#WTH Should I read this summer?

Chip War, by Chris Miller

A small housekeeping note from your WTH team: Normally, in August, Marc and Dany take a short hiatus from pod life. This year, however, we decided to do a short book series with some of our most admired authors about books that, in our somewhat humble opinion, really deserve a read. We begin this week with a pod with our colleague, Chris Miller, on his brilliant book, Chip War. He’s an Associate Professor of International History at Tufts University and a Jeane Kirkpatrick Visiting Fellow at AEI.

In case you’re already thinking to yourself that a book called Chip War might not be for you, note that the New York Times called it, “a nonfiction thriller — equal parts “The China Syndrome” and “Mission: Impossible.” And it really is. An incredible analysis of the business and geostrategic importance of the silicon chip, written for non-tech readers, the book will lay bare for you the critical significance of the island of Taiwan, the implications of falling behind in the Chip War now ongoing (another Great Depression?), and so much more about what is truly a tiny, virus-sized piece of tech magic.

It is actually difficult to overstate how important chips, hailed as the “new oil” of our time, are. This is the single most advanced technology humans have created, and one that doubles in power every other year — a blistering pace. If you care about your business, or your iPhone, or your national security, or your job, or even just your washing machine, this book is for you.

Chip War was FT Business Book of the Year, Economist’s Best Books of 2022, New Yorker’s Best Books of 2022, a NYT recommended book, One of 12 best business books by The Times, a New York Times Bestseller, #1 on Fortune’s Spring CEO Survey of the Best Book They’ve Read in the Past Year and so on.

HIGHLIGHTS

What’s the genesis of Chip War?

CM: I started planning to write a book on the evolution of military power, but I quickly came to realize that if you look at how militaries have changed the past half century, it's by applying computing power to every system that they have. And I started studying, "Well, how is it that we've produced all of this computing power?" And learned in the process about the semiconductor industry, which is an industry that produces chips in all of our devices, not just military systems, but smartphones and dishwashers and coffee makers, and is one of the most concentrated industries in the world with almost all of the world's most advanced chips being produced in Taiwan. And so there was an extraordinary link between the centrality of chips to our military and our prosperity and the concentration of the industry in a region that is an increasingly tense and geopolitically contested area.

Give us a teaser…?

CM: The production of chips is not just about engineers and material scientists. It's also about extraordinary business people that have founded billion and now trillion dollar companies. And it's the story of spies and intelligence agents trying to pilfer technology from their rivals. And so if you think about Mission Impossible, there's actually a shocking relationship between intelligence agencies, defense ministries around the world and the success or failure of their chip industries. And it's something that I really hadn't expected from an industry that I thought was mostly about making chips for my iPhone.

What don’t people understand about chips?

CM: A lot of people talk about the industry as being globalized, and certainly there's a links between the Dutch company you mentioned, the Taiwanese company you mentioned, US firms that are also critical. But it's actually an industry that's really defined by just five countries, the US, Japan, Taiwan, Korea, the Netherlands. There's no other countries that play pivotal roles, and China is actually the world's largest importer of chips because it can't produce cutting-edge chips domestically. China, in most years, spends as much money importing chips as it spends importing oil.

What’s China’s problem?

CM: It's been hard for China for two reasons. One is that the technological capabilities involved are the most complex that humans have ever undertaken. So the chip inside of your iPhone, Dany, will have 15 or 20 billion tiny electronic components carved into it, each one of which is roughly the size of a virus. This is the most complex manufacturing we've ever done. And so in some ways it's not a surprise that China hasn't caught up because there's nothing harder that humans have undertaken. But the second reason is that the way China's tried to catch up has actually been a relatively bad strategy. The Chinese government has tried pouring billions and billions of subsidies into a chip industry to basically copy what's being done abroad and trying to develop it domestically. And copying is a bad strategy in an industry where the technological frontier races forward so rapidly.

Why is there so much concentration in the industry?

CM: I think in the chip industry, what's been really remarkable in terms of why there's so much concentration is that there are huge economies of scale. If you build a bigger factory, you can produce more cheaply. You can hone your technology and therefore win more and more market share. And that's something the Chinese have really struggled with because they've got all of these different subsidy programs, each one run by a different provincial governor. And so they throw lots of money at their industry, but these companies don't have any customers. Whereas, if you look at Taiwan's manufacturing company, the biggest chipmaker in the world, they're serving all of the world's biggest consumer of chips, companies like Apple, Nvidia, AMD, they all go to TSMC to produce. And so they've got scale that nobody can match.

And the second thing is that we often think of unique know-how as being something that you can patent, and in some industries that's certainly true. But in this industry, there's certain things you can patent, and people do. But there's a lot of know-how that you can't really get a patent on but is actually critical to your ability to manufacture at scale. And so if you think, what does it take to have the know-how to produce a 100 million chips for Apple each year, each one of which has 15 billion transistors the size of a virus on it? That's not a process you patent, but that's a type of know-how that's deep in the head of all the thousands of engineers inside of your company, and it's just very difficult to steal or replicate that type of knowledge.

Not all chips come from Taiwan, but almost all advanced chips do. What does that mean?

CM: It's worth noting the most advanced chips are the chips in your smartphone, the chips in your PC and the chips in the data center that manage and process all of your data. So we're all relying on these most advanced chips every single hour of the day. And it does create risks because, of course, China's military power grows and grows. Xi Jinping's volatility grows and grows. And the risk that China tries to blockade or maybe even attack Taiwan, I think, is meaningfully higher today than it was a decade ago.

What would happen if China took Taiwan?

CM: It would really be economically catastrophic. If you look back to the chip shortages of the last couple of years, one thing that people don't generally realize is that the chip shortages of the pandemic era were driven not by declines in supply, but by increases in demand. So the world actually produced more chips each year of the pandemic, more chips in 2020, more chips in 2021, even more in 2022. But demand grew faster, and so in certain industries like the auto industry, there were huge disruptions. And so in a typical car, there's a thousand chips or so if it's a new car. And car companies found themselves missing just one chip, often cheap chips that cost a couple of dollars, but they couldn't produce the cars that they need. And so the auto industry, it's estimated several hundred billion dollars of unsold cars globally as a result of the chip shortage.

Now imagine if we lost access to all of the 90% of the most advanced processor chips that are produced in Taiwan and many more of the less advanced processor chips that are produced in Taiwan. The implications would be far, far worse than the several hundred billion dollars that the auto industry faced the past couple of years. It would be smartphones. It would be PCs. It would be telecoms infrastructure, data centers, but also a lot of goods that don't require advanced chips but still require less advanced chips, of which Taiwan produces many dishwashers, microwaves, cars, toys. Today, almost anything with an on-off switch has a chip inside, often dozens or hundreds. And a huge share of those chips today can only be produced in Taiwan.

Didn’t the CHIPS and Science Act fix this problem?

CM: So I think there are two issues that we're trying and only doing a decent job of grappling with. The first is that there's not a level playing field for chip investment. And so what's happened over the past 30 years is that governments, especially in East Asia, and in quantitative terms, the most important is China, have been throwing billions and billions of dollars at chip firms, both Chinese chip firms, but also global chip firms to invest in China. And so there's been great studies done over the past couple of years finding that US, some 900 companies received more money in subsidies from foreign governments than they did from the US government.

And so if you ask why is there less chip making in the US today than there was 30 years ago, that's pretty clearly part of the answer. And in theory, it would be great to have the WTO or some sort of deal between different governments saying, "None of us are going to subsidize," but that's just not the world we live in. And so the reason Congress passed the CHIPS Act is partly with that in mind, it was an effort to level the playing field, given that all other efforts to restrain other government subsidies, especially the Chinese governments hadn't worked. And the effect had been to encourage US firms and other countries firms to move production to Asia and especially to China. So that was one problem.

The second problem is the deep interlinkage between the entire US electronics industry and China. And this is especially true when it comes to assembling devices. So you produce your chip for your smartphone in Taiwan, but in most cases, that phone is still assembled in China. And these are two related but different issues. I think that the CHIPS Act does do something to level the playing field. It says, "If China's going to pay you to build a factory in China, we're going to offer something roughly comparable so that you don't feel a huge need to move all your production to China." Now, there's inevitable problems with programs like this. They always get politicized. This is a problem of every industrial policy program.

What are the industrial policy problems the Biden administration has introduced?

CM: We've seen the Commerce Department not only focused on chip investment, but also green energy and childcare and taking on other requirements, which I think is a distraction from what the real goal ought to be, which is creating a level playing field that doesn't disincentivize firms from investing in the US. But I think Derek is right, that even if you get a level playing field, that is only a small part of the solution when you've got every US tech firm deeply reliant on assembly in China. And we've seen some moves, I think, by firms who are looking at the landscape, getting nervous, moving production to Vietnam, moving to Thailand, moving to India. But the rate is slower than I think we probably need for security reasons just because today every trillion dollar US tech firm could just simply not operate without chips made in Taiwan and assembly of their devices in China.

What should we have done?

CM: Well, I think Congress and the administration have gotten part of the answer right, which is that you've got to do something to match the tax policy, the tax incentives that Asian governments, especially China are giving. But they've missed that part of the cost differential, why it's more expensive to build chip-making facilities here versus in China is that we impose regulatory burdens that drive up costs.

The big issue in quantitative terms is actually construction. So construction costs in the US are just substantially higher than India and East Asia. And the CHIPS Act, due to the way Congress wrote it, requires the use of union labor in construction. And if you talk to chip firms, they'll say, "Our construction costs are far, far higher than they need to be for that reason." The other aspect is regulation, NEPA, for example. We've talked about permitting in the energy sector, but it's just as bad in the chip sector. And so there's a great study that was recently wrapped up, which found that it takes longer to get the permits to build a chip-making facility in the US than it does in Europe.

Are there other CHIPS problems?

CM: Chip firms themselves have been focusing as you'd probably guess, on getting subsidies rather than pushing back on the cost issues that have been disincentivizing investments. So that that's a real problem. I think the other part is that the administration's rhetoric and also a lot of rhetoric out of Congress is about onshoring. It's about American semiconductors. When you look at what the industry's actually doing though, it's doing exactly what you're recommending, Dany, which is to say, "Let's move production to other countries. Let's actually increase our investment, increase our trade links, increase our tech transfer between allies like Japan, like Europe, like Korea, like Southeast Asia as well."

And so I think there's a mismatch between what the administration's actually talking about, which is ribbon cutting on big factories in the US, which is what they want the new cycle to be about, what companies are actually doing, which is making their supply chains as efficient as possible, while also dealing with the over-reliance on China.

Talk a bit about our military’s dependence on chips…?

CM: Certainly, if you look over the past couple of decades, it is precision strike, the ability to target with a missile hundreds of miles away with almost perfect precision that has been transformative the last 30 years. I think if you look forward to the next couple of decades, it's the application of AI and autonomy to military systems. And you see every major military, US, China, Japan, trying to explore how they're going to be deploying AI to military systems, and it's already happening. We see even in the Russian-Ukraine war ways that both sides are exploring with semi-autonomous or fairly autonomous systems.

And if you want to train a car to drive autonomously, you need a ton of sensors, a ton of data processing, and a lot of very advanced chips on that car. Same thing is true with the drone. You want it to fly autonomously, you need very advanced semiconductors both on the drone itself, but no less critically in the data center that is teaching the drone how to fly autonomously. And so when you look at the types of systems that the Pentagon wants to be buying, they're increasingly autonomous and therefore will increasingly rely on very, very advanced semiconductors that can train AI systems in big data centers.

Chips (or the lack of them) have been a handicap in the Russian war effort against Ukraine as well…

CM: Yeah, it's been really, really interesting looking at the studies that have been done on Russian military equipment that has been captured in Ukraine, and you break it apart and look at the electronics, and it's all stolen chips. It's US. It's Japanese. It's South Korean chips. And partly that's a sign we need to be better at making sure our chips don't get into their systems. And this is a real challenge that I don't think we've done enough to upgrade our enforcement capabilities, but it also speaks to the backwardness of Russian capabilities. They simply can't make modern equipment without access to our chips.

Doesn’t our ability to invent ever more sophisticated chips give us a military advantage?

CM: I think we've got a huge advantage over Russia. We've got a huge technological advantage over Iran. The challenge is that our advantage over China has slipped. We still have an advantage, but it's not nearly as big as it was a decade ago or two decades ago. And that's precisely why this issue has become so much more critical right now. It's that we need to fight harder than ever to keep our advantage and to stop China from using US technology and tools and software to build up its own chip industry. Because the reality is that China used to be 10 generations behind in terms of the chips it can produce, but now it's only three or four generations behind, depending on how you calculate. And if it can access some of our chips by smuggling, it can get closer and closer. And so that's really the issue is that as China's military spending has grown, simultaneously, it's tech capabilities have grown too. They're not as good as ours, but they're getting closer in a way that makes me pretty nervous.

What would happen if China invaded Taiwan and we lose access to TSMC?

CM: Well, I think it's really not an overstatement to say that we're at great depression levels of impact on our manufacturing capabilities. We would struggle to produce cars. We produce basically no smartphones for at least a year, maybe longer, PCs, data centers, telecoms infrastructure. You go across the types of goods we manufacture and they all have chips in them. And not every single one of those chips is made in Taiwan, but Taiwan produces such a huge share of chips and 90% of the most advanced chips. It's just not replaceable in months or even in a couple of years. It would take probably half a decade to fully replace Taiwan chip-making capacity. And so it's really hard to overestimate the economic impact. It'd be the worst economic downturn we've seen in half a century.

Why doesn’t business get that?

CM: I think there are still far too many people who think that the scenario you laid out of a real war is simply impossible. A lot of people just think it couldn't happen. The cost would be too large. And I'm always struck by talking to people in the tech sector, including companies that are fundamentally a 100% reliant on production in China. "Well, why aren't you a bit more diversified?" And I still hear a lot, "Well, they're not really going to do it, are they? It's not really conceivable. Doesn't Xi Jinping... Isn't he really focused on GDP growth?" Well, I don't see evidence for that, but there's a lot of people who haven't gotten their heads around the risk.

And as China makes more and more low-end chips, and as more and more low-end Chinese chips make it into our systems, into our dishwashers, but also cars and airplanes, that risk grows. And I think we've spent the last decade learning about cybersecurity risks and at least trying to address some of them. We've got companies that specialize in cybersecurity. No one has put any effort into hardware security, securing the chips that all of our computers rely on. There's very little focus on that. Companies don't think about it.

Why hasn’t the US government gotten on top of this?

CM: The US government has put a little bit of thinking, but not much into it, and it's a huge problem. I think the challenge is that there aren't great techniques for mitigating these risks. And so what I think we're going to end up doing is just moving towards more of a blanket ban saying, "We don't want any Chinese components in critical systems because we can't verify whether or not they work as they're supposed to."

You're starting to see companies undertake really detailed studies of their own supply chains, learning things about their reliances that they even didn't know when they started. In the past couple of months, we've seen major announcements from companies that make PCs like Dell or HP, that they're beginning to shift their assembly outside of China and deliberately try to reduce their alliance on components from China precisely for this reason because they're getting customers asking them, "How do you know your Chinese-made components work as expected?" And the reality is no one can answer that question with certainty. And so they're beginning to shift not just their final assembly, but all of their components supply outside of China as well.

OK, but that doesn’t excuse the USG ignoring this challenge…

CM: I think that the government for a long time, for some good reasons and some bad reasons, treated these questions as problems for industry to deal with. And in the 1990s and 2000s, when Chinese military power was really quite limited, and the Chinese couldn't cause major problems, the threat to Taiwan was low. That was maybe a reasonable approach to take. But over the last 10 years, Chinese military power has grown dramatically, so has the threat to Taiwan. And so all the supply chains that US companies built over the last three decades now need to be reassessed. And it's taken government a long time to realize that it's not just a problem for these companies and their supply chain operators. It's a problem for all of us because we're all hugely exposed, the US military as much as anyone, to these vulnerabilities that we're only dimly aware of.

Are we doing anything to stop Chinese hardware getting into supply chains? Or stopping US exports to the Chinese?

CM: This is where there's really been a huge change in what the US government's doing. About five years ago, six years ago, it began ratcheting up the controls and the transfer of US technology to China, the types of machines that make semiconductors, for example, as well as certain types of advanced chips. And equally, importantly, it got allies like Japan, like the Netherlands to implement similar controls. And so five years ago, you could import basically any type of advanced chip making tool you wanted to China with very few questions asked. And today, it's vastly harder to import not just the cutting edge, but also one or two generations behind the cutting edge. We're making it really tough for the Chinese to access anything close to the cutting edge. And I think that's the right approach because for too long we were very, very supportive actually of China's technological catch-up. And thanks to that, they caught up to a substantial degree. And it's only really in the past five years where we've turned around in a really big way and said to China, "No, you can't use our technology to pursue your own catch-up strategy."

Are we doing enough?

CM: My sense is that we're not doing enough, that companies still underestimate the risk, that their shareholders don't pressure them to ensure themselves against this risk. And as a result, the entire corporate landscape is still a bit naive about the prospect of potential war in Asia. But I understand as well when companies say, "Well, we can't do it all overnight, or we could, but the cost would be so dramatic that we're just not able to afford it." And so there is a balance to be struck, but I think there's more we ought to be doing because, as you say, Chinese military power grows year after year. And as it grows, the risk of an attack grows too.

Full transcript here.

SHOWNOTES

Related Pieces by Miller

As Chinese Cars Speed into Global Markets, Tensions Will Only Escalate (Financial Times, July 13)

The Chip Patterning Machines that Will Shape Computing’s Next Act (MIT Technology Review, June 26)

U.S. Allies Are on Board for China Chip Tech Controls After All (Nikkei Asian Review, April 4 2023)

Where are we in the Chip War?

US Actions

October 2018: Trump admin cuts off Chinese chipmakers (Fujian Jinhua Integrated Circuit) from its US suppliers

May 2019: Huawei is placed on Commerce’s “Entity List”

January 2020: Trump admin mounts campaign to block sale of Dutch chip tech to China, ASML unable to sell lithography machine to Chinese customer

May 2020: Trump admin blocks shipments of semiconductors to Huawei from global chipmakers

December 2020: US adds China’s top chipmaker SMIC and dozens of other Chinese firms to trade blacklist, denies licenses for SMIC to produce devices of <10nm

August 2022: US passes CHIPS Act

September 2022: Nvidia and Micron say US officials told them to stop exporting AI computing Chips to China

October 2022: Biden admin publishes sweeping set of export controls

December 2022: US adds Chinese memory chip maker YMTC and dozens more Chinese firms to trade blacklist

June/July 2023: Netherlands planning to curb sales of ASML equipment to China, US expected to push for further restrictions

Chinese Actions

December 2019: China begins to stockpile US chips to give flexibility, hardly long-term solution

December 2022: China initiates WTO dispute complain targeting US semiconductor chip measures

May 2023: China bans Micron

July 2023: China places restrictions on export of gallium and germanium starting Aug. 1

Efforts from the Chinese have ramped up as of recent, but ultimately

China struggles to retaliate against U.S. chip controls at the center of a ‘technology war’ (Joe McDonald, Fortune, April 4)

China Can’t Win Its Chip War Against the U.S. Here’s Why. (Tae Kim, Barron’s, April 5)

In Brief: CHIPS and Science Act
The CHIPS and Science Act: Here’s what’s in it (McKinsey & Company, October 4 2022)

Fix CHIPS and Science, or Else (Derek Scissors, AEIdeas, September 9 2022)

US Export Control Regime

Three key notes (Gregory Allen, CSIS, May 3)

Client Alert: United States Creates New Export Controls on China for Semiconductors Manufacturing Technology, Advanced Semiconductors, and Supercomputers in New Phase of Strategic Tech Competition (Gibson Dunn, October 13 2022)

‘An Act of War’: Inside America’s Silicon Blockade Against China (Alex Palmer, NYT, July 12)

China’s Retaliation:

Micron: With Ban on Micron, China Escalates Microchip Clash with U.S. (Paul Mozur and John Liu, NYT, May 22)

Raw Materials: China just played a trump card in the chip war. Are more export curbs coming? (Lauren He, CNN, July 5)

Chiplets/Chipset Technology

Chip wars: How ‘chiplets are emerging as a core part of China’s tech strategy (Jane Lee and Eduardo Baptista, Reuters, July 13)

China’s Strategic Chiplet Advancements Amidst U.S.-China Tensions (Ayush Pandey, Asiana Times, July 13)

Silicon Curtain?

The Cold War With China Is Changing Everything (David Brooks, NYT, March 23)

The US-China chip war is spilling over to Europe (Michelle Toh, CNN, November 25)

Chip 4 Alliance

An overview of the Chip 4 Alliance and its ramifications (Zachariah Peterson, GlobalSpec, December 21 2022)

The Chip 4 alliance will struggle to find cohesion in 2023 (Economist Intelligence, December 8 2022)