---

Chris Miller is an associate professor of international history at the Fletcher School of Law and Diplomacy at Tufts University, Jeane Kirkpatrick visiting fellow at the American Enterprise Institute, and Eurasia director at the Foreign Policy Research Institute. He is the author of several books on Soviet and Russian politics and foreign affairs, as well as the recently released Chip War: The Fight for the World’s More Critical Technology (2022), which was excerpted in The Wire last week. In this lightly edited Q&A, Professor Miller discusses the historical roots of semiconductor competition, how the U.S. came to adopt a more aggressive technology policy in relation to China, and what the recent restrictions on U.S. chip exports and expertise mean for China’s chip industry.

Q. You are a historian by training, and previously wrote books on Russian foreign policy and domestic politics. How did you decide to write a book about the history of the semiconductor industry?

A: I actually started this book intending for it to be a history of missile systems and the Cold War. We know that the Soviet Union had a lot of scientific and technological abilities, but they weren’t able to develop guided missile systems with anywhere near the accuracy or the capabilities of the United States. 

If you look at the big shift in military technology over the second half of the 20th century, it was a revolution in military affairs defined by precision targeting. That’s something that the U.S. could do, and the Soviet Union couldn’t, and it explains the military gap that opened up. I wanted to understand why that happened. 

The answer to that question led me to dig into the history of guidance computers in missiles, and it turns out that the challenge of miniaturizing computing power so that it could fit in a missile guidance computer was one of the core engineering challenges of the early Cold War. It led to a number of early important innovations, one of which was a type of circuitry on silicon that we now know as a chip. The first big order for chips was for the Apollo spacecraft guidance computer, the second was for the Minuteman II intercontinental ballistic missile, so there’s a deep relationship between chips and missile technology.

I was doing this research about five years ago just as the U.S. restrictions on ZTE and Huawei were being put in place, and I was reading the news about the extent to which a small number of companies and countries produced chips on which the rest of the world relied. And I began to string together a realization that actually this technology that was forged in the early Cold War and proved decisive in the arms race was actually very relevant today too in terms of U.S.-China relations. 

And the third data point that caused me to write this book was looking at trade flows in Asia in particular. I had a rough sense of what globalization looked like in terms of the types of goods that were traded, but when I realized that China spent more money importing semiconductors than it spent importing oil, I realized that my image of globalization was at best horribly incomplete. 

… we’re headed back towards the former policy where competition is seen to be important and close enough that key technologies have to be controlled a lot more strictly than they had been before.

We’re seeing the Biden administration take a very aggressive approach to technology policy, a change that began under Trump. But the U.S. has historically adopted a protectionist and aggressive approach to technology policy surrounding chips, including towards the Soviet Union and emerging competitors such as Japan in the 1980s. Could you talk a little about that period, and how it changed?

If you step back and ask yourself when do countries restrict technology versus share it, I think what you find generally is that they tend to share it more with their friends than with their competitors, and that they’re more willing to share when they’re so far ahead that sharing doesn’t seem scary. Where there’s a race going on for technologies that are thought to be important, there’s a lot more sensitivity as to who gets what. During the Cold War there was certainly a sense of really substantial competition between the U.S. and Soviet Union; not only in the sense that both countries saw each other as a threat, but also that they both thought they were fairly closely matched in a lot of key parameters. 

So the U.S. imposed really wide-ranging restrictions — and pushed allies and Europe and Japan to do the same — to restrain, in a really broad-based way, Soviet industrial, and then later computing, potential. So there were restrictions about types of machine tools that could be transferred to the Soviet Union, types of computers, types of semiconductor manufacturing equipment throughout the Cold War. 

When the Cold War ended, it produced a couple decades that were unique, because it left one country in the world that was by far the most technologically advanced and militarily powerful, such that everyone forgot about geopolitical competition for the better part of two decades.

Then, as the U.S.’s military and technological advantages waned, suddenly the question of who’s got what technology became more relevant, in a way that it really hadn’t been in the 1990s or 2000s, because of the unipolarity that had existed. 

If you look at [National Security Adviser] Jake Sullivan’s speech on technology policy from a couple weeks ago [September 16th], we’re headed back towards the former policy where competition is seen to be important and close enough that key technologies have to be controlled a lot more strictly than they had been before.

There was a really interesting quote in your book, from a Trump appointee repeating something an Obama official had allegedly told him during the transition about China’s technological advances: “This is really important, but there’s nothing you can do.” Could you talk about how the Trump administration laid the groundwork for a more combative approach to technology policy? 

The national security bureaucracy and Congress over the last decade has been rethinking its assumptions about the relationship with China, and how to make sense of that. So separate from the changes to the presidential administration, you had a broad trend in U.S. establishment thinking about China, which has been to think more about risks, to focus more on the military power balance, and to worry more about Chinese intentions. That’s a trend that has been present across administrations from late Obama, through Trump to Biden. 

What was distinct between the late Obama period and the Trump period was that while people serving in the late Obama administration were already focused on some of these key issues, there was more willingness in the Trump administration to take costly steps with repercussions in other spheres. 

The [Trump-era] tech export controls [between 2018 and 2020, which forbade U.S. companies from selling technology and software to companies including Huawei and ZTE] were an example of that, addressing concerns that had been in place for some time. For example, the possibility of using export controls against Fujian Jinhua was not something new. 

What was new was the willingness to take this step and risk potential Chinese retaliation, risk loss of sales for the firms that were selling to Fujian Jinhua, and other downsides. The Obama administration had always been much more hesitant to accept this risk and more willing to tolerate the status quo.

[Note: Fujian Jinhua is a Chinese state-owned manufacturer of memory chips which was accused of stealing trade secrets from American chipmaker Micron. After Micron sued Fujian Jinhua for patent infringement, Fujian Jinhua counter-sued Micron in a Chinese court and won. In response, the Trump administration imposed tough export controls on Fujian Jinhua.]

You profile many Chinese chip companies and their origins, including Semiconductor Manufacturing International Corporation (SMIC), arguably China’s answer to TSMC, founded by Richard Chang, the Taiwan-raised, U.S.-educated chip industry veteran. You write that at the start SMIC had all the ingredients for success. So why isn’t SMIC the giant today that it had the potential to be?

There are three main reasons. First is that it’s just really hard to compete against TSMC, because it’s done such a good job over the past decade executing on its technological advances and its business model. Every company that’s tried to compete against TSMC in the business for contract manufacturing has struggled. So SMIC isn’t unique in that regard. 

Second is that in the latter years of the 2000s and the early 2010s, SMIC faced financial issues, including from an intellectual property lawsuit with TSMC, and the Chinese state got more involved than it had been before. In the early days, the Chinese state was not that involved — investors then included Goldman Sachs and Toshiba and it was really more of an international, market-oriented company than it is today. But the state got more involved, helping to bail out the firm, but this also meant that the state interfered more with SMIC’s trajectory, and that had an impact on the ability of the company to focus on technology and its ability to serve its customers, inhibiting its growth. 

Third, more recently over the past five or so years, SMIC can’t access advanced technology and tools needed to make chips. Going forward this will be a huge impediment to commercially viable production for the firm, because it’s going to struggle and probably find it close to impossible to produce commercially viable, advanced chips. 

The thesis of China’s industrial spending in the past seems to have been to tolerate large amounts of waste if it means that a frontrunner can emerge. But the recent roundup of chip executives for alleged corruption suggests that there may no longer be an appetite for that. You wrote about some of the most flagrant cases of wasteful spending by companies like HSMC and Tsinghua Unigroup. How do you see it?

The challenge for the Chinese government is that chips are its biggest import and it’s hugely reliant on foreign policy rivals for this core technology; so if it’s going to pursue the current foreign policy, then the chip dependency is a huge risk.

The question is, what can the government do about it besides spending a lot of money? Well, it’s hard. In addition to spending a lot of money, China could also try to integrate deeply with international supply chains and learn from them and move up the value chain slowly. That strategy was viable 20 years ago, and it worked in some cases, like for SMIC in its earliest decade. But today, that’s not a viable strategy for Chinese companies because of the restrictions the U.S. has put in place. 

…the question Chinese leaders will be asking themselves is, is there a sector of the economy where they can cut off goods in a way that hurts the U.S. more than they hurt China? 

So it’s not crazy for the Chinese government to want to spend a lot of money. So long as its foreign policy is where it is, China’s going to face restrictions on its chip industry, it’s got to find some way around that. The government is in a complicated situation because spending money alone is probably not going to solve all the problems that China’s chip industry needs to solve. 

So much in this industry changes quickly, and your book is certainly timely. But at the time of publishing you write in your book that the U.S.’s assault on China’s tech firms has been a “limited strike,” and that it’s surprising China hasn’t done much to retaliate. Since then it’s clear that the “limited strike” has expanded significantly in scope. Do you think it’s likely that China will now retaliate, and how can it?

Chinese leaders have thus far concluded that retaliation has more cost than benefits, which is probably correct. But there’s still a lot they could do: they could restrict the export of certain goods to the United States; they could cause problems with Apple’s production processes; they could restrict the export of rare earths. All these things are possible. 

Export controls work when they’re more costly to the person you’re trying to punish than they are to you. And so the question Chinese leaders will be asking themselves is this: Is there a sector of the economy where they can cut off goods in a way that hurts the U.S. more than they hurt China? 

Thus far, they’ve concluded no. Maybe this time they’d be more willing to try it because they’re more willing to accept some pain domestically, just to show they’re willing to impose pain on the U.S.. That certainly is a risk. But the conclusion thus far for Chinese policymakers has been to not retaliate.

One question going forward is how these export controls will affect the Chinese tech sector, whether it’s cloud computing or companies that rely on artificial intelligence. One question is whether China’s regulations on accessing foreign cloud computing will allow its companies to take advantage of more advanced computing abroad. Can a company like Alibaba move all their data to servers in Singapore and do their computing there using advanced chips that can’t be imported to China, and then have the result brought back into China? This raises a set of interesting questions about who supervises that process, and the impact on data protection and security laws. This is an open question that China has a major say in, if it wants to do that.

Some analysts have suggested that China might retaliate by ignoring intellectual property rights or by trying to reverse engineer technologies in order to get ahead. But you paint a skeptical account of how successful that strategy is, drawing on the example of the Soviet Union’s attempts to copy U.S. technology in the 1980s. How do you see it?

You can do it in certain spheres, but that’s hard to do across the chip making process. So, for example, you need certain ultra-purified chemicals to make chips, but you don’t need IP to make the chemicals themselves. No matter how much IP you have, you still can’t make the chemicals. The same thing with machine tools: you can try to reverse engineer them, but the reality is you’re talking about some of the most precise machinery ever made. So it’s really hard [to replicate these technologies], and if it was easy, people would’ve done it. So I don’t think intellectual property restrictions are the key issue. In certain subsets of the chip design space that might happen, but I don’t think that’s the key challenge. 

The key challenge is it’s really hard to reverse engineer the machinery, and it’s really hard to know how to use it effectively in a fab, which is what TSMC is so good at. There’s a lot of unique knowledge in chip design firms in terms of how to design chips to maximize performance, and replicating all this requires really niche skill sets that a small number of companies have access to, and that the rest of the world doesn’t. It doesn’t matter how smart you are, it’s really hard to replicate this type of knowledge unless you’ve been trained in a small number of institutions that have it. 

So it’s going to be pretty difficult to replicate all of this domestically.

Is there any way that these restrictions can come back to bite American companies? We’re seeing firms like Intel cutting their capital expenditure in light of the global slowdown in chip demand. Will limiting chip companies’ China sales damage their ability to invest capital on reshoring in America?

It’s definitely going to be costly to the companies in question in the short term, because for a lot of companies that design chips, or make chips, or make chip making equipment, China was an important market. It’s not disastrous for any of the companies involved, but is it costly? No doubt. You’ve seen that reflected in their stock prices, which fell after the restrictions were announced. In the longer term, the risk is not the loss of China sales as much as the potential rise of Chinese competitors. 

…a key source of knowledge transfer in the industry is not so much textbooks, or things to be written out or copied and pasted, but knowledge built up by individuals over their careers, especially in these really niche areas of chip making.

And here, if you look at different segments of the chip market: in the chip design space, it’ll be easier to see Chinese designers begin to replace some U.S. firms. In the machinery space, it’ll take a long time. It depends on how easy it is to imagine competitors. But more competition is definitely the risk of this. 

A recurring theme in your book is of talented individuals spearheading transformations in the industry: Morris Chang, Akio Morita, Gordon Moore, Andy Grove and so on. What are your thoughts on the competition for talent, and the effect of recent human talent controls like the new restrictions on American persons working with Chinese industry?

It’s going to be significant. As we’ve discussed, a key source of knowledge transfer in the industry is not so much textbooks, or things to be written out or copied and pasted, but knowledge built up by individuals over their careers, especially in these really niche areas of chip making. A big chunk of China’s chip industry grew with deep connections with the outside world — with Taiwan, Singapore, and the U.S. There are many people in the industry in China that have multiple passports, or some sort of exposure to the U.S. that leads them to get caught up in these rules. So people will have to make a choice to stay or leave their current jobs. It certainly will break some of the ties that currently exist between the international chip industry and the chip industry in China. 

We saw a little bit of this in Taiwan, which has also been tightening its rules about the ability of people who work in sensitive sectors in the chip industry to then work in China. It’s in response to a similar concern that the easiest way to transfer knowledge is by hiring individuals that have worked in leading firms, and bringing them to firms that are closely, but not yet, leading and trying to have a catch up that way. Certainly what the U.S. government was trying to accomplish was to make that knowledge transfer via individuals harder. 

---