Semiconductors are arguably the most important core technology in the modern world. You can’t fully understand the current state of politics, economics, or technology until you consider the role they play. Rob and Jackie sat down with economic historian Chris Miller to discuss the extent to which microchips are the new oil.
Semiconductors are arguably the most important core technology in the modern world. You can’t fully understand the current state of politics, economics, or technology until you consider the role they play. Rob and Jackie sat down with economic historian Chris Miller to discuss the extent to which microchips are the new oil.
Rob Atkinson: Welcome to Innovation Files. I'm Rob Atkinson, Founder and President of the Information Technology and Innovation Foundation.
Jackie Whisman: And I'm Jackie Whisman, Head Development at ITIF, which I'm proud to say is the world's top ranked think tank for science and technology policy.
Rob Atkinson: Yeah. This podcast is about the kinds of issues we cover at ITIF from the broad economics of innovation to specific policy and regulatory questions about new technologies. And today we're going to talk about what I think, and all we say, is the most important core underlying technology, and that's semiconductors.
Jackie Whisman: Our guest today is Economic Historian, Chris Miller. He's associate professor of international history at Tufts University. He's a visiting fellow at the American Enterprise Institute and he's Eurasia director at the Foreign Policy Research Institute. He's authored four books and the most recent is a favorite among ITIF analysts, called Chip War: The Fight for the World's Most Critical Technology. Welcome, Chris.
Chris Miller: Thank you for having me.
Jackie Whisman: So the premise of your book is that microchips are the new oil and, "To make sense of the current state of politics, economics, and technology, we have to understand the vital role played by chips." Can you expand on this to get us started?
Chris Miller: Well, the typical person relies on hundreds or thousands of semiconductors as they go about their daily life, but most of us never see a chip except maybe in a picture on the internet. And so, I found that when I started this research, I was only dimly aware of the extent to which the world, as we know it, wouldn't exist if it weren't for semiconductors. And I think what's more interesting and important than that, is that the production of semiconductors has become so advanced, so complex that it's the hardest manufacturing challenge humans have ever undertaken. And the sophistication in the companies that design and manufacture chips is extraordinary to what's provided us the advances in computing capabilities that we all take for granted. And so, the reason I decided to write Chip War was because I thought that most people were fundamentally reliant on semiconductors and yet had no idea how they were produced, where they came from, or just how critical they were to our lives.
Rob Atkinson: There's so many interesting things in the book, but you have an interesting statistic, I guess. For those who don't know how chips are made, super complicated, but they have essentially lithography, they etch on onto the silicon. And the newest extreme ultraviolet lithography machines, EUV machines, you write, have 457,329 component parts. And the laser itself is just part of 100,000 parts in an EUV machine. So in a lot of ways, this is perhaps the most complicated thing humans have ever done.
Chris Miller: Well, I think that's really accurate. Then what's striking, I think, about the industry is that you need not only one of those machines but multiple other machines that are almost just as complicated to produce advanced chips. And then, even once you've got the machines, you need the expertise to know how to use them. And then you need to know how actually to design the chips you're looking to produce. And so, the design of chips is, in many ways, no less complex than the actual manufacturing. So when you look at the process from start to finish, you can't produce an advanced chip without using intellectual property, design, software tools, chemicals, manufacturing equipment from dozens of companies in a half dozen different countries. And there's no company and no country that can, on its own, produce the most advanced semiconductors. So it really is one of the most complex supply chains, stretching from Europe to the US to Japan, and in many cases Taiwan and Korea as well. In that sense, it's a real triumph of human cooperation between so many different firms and countries. And it would be impossible were it not for the ability to tap into the expertise in all of these different firms and locations.
Rob Atkinson: The other part of that, obviously, is just the continuing process of what people call Moore's Law. Gordon Moore coined that phrase, I believe in the '60s, doubling every 18 to 24 months. And it's almost magic. Arthur C. Clark once said, "Any technology that's not understandable, you think it's magic." It's almost magic that we've been able to keep doing this. It's crazy.
Chris Miller: And I think it's worth understanding Moore's Law. It's helpful to put it in the context of the rest of the economy because as you said, chips have doubled in their computing capacity once every two years or so since the middle of the 1960s. And almost nothing else in the economy has had anywhere near the same growth rate. I mean, imagine if planes flew twice as fast every two years since the 1960s. I would be flying back and forth to Mars on a daily basis. And so, that exponential growth rate has been really unparalleled in the rest of the economy, and it's provided so many of the productivity improvements and advances that the rest of the economy has relied on, which is why the continued progress of Moore's Law, which is getting harder and harder to make work, is so critical to not only the tech sector but the entire economy and really all of human progress.
Rob Atkinson: We want to talk about that a little bit later, but you note how complicated this is and how broad this is, but the US did lead early on. We were the ones developed that, and you're right about that, partly the role of the Air Force in particular with procurement and all. Incredibly complicated question, could take five hours to answer, but I'm going to ask you to answer really briefly. Why does the US not lead in production anymore? We lead in certain parts, but we went from about 42% to say 12 or 13, and it's a big reason why Congress passed and funded the CHIPS Act. So what was behind all of that?
Chris Miller: So I think first off, just to underscore your point, that there are a lot of parts of the chip making process where the US does lead, but certainly in the manufacturing of chips themselves, US market share has declined. And when it comes to the most advanced processor chips, many of them can only be made offshore in Taiwan or in South Korea.
So why is that no longer possible to do in the US? There's a couple of reasons why. First off, in other countries like Taiwan in particular, the conditions for chip making have been more favorable, more favorable tax policy, more government support for workforce development and more favorable currency policy, easier access with permitting, for example, for new facilities. Governments have just tried harder to make it easier to build chip making facilities. And over time, that's had a really big effect in new investment in places like Taiwan and South Korea. So I think that's reason one.
Reason two is that I think the US ecosystem, in general, has underplayed the importance of manufacturing expertise relative to software expertise. And software is undeniably important in a million different ways. But I think when we think about tech, we think about social media and search engines. Critically important, no doubt, but we've underestimated the importance of the types of manufacturing capabilities that make all other tech possible. And so, I think there's been a cultural aspect to this as well, which we need to turn around.
Then thirdly, there've been a couple of companies that have slipped behind that previously were major players and since have fallen behind technologically from foreign competitors. And so, you can look at a number of specific companies in the US which had they been more successful in their technological development or their business models, the entire country would be in a different position today. Now, we obviously can't make policy betting on the success or failure of any individual company, but I think the policy environment in the US has been insufficiently focused on making life easy and making investment attractive for advanced manufacturing firms. And that's something that other partners like Taiwan and South Korea and others have done better than we have.
Rob Atkinson: There was a head of a think tank recently, I can't say who it was because it was one of those private conversations at a dinner, but it was asked, "How much manufacturing could the US lose and still be okay?" To be fair, this was like five years ago. And he said, "All of it. You don't need any manufacturer." And so, that was an extreme version of what you were just saying, Chris, is we didn't take manufacturing seriously. I remember we did an early event years ago on advanced manufacturing policy and somebody came up to me before the event, said, "What are you guys doing with manufacturing? I thought you were a tech think tank."
Jackie Whisman: Silly us. What's your take on the CHIPS and Science Act? Is this going to be the critical policy to jumpstart domestic chip production and technology dominance or is it too little too late?
Chris Miller: Well, I think it's a lot to ask one piece of legislation to transform an industry as big as this one, but I think it's a step in the right direction. Far from perfect, far from probably enough, but I think it does show that the US government is realizing it's been behind in its willingness to take the issue seriously and to provide a supportive environment for chipmakers. And in that sense, it's a really important step. I think the tax policy aspects of it, providing larger investment tax credits, for example, are really meaningful in terms of leveling the playing field between opening new facilities in the US versus in other countries, especially Asia. I think the workforce development aspects of the CHIPS Act are going to be absolutely critical in convincing firms to open new facilities here.
Then I think the R&D focused provisions could be really transformative. I mean, the US government, although not every R&D investment that it makes works out well, some of them have historically been really profoundly impactful. And I think in the chip industry in particular, you can find a lot of instances where the US government has invested in R&D and really driven technology forward in a way that has made a difference. And the earliest chips, of course, emerged in no small part out of defense industrial investments in microelectronics. And the defense industrial base has been a key driver of advances in chip making technology and chip design for several decades. And I think it's a good thing, in general, to see the US government recommitting to these investments via the CHIPS Act.
Rob Atkinson: Yeah. It reminds me of another salon dinner I was in where I was doing my typical thing of talking about how critical the role US government has been in certain technology areas, not all, but many of them. And somebody from, shall we say, free market think tank, said, "I think we've had enough of this stuff about how government plays a role in technology. I mean, just look at the internet, government didn't have anything to do with the internet." And I obviously didn't understand where ARPANET came from and TCPIP came from and all. But even Google, for example, was originally done on an NSF grant that Sergey Brin was working on. So I think your point you make in the book about the role of government is critical.
The other thing I wanted to ask you about is, I've seen this kind of push back against the CHIPS Act where these people say, "Oh. These companies are already profitable. Why do we need to subsidize these profitable companies? They're not losers." Which makes me think, do you want to subsidize losers? It doesn't seem like that's the company you want to subsidize. But you talk about the cost differential and that's part of what CHIPS is trying to respond to. Could you say a little more about what's that intended to do?
Chris Miller: Well, there's been some great studies done of the cost of opening a new facility in Korea or Taiwan or China versus cost of opening a similar facility in the US. And what they universally find is that it's more expensive to operate in the United States for a number of reasons. Tax reasons, land reasons, permitting reasons, direct government subsidy reasons. And that explains a big chunk of why manufacturing has seen more investment offshore than onshore over the past several decades. And I think there's no reason to think that this dynamic will shift unless the policy environment shifts too, which is what the CHIPS Act is responding to. And I think, especially in the context of China's tremendous investments in its own semiconductor industry. I mean, China is spending a CHIPS Act a year and has been for the last decade in its own chip industry. And although Chinese firms are less technologically advanced, if they keep spending that amount and keep building out their capacity, they will win more market share and they will move up the value chain slowly but surely. And that's been the trend the past decade. China started far behind, now it's still behind but less far behind than it used to be.
And so, I think we need to take that seriously. We can't just write that off and assume it will never really dent into advanced cutting edge technologies. Because actually, Chinese firms have shown that they've got some real capabilities, especially when they're given blank checks from their government. And so, when China's spending tens of billions of dollars a year subsidizing its ship industry, I don't think we can stand by and do nothing. And in that context, I think the tens of billions of dollars over five years that the US CHIPS Act intends to spend looks measured by comparison to what our adversaries are doing.
Rob Atkinson: What's interesting about that is if you just look at history, which is what you do, you wouldn't have thought Taiwan would've been a leader in chips in the '70s but they cut a deal with RCA to license their technology. You wouldn't have thought South Korea would've been a leader in the early '80s, but they focused on it. So the idea, somehow, that these countries can't catch up really if they're willing to put talent and money behind it, to me, is one of those kind of American delusions we have. And we've seen the results, we've seen it in 5G equipment and we used to be the world leader in 2000, now we got pretty much nothing. So I'm glad you're making that point. It's almost like a warning cry to say, "We can't just sit in our laurels. We can't just assume the Chinese..." Yeah. They've had a bunch of failures recently, but doesn't mean they're going to give up or it doesn't mean they're going to fail. We don't know.
Chris Miller: Yeah. And they've had failures, they've also had successes. And I think that's really worth... I mean, I think that the challenge that the Chinese face now is that the way that the South Koreans and the Taiwanese caught up was by integrating really, really deeply into international supply chains by working really closely with customers, with component suppliers in Japan and the US, and Europe. And right now, China is facing more and more difficulties on that front than ever before due to policies in Washington with export controls, also policies in Beijing, more state control of the industry there. And so, I think it'll be more difficult than ever before for China to try to replicate the Korean or the Taiwanese model, but that still doesn't mean we should write them off and just give them the amount of funding going in and give the amount of existing expertise in the Chinese chip industry. There's a real strong base to build from.
Jackie Whisman: What do you make of National Security Director, Jake Sullivan's speech on this issue and the recent, more aggressive, Biden policies around chip manufacturing export controls on China?
Chris Miller: Well, it's clearly a big expansion of export controls in the chip sector, and more importantly, a big change in strategy. I think there's a really delicate balance to be struck here. On the one hand, we've got to have a market that is open enough for US-based firms to sell their products to as much of the world as possible. That's how companies invest in R&D, is by selling globally and profiting from those sales. On the other hand, it seems, to me, hard to justify providing China with tools that will be used to make components that will go into both intelligence and military systems at a time when our military advantages over China are deteriorating dramatically. And that's what every report that the Pentagon produces on Chinese military power concludes.
And so, I think the Biden administration has done a reasonable job of trying to weigh those balances carefully. I think they put a lot of serious work into the export controls, trying to understand exactly what the impacts would be. I know industry is cognizant of the economic cost that some of the controls will bring, but the reality is, we've always got to strike some balance between the interest of industry and the security concerns. And I think that's something that the Biden administration did a reasonable job of striking a balance there.
Rob Atkinson: I mean, one of the advantages the US has is, its reach is so broad, it can tell Korean firms not to export to China if they included some kind of American inputs and components. What I worry about is two things. One is that eventually countries are going to say, "Hey. We don't really like that and we're going to divest out of the US components 'cause we just don't like that level of control." I was talking to a European firm a couple years ago. I was over there in their factory and it's a machine company and they make machines that include US semiconductors, and they sell them around the world, including China. And they're really worried that they were going to get swept up in a ban. And the chief engineer told me, he said, "We're just thinking of re-engineering around that and not having American components in our machines." That's one thing I worry about. The second thing I worry about is diversion or other countries not buying into this and somehow they're able to get what they need, and therefore China still gets to go ahead a little bit slower, of course, but we get hurt. What are your thoughts, Chris, to both of those concerns?
Chris Miller: Well, I think those are absolutely valid concerns, and in a lot of sectors they would make me really hesitant about tightened export controls. I think in this sphere of the types of advanced chips that have been banned, so GPUs intended for use in data centers and in semiconductor manufacturing equipment, I think there are market dynamics that make me less concerned about both those factors. In terms of getting around designing out US components, I think it'll be hard in the semiconductor space. US components are just so central to the manufacturing equipment that's needed and there's such a small number of firms in just really three countries, Japan, the US and the Netherlands, that it seems pretty plausible we'll be able to prevent a large scale designing out. And I think there's a big effort underway right now to get the Japanese and the Dutch governments to sign on board to some sort of similar set of controls. And I certainly hope that they do, but I think given the small number of firms and the small number of tools we're discussing, the ability to design out US components is going to be something that we can monitor and it'll be hard to do at all.
In terms of diversion, I think there's a bigger challenge with that because chips are small and they're high value, so it's pretty easy to smuggle them, as we've seen with all of the US and Japanese and South Korean chips that are used in Russian missiles that have been captured in Ukraine. These aren't legally acquired by Russian [inaudible] industry, they're smuggled in. But the new US controls in October, they target chips that are installed in vast data centers. And so, the number of facilities that they're installed in China is actually quite small. It's a reasonable number of facilities that the intelligence agencies will have a new task in tracking what's inside of Chinese data centers, but it seems like a very surmountable task to know whether or not banned chips are installed. So I think the commerce department, which enforces these controls, is going to have more work to do. Any set of controls is never perfectly flawless, there's always some amount of leakage. But my sense is that the Biden administration actually did a reasonable job designing these controls to mitigate the two dynamics that you mentioned.
Rob Atkinson: It's our sense as well. I'm glad to hear you elucidate that. One of the things that I hear from, if you will, super China hawks, because I think we're sort of China hawks, but we're not really extreme China hawks, we still want to sell tide detergent to the Chinese. And some of these folks say, "Well, we shouldn't be selling them any chips at all." And to me, that goes way, way overboard because we want to sell them. If we sell them $10 of chips, that's $10 more that American companies get to do R&D and get scale. It's $10 less than they have. So that's where I think your point about the Biden administration kind of getting it right, going right down the top of that knife-edge curve, if you will, is we don't want to do too much and say, "Well, you just can't buy chips at all," because that would hurt us too much. But if we keep it focused, we can do more damage to them than to us. Is that your sense as well?
Chris Miller: Yeah. I think that's right. And if you look at the share of chips that's impacted by the recent export controls, that's a really tiny share of chips. All chips in smartphones, all chips in PCs, none of these are affected by the controls. It really is just high-end chips used in data centers, and everything else is left unimpacted. And there's good reason for leaving everything else unimpacted because for many types of chips, the low-end, China can produce them domestically. And so, if we ban US chips from selling, China will just build up its own domestic capabilities, and at the low end, it's able to do that. And similarly, the question of, "Can we get allies on board," it'll be a challenge to get allies on board with these focused high-end chip bans, but it'd be much more difficult to get allies on board if we go into lower end chips as well. And that's why I think a broader ban doesn't make much sense, even though this more targeted high-end policy probably does.
Jackie Whisman: We have one last question to close, and it might be a little too complicated for our two minutes that we have, but we'll try our best. One of the key questions in terms of the US and allies leading is if Moore's Law slows down. If it doesn't slow down, it will be harder for China to catch up. If it does slow down, it'll be easier for them to catch up. Your thoughts on this?
Chris Miller: Well, I think one of the interesting dynamics of Moore's Law is that every 10 years there's a new influential chip industry figure who predicts the death of Moore's Law in 10 year's time. And you can go back historically and find brilliant scientists and visionary CEOs who have predicted Moore's Law's death, including Gordon Moore himself, who was unsure in the early 2000s whether Moore's Law would continue into the 2010s. And sure enough, it did. So I remain pretty bullish about Moore's Law continuing. There are so many different strains of research underway that could produce the advances that will deliver us a pathway for continuing Moore's Law past 2030. I just have to think we're going to find ones that will be cost efficient and plausible to actually undertake. But certainly, there is a risk that when we get past 2030, it just becomes too hard to keep shrinking transistors further. But my guess and looking at how this debate has played out historically, is that the current pessimist will be proven wrong and that will have plenty of advances well into the 2030s that'll let us keep shrinking transistors and making chips even more powerful.
Rob Atkinson: And to your point, that's one of the big reasons why the 12 or $13 billion in the CHIPS Act for advanced R&D is so critical. It's kind of like saying, "If we're going to win a race, we got to be able to run faster and not just hope that the other guy slows down." So the R&D component is critical.
Chris Miller: Yeah. I completely agree. And I think the US now has settled into a strategy of both running faster and making sure it's not providing critical equipment that's supporting China's effort to run faster. And striking that right balance is going to be critical going forward, but it's important to focus on both pieces of that equation.
Rob Atkinson: Well, that's great, Chris. Really enjoyable, really interesting. And I encourage everybody to buy a copy and read a copy. When I was urging people with my book, I said, "I don't really care if you read it, I just want you to buy it." But in your case, you should read it and buy it or buy it and read it.
Jackie Whisman: It's the book of the month at the ITIF book club, I think.
Rob Atkinson: There you go. So anyway, thanks, Chris.
Chris Miller: Thanks so much for having me.
Jackie Whisman: And that's it for this week. If you liked it, please be sure to rate us and subscribe. Feel free to email show ideas or questions to podcast@itif.org. You can find the show notes and sign up for our weekly email newsletter on our website itif.org. And follow us on Twitter, Facebook, and LinkedIn @ITIFdc.
Rob Atkinson: We have more episodes and great guests lined up. New episodes drop every other Monday, so we hope you'll continue to tune in.