Decapitating Huawei & Recapitating America

Interview with Tufts Professor Chris Miller

Donald Trump’s trade war with China is a monkey wrench dropped into the machinery of globalization. It has disrupted international supply chains, making it far more difficult to manage both how, and where, everything from soybeans to semiconductors get produced.

Over the years, the United States gradually lost control of entire segments of the semiconductor supply chain. You can argue how much of an effect that’s had on the diminishing competitiveness of U.S. chip companies, but the fact is that the competitiveness of U.S. chip companies is diminishing. The trade war both exacerbates and emphasizes that problem.

One of the concerns is that the U.S. has limited capacity to manufacture the most advanced chips domestically. Intel is the last of the leading-edge manufacturers in the U.S., but it stumbled recently.

The U.S. semiconductor industry has persuaded Congress to help. Several proposals for legislation have been bundled into a bill being worked on now called the CHIPS for America Act. That’s a tortured acronym, by the way. It stands for Creating Helpful Incentives to Produce Semiconductors for America Act. The CHIPS Act is focused largely on manufacturing, but it is also considering support for other aspects of semiconductor manufacturing, including packaging, which is becoming very, very important.

Intel, which has been carrying the burden of maintaining world-class production capabilities in the U.S. pretty much alone now, is lobbying for Congressional support under the CHIPS Act.

Last May, the Administration induced TSMC, the world’s largest IC foundry, to say it would consider setting up a fab in the United States. TSMC is looking for some Federal support too, however, before it makes its final decision.

And if Congress is going to spend money to support domestic manufacturing, the last large foundry based in the U.S. – GlobalFoundries – believes it deserves a slice of the pie, too. As do other foundries, private manufacturers from the very large, such as Micron Technologies, to the very small, such as startup Spin Memory.

But at least one influential scholar is asking: Is manufacturing the best place for the country to spend its limited budget?

Chris Miller is a professor at Tufts University and an economic historian who is rapidly gaining an international reputation for his expertise in international trade and relations. He published his first book in 2016. “The Struggle to Save the Soviet Economy” is about Russia after the collapse of the Soviet Union. He followed that up two years later with “Putinomics: Power and Money in Resurgent Russia.” He is currently working on a history of the semiconductor industry.

Our EE Times colleague George Leopold recently quoted Chris Miller in an article about the CHIPS Act. George invited him on to this podcast to have a longer conversation about the trade friction between the U.S. and China, and how the U.S. should support the semiconductor industry. Here’s George:

GEORGE LEOPOLD: Semiconductors have emerged, to use the military term, as a chokepoint, pitting U.S. technological superiority against China’s determination and deep pockets to build an indigenous system. You’ve argued that, in trying to decapitate… the U.S. efforts to decapitate Hauwei, that’s sort of a signal of a decline in American technical prowess. In your view, what’s the best way to meet the tech challenge from China?

CHRIS MILLER: I think you can understand what the administration is trying to do vis a vis China. It’s worried that China is catching up in a number of key technologies, semiconductors being one of them, and it wants to keep the gap between the U.S. and China as large as possible. And one way to do that is to reduce China’s ability to make progress. In many ways, there’s nothing new about that strategy. There’s long been export controls on certain types of technology that are needed to manufacture semiconductors. The environment I think has been much less intense in recent years compared to earlier decades.

But in addition to the strategy of trying to hold China back, there’s also a strategy of trying to push U.S. innovation forward, and I think an optimal strategy includes a bit of both. But I think right now it’s fair to say that government is focused a lot on holding China back and not nearly as much as it ought to be on pushing the U.S. forward. So that raises the question of, “What can we do better to foster a climate where you get more innovation and more investment in this type of technology in the United States?”

GEORGE LEOPOLD: Right. So it seems that there’s sort of a bipartisan consensus now that we need what Jim Lewis calls the next generation industrial policy, and the expression of that, of course, is the CHIPS Act, which has been folded into the National Defense Authorization Act. Chris, what are you hearing in terms of where that stands? Is the conference meeting? And then what issues are the House and the Senate conferees grappling with?

CHRIS MILLER: There have been a number of twists and turns in the debate so far regarding what actually makes it into the final Bill. I don’t know that I’ve got the crystal ball to say what’s going to come out of that process. Obviously the NDA [National Defense Act] is always one of the more contentious pieces of legislation that eventually gets passed. It always eventually does get passed. But I think there are a number of different schools of thought as to what’s most important. One school of thought is to say, “We’ve got to find a way to increase fabrication of semiconductors in the U.S., which some people think requires putting more government money behind firms that are willing to build or further invest in existing fabs in the U.S.

I see the logic behind that. There’s no doubt that there are spill-over effects from having some fabs into other parts of the semiconductor supply chain. But it also seems to me that if you look historically, the U.S. government has been relatively good at funding an ecosystem when it wants to fund an ecosystem, and less good at targeting specific parts of the supply chain or specific types of technology, specific industries. So when I look back at the past half century of government support for semiconductor technology, it seems to me that the ones that are the most successful at finding commercially relevant products are not those policies that have picked out a specific type of activity and said, “Let’s fund that.” But rather, there are strategies that said, “Let’s find a way that supports the broader ecosystem.”

GEORGE LEOPOLD: You’ve questioned parts of the CHIPS Act, especially those emphasizing reviving domestic U.S. manufacturing. If I read it correctly, you’re talking about, Let’s put our basic research money into next-generation technologies.

One of the examples that we’ve reported on, you’re aware, I’m sure, is DARPA’s electronic resurgence initiative. Is that one example of a way forward, a good investment by the U.S. government?

CHRIS MILLER: I think that’s right. There’s no doubt that an increase in domestic manufacturing would be a good thing. I think the question is, Should that be where our marginal dollar goes? Especially at a time when we’re likely to see fewer dollars left around to spend on different industrial priorities. So given that, I’d rather see money put at the leading edge rather than put into potentially mature technologies that probably will be important in the future but probably not at the cutting edge. And probably not where most of the dollars are made.

I think one interesting thing to look at is, where are there going to be extraordinary profits made ten years down the road? And it’s probably not going to be in doing things that are easy to do today, or even things that are hard to do today will, in a decade’s time, be relatively straightforward.

So what I’d rather see is people looking and saying, How can we invest in what’s going to matter most in a decade or two decades? That’s where you can help build firms that are going to have a competitive advantage vis a vis their competitors, and also how you can support the industry as a whole. Because you ultimately need not just firms that are at the world standard and can compete with Taiwan and China and South Korea, but firms that are driving ahead of the world standard to really attract resources and really attract talented people into the industry.

GEORGE LEOPOLD: I noted yesterday GlobalFoundries had a media briefing, and the subject of the CHIPS Act came up. The word “subsidy” was used, and the CEO, Tom Caulfield, bristled at the expression and said, “We prefer to call this ‘co-investment.'” And I guess they could make that argument since they always point out that they’ve invested $15 billion in fab capacity in the U.S., and obviously they’re a trusted supplier to DOD and so forth.

As we’ve reported, the industry’s waging a pretty heavy lobbying campaign right now to shape the outcome of the CHIPS Act. How do you see it?

CHRIS MILLER: You can debate over what the right language is, subsidy or co-investment. But it’s clear that not everyone can get a co-investment. So it’s not a surprise to see the firms lobbying the hardest find more pro-investment dollars than their peers.

I think when you look historically at the semiconductor industry, what you find is that the places where the most lobbying dollars were being spent were places where you often got a lot of legislative activity. That’s not a surprise. But there were often also places where you didn’t have the type of innovation that was making a difference a decade later.

I look back to the debate over Japan’s semiconductor industry in the 1980s, which, if you looked at volume and if you looked at sales dollars in the late ’80s, was over taking the U.S. by some metrics. There was a big offer by the industry then to both pressure the Japanese to open their markets and to even have agreed-upon sales targets that they would hit from U.S. firms, but also to threaten tariffs and to have a variety of subsidy programs for U.S. industry. And I think what’s really striking about that period is that there was a huge focus on memory chips at the time, which were the bulk of the market. And if you fast forward ten years after that, memory chips were an important part of the market, but not where most of the money was being made.

There’s an argument that actually a lot of the legislative activity that occurred as a result of this lobbying in the 1980s really was kind of missing some of the key points, which is that there was a whole other set of technology that didn’t have all the lobbying dollars behind it, but was actually more important for the health of the industry in the long run.

GEORGE LEOPOLD: Chris, you’re writing a book on the history of the computer chip, the semiconductor… Can you tell us a little bit about where that’s at and when the book might come out?

CHRIS MILLER: The book’s about two years away from publication. My goal is to put the computer chip at the center of both the global economy over the past half century or so, and also the center of international politics. Because it seems to me that if you want to understand what has undergirded American power over the past half century, it’s America’s dominance of computing power, which obviously relies on America’s central role in design and producing and selling chips.

What you see with Hauwei right now is actually nothing new, in a sense. The U.S. has long tried to use access to computer chips technology which the U.S. has historically dominated as a tool with which to punish its rivals. We saw this with the Soviet Union, for example, in the Cold War in a very significant way, more significant even than we see with China right now.

And we also see chip technology and semiconductor supply chains as being ways the U.S. binds allies more closely towards it. So we look at South Korea, Taiwan or Japan most interestingly. On the one hand they’ve done exceptionally well over the past couple of decades. On the other hand, they’re extraordinarily reliant on both U.S. technology and the U.S. market. So they’ve been sort of bound into the U.S. system as a result of these supply chains.

My argument is that you can’t understand how either of these processes works. The U.S. competitive strategies versus its rivals or its strategies to bind together its allies without looking at the chip, both as a piece of technology but also as a set of networks that cross borders, a set of trade linkages, as a set of intellectual property linkages. All of which have enhanced American power really substantially.

And the key question, of course, in this race going forward, is, Will this last? Will the U.S. remain at the center of this global network in the next couple of decades? Or it is eroding? I think there are reasons to think that it’s eroding to a certain extent as China invests a lot more, in particular. But also there’s a whole lot of pack dependencies that keep the U.S. in a central position. So it’s not going to disappear overnight, even if the trend lines don’t look that positive today.

GEORGE LEOPOLD: You’ve made the argument that if we do go after Hauwei and TenCent and these companies, it’s going to force some of these companies relying on U.S. technology to diversify and create their own indigenous capabilities, right?

CHRIS MILLER: They’re going to try. The challenge that companies like Hauwei faces that it’s an extraordinarily difficult struggle if you’re going to try to create your own not only semiconductor design but also manufacturing and manufacturing equipment firms. Because right now Hauwei’s cut off from almost all of it. That’s something you can’t create overnight. Hauwei’s done, actually, I think a pretty impressive job on the design front with high silicon, but that faces real severe challenge right now given the new U.S. regulations that severely restrict sales to it.

So I think no matter how many billions China pours into the industry, it’s not something that they’re going to have developed independently within five years or even within ten years. It’s worth noting that the Chinese have poured billions of dollars into their semiconductor industry for now several decades, and there’s been just as many failures as there have been successes. Perhaps even more. So I think the Chinese still remain pretty far behind, even though they’ve obviously narrowed the gap from where they’ve been in the past couple of decades.

GEORGE LEOPOLD: Yeah. At least I think probably about two generations of process technology. If you were able to craft the perfect legislation, I mean obviously you think that more money, investment in manufacturing is generally a good thing. What else would you like to see in this pending legislation to advance our technological capabilities over the next couple of decades?

CHRIS MILLER: I think there are a couple of different questions at play here. One is specific to the defense sector and the defense supply chain. That might be a case where there’s a really strong argument to have manufacturing on shore. The amount of manufacturing you need on shore to meet defense needs is actually pretty small. The defense sector makes up a couple of percentage points of overall consumption of chips in the U.S. So that’s not a huge market, but that’s kind of a separate question than, Do we need to manufacture every iPhone chip on shore?

But I think you need to look at not only the entire supply chain but the entire ecosystem. So start with education. Are we producing enough of the engineers we need to provide the types of talent to chip design firms, to companies that are fabricating chips, etc. Do we have the right set of tax incentives and environmental policies to make it possible to build a fab at a competitive price? It seems like the answer right now in many cases is no. And so we need to look at what the right way to deal with that is. It might be the case that the only way you can make manufacturing in the U.S. competitive is by handing out money, but it’s probably not. And it seems to me that there are probably regulatory questions that can be looked at next to the subsidy questions or the co-investment, if you will, to make fabrication more cost-competitive.

But I think the key point is that it’s not obvious that we need to focus on fabrication. It would be great to have more on shore, but there are a lot of other important parts of the supply chain, and a lot are more valuable than fabrication, too.

GEORGE LEOPOLD: Right. Right. One last question for you, Chris. What do you think the likelihood is that TSMC is going to build that fab in Arizona? And would it matter?

CHRIS MILLER: Would it matter? Obviously it’s not going to be the centerpiece of TSMC’s global production. Obviously it’s going to be not leading edge by the time it’s built. It’s going to relatively small as a share of TSMC’s overall production. We know all that. That said, there is a benefit to having more manufacturing. You just have a larger ecosystem, more people working in the field. That’s a good thing. Is it transformative? Probably not. But is it positive on the margin? I’d say yes. Will it be built? I don’t think I’ve talked to anyone who thinks it’s a coincidence that TSMC offered to build a fab in a swing state in an election year. So I think we’ll have to reconvene after November 3rd to see who’s in the White House and then assess again whether they’re going to go through with that investment.

I’ve been struck by talking with people in Washington how few people think that’s a done deal. Even people who I would think would be obligated based on their job in government to say that it’s a done deal have prevaricated on whether it’s actually going to happen. So it seems like there are a lot of big open questions around it.

To me, I think that just speaks to the broader challenge. We shouldn’t be focusing on a specific project. We should be focusing on what’s the broader system look like; what’s the broader ecosystem look like. Because that’s where we can actually count on government to make intelligent bets over the long term. Is the government going to make the right bet in terms of where TSMC should be build its next fab or what that fab should be focused on? I doubt it. I don’t think that’s what Congress is good at. Their track record is I think actually quite poor on that front. But it you look at DARPA investing in advanced technologies, there’s something where actually they’ve got a great track record. And I’d rather have my marginal dollars go in that direction then go towards subsidies or co-investment, whatever phrase you’d like to use.

GEORGE LEOPOLD: Lots of moving parts. We’ll continue tracking this. Chris Miller from Tufts University, thanks for being with us. We appreciate it.

CHRIS MILLER: Thanks for having me.

BRIAN SANTO: George Leopold has been following the progress of the CHIPS Act in the pages of EE Times. We’ve got links to his coverage on the web page dedicated to this podcast episode, which – if you’re not already here – can be found at www.eetimes.com/podcasts.

By the way, since George is one of our favorite people, and a fine reporter to boot, I am delighted to take this opportunity to flog a biography he wrote called, “Calculated Risk: The Supersonic Life and Times of Gus Grissom.” It’s a fine read, and I think you’ll like it. There’s a link to that on the podcast page, too.