bad."

CHRIS MILLER: I think we still have some really severe cognitive dissonance over this issue. I think if you were to ask a typical person five years ago and then ask them today what is the likelihood of something going wrong in the Taiwan Straits over the next five or ten years, I think almost everyone would give you a higher probability today than five years ago, but I think if you ask people what costs would you be willing to bear to address that risk today, the answer is, "Well, not all that much."

We see that in the way that companies make decisions, we see it in the way that the government has made decisions, and I think it is basically an irrational result of our difficulty in planning for crises in the future. No matter what probability you put on the worst case scenario the cost is measured in many trillions of dollars, and so you ought to be willing to buy a very expensive insurance policy right now to begin to prepare for that.

We can debate what the best way is to prepare for that, but I think we are not doing nearly enough to prepare for that scenario. Even though I think it is not the most likely outcome, even though I think we still can avoid a crisis, and even though I hope that it does not happen I still think the reality is the chance is certainly above zero and I am afraid it is above a 10 or 20 percent probability over the next five years, so I think we ought to do more to prepare.

TATIANA SERAFIN: One way to prepare—just reminding our audience—is to get Chip War: The Fight for the World's Most Critical Technology by Chris Miller. You will learn a lot about many things. You have done so much research and talked to so many people.

I want to talk about some of your sources because I like to leave our audience always with where they can get more information. Some of the sources that you mention, if people want to just learn more about the chip industry, especially in China, was the Nikkei Asia reference that you made. You did a lot of work with Georgetown University, our alma mater. Is there anywhere else that you found that was really helpful in putting this book together that our audience can follow current developments as they happen?

CHRIS MILLER: You mentioned the Center for Security and Emerging Technology at Georgetown. I would reiterate the fact that they do extraordinary work not only on semiconductors but also all other types of advanced technologies—artificial intelligence, biotechnology—and their implications for security. I learn a lot from them, and I think all of your listeners would too.

TATIANA SERAFIN: Great. Thank you so much for your time. We hope to see you soon for your next book. If you want to share with us, we will be willing to listen to what your next book is, but for now Chip War—read it, buy it, and talk to us online about it. We want to hear from you. Thank you so much for your time today, Chris.

NIKOLAS GVOSDEV: Thanks, Chris.

CHRIS MILLER: Thank you for having me.

Voir la vidéo complète

Carnegie Council for Ethics in International Affairs est un organisme indépendant et non partisan à but non lucratif. Les opinions exprimées dans ce podcast sont celles des intervenants et ne reflètent pas nécessairement la position de Carnegie Council.

Vous pouvez aussi aimer

3 FÉVRIER 2022 - Podcast

The Doorstep: Hacktivisme 2.0, avec Joseph Marks

Dix ans après que des hacktivistes comme le collectif Anonymous se sont fait connaître par des cyberattaques contre des agences gouvernementales américaines et des multinationales, et qu'ils ont ensuite été étouffés par ...

13 MAI 2024 - Podcast

L'exploitation continue du commerce mondial du sucre, avec Megha Rajagopalan

En collaboration avec la Social Justice Academy du Marymount Manhattan College, Tatiana Serafin et Megha Rajagopalan, journaliste au New York Times, discutent des droits de l'homme et du commerce mondial du sucre.

16 DÉCEMBRE 2021 - Podcast

The Doorstep: Evaluation de la "politique étrangère pour la classe moyenne" de Biden/Harris avec Mo Elleithee

L'équipe Biden/Harris avait de grands projets pour réengager les États-Unis dans le monde après quatre années de repli sous Trump. Mais la poursuite ...