Lawfare Daily: Mike Schmidt Talks CHIPS and U.S. Industrial Policy

[Intro]

Mike Schmidt: The core objective of industrial policy shouldn't be making money, and I think that, you know, sometimes a profit objective will run counter to the national security objective.

Peter Harrell: It's the Lawfare Podcast. I'm contributing editor and host of the Security Economics podcast, Peter Harrell. For today's episode, as part of our new Security Economics series, I sat down with Mike Schmidt, distinguished visitor at Princeton University's Griswold Center for Economic Policy Studies, a non-resident senior fellow at the Institute for Progress, and for today's purposes, the former director of the CHIPS Program Office at the Department of Commerce. Basically, the person who turned the 2022 CHIPS Act into a program that actually secured hundreds of billions of dollars of planned semiconductor factories for the United States.

Mike Schmidt: China is manufacturing as a total powerhouse. If our strategy is to focus only on what we're building here, it's gonna be very, very difficult for us to compete.

Peter Harrell: The Security Economics podcast explores the intersection of economics and national security, how countries manage their economies to support national security, and how they deploy economic tools in support of geopolitical objectives. Today, Mike and I discuss how the CHIPS program worked in practice, what kinds of industrial policy tools a government needs, and how we should be thinking about industrial policy as a nation.

[Main Podcast]

I wanna begin with a question just to jump in about where the semiconductor industry is today. When you were setting up the CHIPS office back in 2022, 2023, a lot of the industry was in the doldrums and people were worried, for example, about the future of Intel. Memory semiconductors were going through a cyclical bust and there was, I think, a very strong impression both in Congress and in industry that, oh if we didn't have the CHIPS program, there's no way these companies would be able to afford to build here in the United States.

Today of course we're in a very different situation . We've seen Intel stock which of course the U.S. is now a 9% owner of, I think more than triple over the last couple of years. The memory companies have all joined the trillion dollar club. They're all making money hand over fist. When you look back from where we are today to where we were a couple of years ago, do you think we actually needed the CHIPS Act to build semiconductors in the U.S.? Or, could these companies all actually be paying for it out of pocket?

Mike Schmidt: Yeah, you know when I think back to the dynamics in the industry when we started, I think the other characteristic that I would flag and reflect on is the fact that when CHIPS passed, mobile, particularly on the leading edge, mobile was still the huge driver. They ruled the roost in terms of demand with Apple being the big gorilla in the room, and of course companies like Nvidia and AMD were big drivers of demand, but the whole shift towards AI being the driving force of the industry hadn't happened yet. CHIPS passed in, in August of 2022 and it was that fall that ChatGPT had its viral moment. So while we were implementing the CHIPS Act, the industry was rapidly shifting and adapting to what was gonna be, you know n- now clearly is years and years of a structural change with AI really, really driving the demand.

I think that there's no question that there would be huge demand for chips and chip production driven by AI and that these companies like TSMC and Micron and Samsung and Intel and even others would, would benefit from that. The question for CHIPS was, you know, where's that gonna happen? Mean- meaning where, where is the manufacturing gonna happen?

And, you know, I think that there were some kind of incentives in the system towards geographic diversification so you could imagine some measure of onshoring or supply chain diversification. But I don't think we get anywhere close to where we are without the incentives of the CHIPS Act and wi- without Chips implementation because now the United States really is a premier dis- destination for semiconductor manufacturing particularly on the leading edge. We're the only economy in the world that has all five leading-edge producers building projects here, really large projects. There's no other economy with, with more than two and so my, my assessment is that, that the shape of, of global supply has, has really been impacted by the CHIPS Act.

Peter Harrell: So maybe in a sense actually what you're saying is that we, the United States captured a larger share of this build out, this AI driven build out that we are seeing, because of the act that got the, the companies thinking about the United States as opposed to South Korea or Singapore wherever.

Mike Schmidt: That's right. That's right. A- and, and then the other thing I would say is that when I left the CHIPS office in, in early 2025 we felt like we had put the industry or put the country in a, a really strong position with Micron and TSMC, and Intel, and Samsung, SK Hynix. But there were a, a lot of risks kind of still embedded in many of those projects, and strong demand has a, a way of making risk disappear or at least be mitigated.

And so I think when you look at the strength of Intel, or you look at the e- the strength of Samsung and their foundry business, when you look at just the overwhelming demand for DRAM that's driving micron chip production and therefore how important their project in New York and their project in I- Idaho are gonna be, I think that, that there's no question that kind of once we've, once we s- the CHIPS Act kind of set the industry up in the United States as we did, that the demand has been really helpful in kind of solidifying that trajectory.

Peter Harrell: That makes a ton of sense. Let me step back a little bit with you, Mike, because I think that for many of my listeners, when they think about the chips program, they are thinking about the grants program, which was kind of the signature piece of it, or certainly the piece that, that generated most of the attention. But obviously, the CHIPS Act had a couple of different pieces to it. There were grants, tax credits, R&D. Maybe if you could just step back and walk through how you thought about the structure of the chips program and how the different pieces of the, the chips program toolkit fit together.

Mike Schmidt: So I think the tax credit was foundational, and I think to your point gets, gets a little bit less attention. But the tax credit was a 25% ITC, and it was direct pay basically fully refundable. So regardless of tax liability, that's 25%, that's as good as cash if you make, make that investment.

So we always thought of that as the, the baseline incentive. And then our job at Commerce Department with we had, we had grants and we had loans. We can get to the loans, but really we were mostly focused on the grants. Our job was to say, or the way we conceived of our job was to say, “how can we use these grants to get companies over the edge in terms of making investments in the United States?” And we wanna give them just as enough subsidy with the grants and the tax credit combined so that it makes sense for them to invest here as opposed to in Taiwan or Korea or Singapore or Japan or, or, or elsewhere.

So that was the, and, and we built a whole operation, a whole process, a whole team that we felt was positioned to engage with companies in that way to do the bottom-up analysis of project economics and financials and determine what the subsidy level was that was necessary to secure the investment in the United States.

We also had loan authority, which we tried to deploy as a, a kind of another tool in the, in the toolkit. It ended up being substantially less powerful than the grants for obvious reasons. You have to pay loans back. And also fundamentally the, you know, the CHIPS Act, the, the problem it was trying to solve was that it's more expensive to build chips here than in the United States. These aren't companies that really need access to financing. So we were able on the margin to do some loan negotiat- negotiations that, that were, were helpful but that wasn't, you know, that wasn't really the, the major driver of, of the, of the program.

Peter Harrell: Can I actually ask you a question on loans real quick, Mike, just to jump in because it's, it's interesting when I look at other recent pieces of U.S. government industrial policy we did see more of an emphasis on loans. So for example if I look at the work that the Department of Energy Loan Program Office did, obviously loans, with, with under the previous administration But there was a, you know, really a lot of lending into clean energy and clean energy materials projects. Or if I look at the Trump administration's kinda current focus on critical minerals and various other kinds of Trump administration industrial policy, there's equity which we can come to in a minute and I'd be interested in your views, but you really see a lotta loans to a buncha these mining companies, and I'm, I'm interested in like, when you think the government should be thinking about a loan versus a grant how the different tools fit together.

Mike Schmidt: I think you need to start by asking the question: What is the problem here that the market itself is not solving? And so, in the case of chip manufacturing the problem was never that TSMC and Samsung and Micron couldn't issue bonds, couldn't access capital. These are the some of the most largest, most creditworthy companies in the world. The problem is that they're shareholder-driven enterprises that wanna make investments that are most profitable and-

Peter Harrell: Logically enough.

Mike Schmidt: Logically enough, and it is more expensive to do it here than to do it elsewhere. And so incentives have the function, direct incentives grants have the and tax credits have the function of offsetting that delta to making it economic for them to do it, do it here. There are a lot of businesses where that's not the primary problem, where actually accessing capital or accessing capital on terms that are gonna, you know, allow the a, a project to bear out commercially is much more difficult. And there I think that the toolkit, the credit toolkit, could be very valuable as well.

Peter Harrell: I do wanna ask about equity authority. Obviously one of the things we've seen the Trump administration do both with the CHIPS Act and also with other authorities is, you know, convert what I think previous administrations used more as grant authorities into authorities that they view as allowing them to take equity stakes, provide capital via, via taking an equity stake. And I'm, I'm interested in your views on, and I know you've written about this too, but I'm interested in your views on, you know, w- when it would be useful or helpful to have an equity toolkit for industrial policy, and when you think the government should generally be relying on these other, other kinds of assistance.

Mike Schmidt: Well, I'm gonna answer that in a couple ways. One is that I think as part of the toolkit, I'm pretty supportive of having it, having access to it, which is separate from saying when, when should you deploy it?

Peter Harrell: Do you think it should be in the toolkit? Makes sense to have it in the case.

Mike Schmidt: Yeah, I, my personal view is the toolkit should be very broad, in that we should have figuring out what the right intervention is for the right company, for a particular company or for, for a particular industry is gonna change from industry to industry, company to company, and a broad toolkit is valuable in that we need to be empowering the state with a measure of discretion to advance our industrial policy objectives in, in different ways.

That is a separate question from the kind of, a set of kind of disciplines or norms or a broader state craft around deploying those tools and when it makes sense to deploy certain tools in, in, in certain contexts. Equity, I think it goes back to, to the question I asked earlier: What is the problem that the markets aren't solving, right? So if you showed me a early stage company with a technology that was gonna be critical to industrial base in some way or critical to national security in some way, and they sh- were struggling to raise capital from traditional venture but it was a really important technology, I would say, "You know, there's a case where government equity could make a lot of sense." In the case of more mature companies that ha- that can raise equity on their own, I struggle to see it in the same way and, you know, equity introduces a whole range of complications, policy downsides, et cetera, that then have to be considered.

I mean, big picture, I think, you know, there can be a bit of a sense sometimes of like let's take equity because it's costless or because the, the, you know, the, the government's gonna make a profit.

Peter Harrell: I mean, Trump keeps talking about how much money we've made on the Intel investment.

Mike Schmidt: Yeah, and, and to me it's like, I, I mean, a couple things on that. One is like I just don't, you know, I think we have other ways of making money as a government, like that I don't view industrial policy as, as being the, you know, the core objective of industrial policy shouldn't be making money, and I think that, you know, sometimes a profit objective will run counter to the national security objective. And if it's something that is profitable on its own, then, you know, probably the market could do it anyways and, and so what are we, what are we kind of doing in there?

And then, and then all industrial inventions will come with costs, and some of those costs are fiscal costs, you know, like tax credits and grants. They, they cost taxpayer money. But some of the costs might be measured in policy drawbacks or downsides. And I think that in the case of equity, that’s, you know, you're, you're bearing some of those costs as well.

Peter Harrell: Let me a- ask you about a different part of the toolkit, which is the the demand side part of the toolkit. And you, you have talked in the past and written in the past about how there wasn't that much of a demand-side part of the toolkit for CHIPS. My sense is you and Secretary Gina Raimondo could, you know, call up companies that are buying chips, you know, computer companies went and say, "Could you please buy American chips?" But you, you didn't really have a good tool to create a demand-side incentive for the fabs that were are being built here in the US.  I, I'm interested in kind of looking back what your views are about where demand-side tools should fit and how you would fit, you know, would, would have fit demand-side tools into the CHIPS Act i-if you would have.

Mike Schmidt: So CHIPS, CHIPS was a supply-side program. It was grants and tax credits to make investment happen here. I think supply-side incentives are hugely powerful, and they are by their nature limited depending on the set of problems you're trying to solve. So, what you describe is, is exactly right. We spent obviously most of our time negotiating with the major semiconductor manufacturers like TSMC and Samsung and Intel and Micron to try to secure their investment here and negotiate the in- or incentive deals with them.

But we also spent a lot of time with their customers. And this is Apple, NVIDIA, AMD, Qualcomm, et cetera. And we did that for two reasons. One was it helped us understand a full picture of the industry so that we were more knowledgeable counterparties in negotiating with the semiconductor manufacturers.

But, two, was because we really wanted to encourage them to provide strong demand si- signals for American-made chips. And those, you know, that could be, you know, we want you to demonstrate a willingness to pay a premium for American-made chips. That could be we really want you to begin to diversify your supply. Obviously, TSMC was and continues to be an incredible dominant player in leading-edge manufacturing. It was really important to us to try to also drive demand to Intel and Samsung because we wanted a diverse competitive ecosystem for leading-edge chips, not just for the United States but, but for the world. But, but our tools for doing so were, were kind of inherently limited.

Peter Harrell: Let me ask, let me ask you a quick question on that. So you wanted a diverse, you know, multiple suppliers. Was that for supply chain resilience purposes? Now k- kinda market competition purposes? Like why, what did you see as the benefit to having multiple leading edge suppliers rather than kind of a “we'll bet heavy on TSMC and make sure it succeeds” kind of approach?

Mike Schmidt: A couple things. One is I think resilience and competition both of those things. A world in which one company is producing every advanced logic chip is, you know, a, a world that has a lot of drawbacks-

Peter Harrell: Yeah.

Mike Schmidt: From a, from a competition perspective, from a resilience perspective, security perspective etc. You know, the other thing I would say is each of the companies in the leading edge ecosystem has the ability to bring different types of advantages to the country or a willingness to bring d-different types of advantages to the country.

So with TSMC for example, we were very successful in getting them to increase their ambitions to do leading-edge manufacturing here and that was a huge success. By January 2024 they were doing leading edge they were actually producing leading-edge chips in the United States. They had committed to three fabs. They then updated that. They committed to six fabs. So huge massive success in terms of what they're willing to build out.

We also spent a, a lot of time talking to them about where their core process technology R&D was happening. That is still happening in, in Taiwan. With Intel, they have an incredible R&D ecosystem in Oregon and with Samsung as part of our deal they actually committed to do some R&D in the United States for the first time out, out of Korea.

So, when you think about a resilient and competitive ecosystem having multiple players is advantageous but also each of those players can bring different things and if you're able to negotiate with different companies to bring different things like that, that's, that's an advantage too.

And then the last thing I would say is we kind of came to the view that there's value in a diversified portfolio, and you don't know where risk is gonna come from. And having, you know, seeding several investments when the future's uncertain was the prudent way to do it.

Peter Harrell: One last toolkit conversation or part of the discussion before we move on to other topics. You have said in the past, you've described in, in the past a moment when you realized that one of the chip companies you were negotiating with had a matter pending with the Bureau of Industry and Security at the Commerce Department which presumably was something to do with export controls. And you said that linking the two would've made getting what you wanted on the CHIPS Act easier but that, you know, the lawyers thought that was a bad idea and generally dissuaded you from that.

I have thought in the past, on a kind of similar track that, you know, you can envision places where export controls could be part of industrial policy in the sense of if we have some technology that we've invented here but we don't want it manufactured in China, for example OLED screen technology, you know, which is now mostly manufactured in China. We could've restricted the export of the technology to China and kept the manufacturing somewhere else. I'm not saying that would've been a good idea. I'm saying that conceptually you can think about how export controls would be a piece of the, the national, the industrial policy toolkit. Curious how you saw that like obviously you can't give details about specific cases but kinda-

Mike Schmidt: Yeah

Peter Harrell: How you saw the pros and cons of mixing export controls an- and industrial policy?

Mike Schmidt: It's a, it's a great question. I think there are few, a few threads to pull on there. One is I think on the export control side, when we were in government, I wasn't involved in, in export controls. But I think the Biden administration was pretty careful and consistent in saying that export controls aren't a tool of industrial policy. They're a tool of, you know, kind of other national security concerns.

And I think that deserves scrutiny and, you know, should be revisited because in the fact pattern you just provided or others, I could imagine wanting to have an integrated strategy. And even if, you know, I, I will say, even if the intent of the Biden era export controls was not related to industrial policy, the effect was. You know, like, like, like arguably the most important structural shaper of the global semiconductor industry right now is the fact that China does not have access to EUV technology.

Peter Harrell: Right. You'd have leading-edge chips being made in China today if they had access to EUV technology.

Mike Schmidt: Right.

Peter Harrell: Yeah.

Mike Schmidt: And, and so what, what does, you know, Samsung and Intel, what do their prospects look like if Huawei is churning out, you know, really good five-nanometer chips using EUV machines from ASML? Or if, you know, TSMC and Intel are competing for ASML's EUV machines with Huawei or whatnot, right? I, like, I, I, I think that, I think that's, that's hugely important.

I think the anecdote you're getting at all, or you, you mentioned gets at another question too, which is whether it's through export controls or through other authorities of the government, how comfortable are we using the power of the state coercively to get what we want when it comes to industrial policy? And there I think that certainly when we were in, we were pretty careful. I think the current administration is much more comfortable u- using more coercive forms of, of intervention.

Peter Harrell: Yeah, definitely like, the way I look at them is like they have objectives they want to achieve that are sort of substantive objectives, and they'll kind of pull whatever leverage they can figure out to pull to achieve those objectives. And you know, that may have long term downsides. I, I personally think there are many ways in which it might but, you know, they're less worried about that. They're kinda how do we, what leverage we have to achieve this objective over the next 12, 18, 24 months.

Mike Schmidt: Exactly. And, and on, on the one hand I have sympathy in the sense of like, you know these are big powerful companies. It's not like I feel bad for them. You know?

Peter Harrell: Yeah, that's fair. That's fair.

Mike Schmidt: On the other hand, I do think that there are some like basic norms about how our system works over the long term where, you know, that, that are worth protecting. And there could be trade-offs between in the short term using coercive power to get what you want in the long term the type of institutions and investment environment and country we have, so.

Peter Harrell: Let me step back a little bit moving beyond just the CHIPS Act to some of the lessons and reflections you might have coming out of the CHIPS Act and some of the research that you are doing now. You've talked in the past about how the CHIPS Act is a non-repeatable one-off, and that now that you're at Princeton and at the IFB thinking about how the government can get industrial policy right, right long term, I'm interested in how you think the government should structure itself to do industrial policy across, you know what I think we would all agree is gonna be a number of important strategic sectors that the U.S. government will want to invest in.

Mike Schmidt: So just to clarify, I don't know that I would say that my view is CHIPS was a non-repeatable one-off, but what I would say is that it was very hard to do. There are a bunch of hurdles to jump through to get through anyth- anything done in government, and it, it took a pretty extraordinary team working really, really hard to make it happen. And so I think from a policy standpoint, we need to think about state capacity and kind of reducing the constraints on the government as part of this, the, the question of what we, we need to make industrial policy succeed.

I think that relates to the, the broader question of what is the right institutional design for industrial policy going forward? One place I start with here is the role of Congress. And the reason I say that is that there are a lot of proposals out there for kind of flexible pools of strategic capital, and my personal experience running the CHIPS operation was there's just this huge amount of value in having Congress set the objectives.

Peter Harrell: Rather than sort of a flexible-

Mike Schmidt: Rather than a flexible-

Peter Harrell: You know, “Here's $30 billion, you figure out what's important.”

Mike Schmidt: It's like the bill passes and you're like, we know that if we succeed in bringing leading edge production here, we will have kept faith with the purposes of this legislation. We will have kept faith with the bipartisan coalition that, that enacted it. And from day one, you are mobilizing around that objective and you are attracting talent that is geared towards that objective, right? People who, who have experience, relevant experience, who wanna be a part of the, the effort. And notwithstanding a bit of a bumpy road after the transition, like there has been actually a fair amount of bipartisan continuity in terms of advancing the objectives of the CHIPS Act which I think would be much harder to maintain if you had a flexible pool of capital and people had different ideas of, of what to, what to do with it.

So, so I think that I, I, my optimal institutional design personally I think starts with Congress saying Here are the five or 10 industries that we care about, and here are the resources that we're going to put against those objectives. You know, high level, not getting into the weeds of you're supposed to do this with the money or that with the money. A lot of discretion when it comes to process, the right mix of tools, you know baseline non-discretionary incentives discretionary incentives but fundamentally Congress is kind of creating the broad framework and then the executive is mobilizing to implement it.

Then the question becomes how should the executive structure itself in that effort? I think that institution building is really important. I think we need to rationalize the various offices and, that are doing this. I could imagine a big bureau maybe in the Commerce Department that kinda does all of it. A big “Bureau of Industrial Investment or “Economic Security” or something like that. I could imagine saying “we're gonna have one at Commerce, one at Energy 'cause energy is its own kind of unique beast and they have the labs and that's, you know, all that. And one at DOD.” Some kind of rationalization in terms of how it all works together and then a set of, you know, special authorities around special hiring authorities contracting lots of admin funding, you know what it takes to really build an institution to do this.

Peter Harrell: It is interesting to me to hear you say you think it makes sense for Congress to set the direction because as you say there have been a number of proposals over the last couple of years that would give a more flexible tool to the executive branch so I can kinda see, intuitively or sort on first glance, I do see the appeal of that like well will, Congress doesn't necessarily know what we're going to need and all the rest but it's, it's interesting to hear you talk about your right wouldn't, if, if you have a flexible approach you will have implementation that might chase too many objectives you might have political support that doesn't endure across administrations. It's just interesting to me to hear, hear you actually liking that con- strong congressional mandate

Mike Schmidt: You know I am not like one of these people who's like oh we need to do what China does when it comes to industrial policy I think we need an American model of industrial policy and I think we're kind of developing that but one thing I think they do really well is set long term objectives and stick to them right? And in the American system the way we do that, the way we create actual policy stability is through bipartisan legislation in Congress. And so I think that that takes some work maybe politically it's too hard don't know But from an optimal institutional design standpoint, that's how I would think about it

Peter Harrell: Well uh that maybe brings me to the next topic I wanted to talk with you about which I know you've been thinking about at Princeton and beginning to do some writing on, which is well, maybe I'd phrase it this way, if you were going to go up into a Hill briefing for Congress on what sectors they should pass legislation to drive U.S. industrial policy towards, I'm interested both in the sectors but also how you think about what those sectors should be.

I'll be very frank here, you know, I worked when I was with the Biden campaign and then very early in the Biden administration, I worked on the president's supply chain EO, 14017, where they laid out, you know, "Here are some sectors where we're gonna work on supply chain." And you know, we had chips and we had kind of EV batteries, we had pharmaceuticals, we had critical minerals. And, and being frank like the way we came up with those was kinda, well it's sort of obvious that these are four we need to work on. There wasn't a ton of scientific process you might say. You've obviously now been thinking about this quite a bit and lived through CHIPS like how should we be thinking in a more structured way about, you know, about what sectors is appropriate to do industrial policy on?

Mike Schmidt: Yeah th- there is a bit of kind of I know it when I see it in the-

Peter Harrell: Yeah, exactly. I mean that was, you know, that was for EO 14070, that was, you know, it was kinda well there's bipartisan support for this. We kinda know these are important.

Mike Schmidt: Yeah. Yeah.

Peter Harrell: You know there'd be political support let's do it.

Mike Schmidt: Which, which by the way like there's going to be art and science here right at some point its gonna to be your applied judgements but I think we need more stronger theoretical frameworks. I would say, so first question is deciding that an industry is critical ,that matters in some deep sense for economic and national security.

And there I think their few categories that jump off the page to me. One is obviously defense technologies. Two is dual use technology. So technologies that are really important for defense, but where you can't really sustain them commercially unless you're, you also have a thriving commercial market supporting that, that production. And then three is broad-based intermediate goods, right? Br- broad-based intermediate inputs into the system. So, you know, think rare earths being a great example of this. Obviously, energy is an example of this, where both for the national security applications, but the broader economic implications of, of disruption or coercive efforts are such that having vulnerability in those industries kind of threatens our political autonomy, right? Our ability to sustain certain policy trajectories. And then the last I would add is areas that are critical to health and human life, right? And so when you think about pharma and medicines and all that, like it might not be that it's a huge supply chain disruption, but it's a really, really big deal if we can't get access to those things because people's lives are at stake.

So that's kind of like this, this bucket of questions around the criticality of the industry. Then I think you need some theory around international dependence, with dependence on China being the key question. But obviously, in the case of chips, Taiwan is part of that. For certain industries, you might have, you know, Russia or other countries. But I think that that needs to be a, a crucial part of the analysis.

I think capital intensity is very important. I think the case for an intervention goes up when things are really expensive to build and when it takes a really long time to build them, right? So, if you think about masks during the pandemic, you know, you might say, "Well, it was really bad we didn't have masks." But you might also say, well it took like a couple of months and then suddenly we did have masks, right? And you know, and for next time, you know, maybe just having a bigger stockpile of masks is what we need so that we can bridge to, you know, when we're able to ramp production et cetera. So I think capital intensity is really important.

And then I think you need some framework around not just kind of addressing the world it as, as it is but the world where it's going, right? And so, you know, robotics might be a good example of this today, right? We don't have major, r- robotics industry will evolve dramatically in the next 10 years, and if we don't act, might we, you know, accumulate a set of dependencies that we, we don't want, and so we should act now, that kind of thing.

So, I would say those are, those are some, some categories I think about all within, you know, some theory of the case of allies and partners and, you know, not, not trying to do everything here, right? It has to be part of a broader, broader effort.

Peter Harrell: And you know, it's interesting you mentioned masks. I've been doing some, for some research I've been doing on industrial policy, I've been looking at the U.S. government's experience trying to build up mask capacity in 2020 and 2021, and I think you are in, y- you are onto something with the lesson there. We basically were able to surge capacity, but also what ended up happening in 2021 is demand plummeted. A bunch of the new entrants we surged capacity to went bankrupt because you know, the demand for the U.S.-made product basically evaporated. And I think there is maybe a lesson that for that kind of a product thinking about better stockpiling with some sort of flexible surge capacity is gonna be more economically efficient and more policy efficient than trying to have some long term mask industry here in the U.S.

Mike Schmidt: Right. If, yeah, it's like regardless, the i- the right, the right answer isn't ex ante to say we need-

Peter Harrell: Yeah.

Mike Schmidt: To invest billions of dollars in max, mass factories because it's a problem we're probably gonna be able to solve when the contingency arises, right? So.

Peter Harrell: On that topic of kinda resilience and import dependencies, you've written about this concept you're developing of vertical resilience versus horizontal resilience. And I'm, I'm interested in how you, how you think about that.

Mike Schmidt: The concept of v- vertical resilience is saying, okay, within a supply chain, within a chip supply chain, semiconductor supply chain, you have equipment that goes into chip manufacturing, like EUV machines. You have the chip manufacturing itself, and then you have the packaging of those chips. All of those are necessary to get a semiconductor to market. And the intuition there is that you can have a goal, we're gonna onshore chip manufacturing, but if there continue to be kind of durable dependencies in other parts of that vertical supply chain, you're probably not having the impact you want to have in terms of overall resilience, right?

And so, to make it concrete, in CHIPS implementation we were really, really focused on chip manufacturing which was, you know, for the advanced logic ships was highly concentrated in Taiwan. But we also thought, man, if all these chips just end up getting packaged in Taiwan anyways, we're probably not filling-

Peter Harrell: They're just being flown from the U.S. to Taiwan for packaging.

Mike Schmidt: Yeah. We're probably not going to be having the impact we want to have. So we pounded the table. Like, we went hard to try to get more packaging in the United States. And we went direct to market to companies that do packaging, and we also applied a lot of pressure on the chip makers themselves saying like, "Hey, if you want federal funds, like one, one thing we need to understand is where these chips are going to be packaged," right?

And so in some instances those, those efforts led to new projects being announced in the United States. Like we had a company called Amkor announce a major advanced packaging project in Arizona, not far from where the TSMC, where the TSMC plant is, is being built. But in some cases, we also had companies commit to changing where their packaging is happening abroad. So saying, "We're gonna, we're gonna build, you know, as part of this deal, we commit that within five years we will have another packaging facility outside of China and Taiwan, you know, in a different, more resilient place."

So that's, that's the concept of, of, of vertical resilience. And there I think the, the big lesson or intuition is, you know, policymakers really should be looking at the full vertical supply chain to say, "Where are there gonna be sticky areas of dependence?" Right? If they're areas of dependence that maybe can be addressed in a few months, you know, like the mask equivalent, maybe you don't need to, a solution to everything. But if they're something like an advanced patching facility, which takes a couple of years to build and is expensive, et cetera, then yeah, I think, I think you, you want to make sure you have a comprehensive policy framework.

Okay, so horizontal resilience is, is saying, okay, like let's say there are five or 10 major industrial dependencies on China, right? Do we have to address all of those in order to shift the strategic dynamic in our favor. And I think the intuition there is, obviously it's great to address all of them, but also there can be a lot of value in addressing the most acute dependencies and that the returns to those investments from a strategic standpoint can be substantial even if a set of other dependencies might remain.

Peter Harrell: Hmm. Let me ask you quickly actually sort of follow on on that. For the dependencies we have whether they're vertical or horizontal, how do you think about what we need to onshore verse friendshore? You know, I think for example, you've said, you know, “We weren't gonna manage to onshore EUV machines,” which are mostly made in the Netherlands or I guess entirely made in the Netherlands by ASML. Like, how did you think about what parts of this n- we need here in the U.S. versus this friendshoring rubric that that the previous administration talked about more than the current one?

Mike Schmidt: Well, so big picture I think that China’s manufacturing is a total powerhouse and that if our strategy is to focus only on what we're building here, it's going to be very, very difficult for us to compete. But if our strategy is to work in concert with allies and partners then I think we're in a very, very strong position. So I think, big picture I think that's essential. It was, I think Lee Kuan Yew said that, you know, China has a billion people, America as it, as its best has six billion or seven billion people something like that. And I think that's, I think that's kind of like how I would, you know, a hopeful optimistic take on American policy and where it can go maybe. But I think that's, that, that should be our, our mindset.

Let me just reflect on kind of how this played out in, in, in CHIPS because there were some big strategic questions about which parts of the industry we were gonna attract here and which we were gonna really kind of focus on and, and prioritize.

So in the case of EUV machines and, you know, ASML w-we were kind of looking at s- limited resources. We had 39 billion dollars which is a lot of money but actually not a ton of money when it comes to the scale of the industry. And it was really about how are we gonna most efficiently allocate our capital and we had discussions that said, "Should we make a real run at ASML?" You know, we decided that we were comfortable having production i-in an allied country like the Netherlands and that we'd focus our, our capital elsewhere and focused on other, other dependencies.

Within chip making itself it was interesting because what you ended up seeing happen was that, kind of, within the context of governments intervening and competing for investment the market also was making a set of kind of pretty sensible decisions about where manufacturing would happen. What types of manufacturing would happen, right? So we, we had great ambitions, or discussions about ambitions I should say to onshore not only leading-edge manufacturing but current and mature technologies. And we had some real success there. We had GlobalFoundries, we had Texas Instruments.

But if you look, for example at what TSMC decided to do, TSMC said, “We're gonna do leading-edge manufacturing in the United States and we're going to build our current mature facilities in Germany and in Japan.” And if you look at each of those countries it makes a lot of sense. The United States is home to Apple and NVIDIA and AMD, the major leading edge customers. TSMC says, “Yeah, okay, I'll build a facility close to my ma-major leading edge customers.” If you look at Japan and Europe obviously they're really important industrial ecosystems there and building lagging edge technologies c- in partnership with those customers makes a lot of sense, so I think that the, at least the experience that we had was that in the context of these single nation efforts to onshore particular parts of you know, of an industry that certain kind of logic of the market still played out in terms of where investment around the world ended up happening.

Peter Harrell: Let me ask you one more CHIPS question. So you wrote an, or I read an essay of yours recently where you talked some about a couple of lessons from World War II, or comparisons to World War II, and I think one thing that really comes through in that essay is we managed to build the factories very quickly in the late 1930s early 1940s and, you know, there's a lotta talk today about the way in which environmental regulations various other, various bureaucratic structures slow us down. What were the barriers o- or the challenges to moving quickly that most surprised you? Like, if you'd asked me, I would have said, yeah, of course they're gonna be environmental issues, so I'm not surprised at all. I think you've said you spent a quarter of your time on NEPA which is maybe higher than I would have expected or like-

Mike Schmidt: That was, to, to be clear that was, that was not for the whole program. That was like a-

Peter Harrell: Got it. Okay

Mike Schmidt: Okay. Yeah but no that was-

Peter Harrell: But what slowed us down? Like what were the things that surprised you that slowed us down?

Mike Schmidt: I think what's really important to understand about this is that it's about the cumulative effect of a bunch of constraints and any single constraint you might look at and say like, “You know what? That makes a lot of sense. I understand why that's there,” and most of them do, not all of them but much of them do.

Peter Harrell: They're not totally random basically. There are, there's a reason for each of 'em. Yeah.

Mike Schmidt: When you add them all up you're sitting there thinking like man this is just too hard and you're doing it in the context of I, I think one of the things that made it most kind of acute for me in my experience was we're literally sitting across the table from private industry, right? So it's like you have this, the kind of counterfactual is very visible and you're trying to be dynamic and keep up with and compe- compete with, with industry in a sense, right? 'Cause you're negotiating, right?

So, I would say just to, just to like walk through some of them. The first one I felt, like in my first few days of work, is procurement, like government contracting. So like, you know, you come in, we had special hiring authorities which were absolutely essential. Without special hiring authority we would not have been successful in implementing the CHIPS Act. But we didn't have any exemptions from procurement rules and so you start, you start work, the first thing you wanna do is build a team. It takes a long time to build a team. We needed quickly some semiconductor expertise in-house. You have to go to a consultant to do that. But if you have to go through a normal government procurement process it's very, very difficult to do that.

I mean, I, I remember a story where we were sitting across the table from one of the big executives at one of the companies we were negotiating with who kind of gave a whole speech about how our incentives weren't sufficient and, you know, he had a fiduciary obligation to potentially invest abroad and all this kind of stuff. And he presented his, you know, independent analysis from a major investment bank explaining why our incentives were insufficient and we also had plans to have an investment bank who would be able to produce similar such analyses but ours was still in procurement, right? So it was like this very-

Peter Harrell: Yeah.

Mike Schmidt: It was like this very kind of like clear distinction. But anyways, procurement, I mean the Paperwork Reduction Act which basically says you need to like go through a process with OMB and a public posting with comments every time you want to solicit information from the public, you know, including major semiconductor companies. Hundreds of hours working on the Paperwork Reduction Act.

Peter Harrell: So it's not reducing paperwork is what I hear you saying.

Mike Schmidt: It's, it's, it's just producing paperwork.

Peter Harrell: Yeah.

Mike Schmidt: We had, you know, in the, in the nitty gritty implementation of something called the Davis-Bacon Act which is a statute that requires s-something called prevailing wage for construction workers, we ran into some, some meaningful challenges. NEPA was very difficult. Of course we ended up getting a correc- congressional exemption from NEPA but you know, it's not, it, you know, when I think about environmental, NEPA is part of it th- that, that requires an environmental analysis or assessment of, of, of your projects but also the underlying substantive permits; Clean Air Act, Clean Water Act Endangered Species Act. You know each of those depending on the nature of project can introduce hurdles that can be hard to, hard to deal with and sometimes they feel like they're advancing meaningful, kind of, substantive goals, sometimes its kinda hard to see ,kind of how they are laddering up into like a sensi- sensible overall permitting policy.

And then, and then we had I would say among the most surprising to me which I really didn’t anticipate maybe I should’ve We had this period where we had to go from our term sheets, we announced a bunch of term sheets to great fanfare, and then we had to go from our term sheets to our final awards. And you know I kind of felt personally felt at the time I was like oh man were at like the 10 yard line. We're almost there right? And then getting into the weeds of finalizing these awards and-and the legal relationship between the government and the companies proved to be very, very challenging in terms of things like what are the circumstances where we're going to be able to stop funds or claw back funds? What are the termination rights of each party? What is the dispute resolution mechanism? What are the covenants; the affirmative covenants, the negative covenants?

Peter Harrell: All sort of, these are big contracts since you're gonna have lots off-

Mike Schmidt: It's a big contract reps, of like just contract detail, yeah. Reps and warranties and events of defaults and indemnities you know, like if the government gets sued for something that happens at the project site is the company indemnifying the government

Peter Harrell: Right.

Mike Schmidt: Those are these big open ended liabilities, so companies and their boards of directors really worry about them. And then, so we’re trying to negotiate with it but turns out there’s DOJ guidance on how indemnities can be used. Our lawyers are actually negotiating with DOJ while were negotiating with other companies. You know, it was-

Peter Harrell: Yeah.

Mike Schmidt: There were, so like I took a trip to TSMC in Taiwan and we just spent a full week in conference room going through these details, and I think that there was, big picture on all constraints, they're there for a reason. Most of them are there to protect taxpayers in some ways, but they come at the cost of adding friction and we need to, you know, the, like…

Like I remember, you know, I asked the, a counterpart at TSMC, I said, “You know, how does our contract compare with Japan's contract?” And he looked at me kind of puzzled. He said, “W- what do you mean?” I said, “You know, Japan, your deal with Japan, tell me, give it to me straight. What's it like?” He goes, “Oh, we don't have a contract with Japan.”

Peter Harrell: Wow.

Mike Schmidt: You know? He said, “We just, we just, we have our agreement-

Peter Harrell: Building in the incentives and then off we go-

Mike Schmidt: Yeah, "We submit our, we submit our invoices and we get the incentives and, and then…” You know, there are other challenges-

Peter Harrell: Wow.

Mike Schmidt: With that. There's a lot of ex post auditing and, you know, so it's not that AOB. But, but it is, it is there was a lot of friction there.

Peter Harrell: If I remember correctly, you went on Odd Lots in 2023, and said, you've said that coming out of that interview where you talked about the CHIPS Act you managed to hire a couple of folks who had kind of decided to apply after, after hearing you. And I, I'm curious what advice you might offer people who are out there right now who might be interested in working for the government on industrial policy. Like you, advice on like offices they should be thinking of applying to or skills they should be thinking of building or just, you know, any advice you might have for listeners who are maybe in college or grad school or think tanks right now, but, you know, want to do this on the inside and what advice would you give them?

Mike Schmidt: The first, I guess first advice would be to encourage it. I think it's hugely important for the country. I think it's, will continue to be a really important part of U.S. policy, and it's awesome and fascinating, awesome, fascinating work.

I think that there are a few angles into it. One is to develop really deep industry specific expertise, right? So we had people on our chips team who, we built a team of one hundred and eighty people. Among that group were people who knew the industry really well and they were of extraordinary value. Some of them came from the industry themselves, some of them came from within the U.S. government, the intelligence community, DOD, et cetera. So there are a lot of different places there you can accumulate that kind of knowledge, but, you know, that was really valuable.

But then there are a set of, there are a set of skills that were valuable, you know, even if you didn't have the industry specific context. So like core commercial skills, you know, the ability to underwrite negotiations, and do financial models, and understand businesses, and the types of things you, you, you'll accumulate in the financial sector or in consulting, et cetera. Those were hugely valuable.  But we also had an awesome workforce team, we had a awesome environmental permitting team, we had a, an independent risk team. So, I think there are different kind of entry points, but you know, a set of skills that, that are, that are really valuable.

And then look, I think it is one of these things where like there's no substitute for actually doing it. And so, if, for people who are interested in doing it, finding, you know, parts of the government now or in the future that are working on issues that feel meaningful and where your skills could contribute, just kinda go for it. You know, see if there's a way to get in the door. And once you start doing it, then the opportunities can kind of emerge from there.

Peter Harrell: Well, that's a great place to leave it. Mike Schmidt, thank you so much for joining.

Mike Schmidt: All right. Thanks Peter.

[Outro]

Peter Harrell: The Lawfare Podcast is produced by the Lawfare Institute. If you wanna support the show and listen ad free, you can become a Lawfare material supporter at lawfaremedia.org/support. Supporters also get access to special events and other bonus content we don't share anywhere else. If you enjoy the podcast, please rate and review us wherever you listen. It really does help.

And be sure to check out our other shows including Scaling Laws, Rational Security, Allies, The Aftermath, and Escalation, our latest Lawfare Presents podcast series about the war in Ukraine. You can also find all of our written work at lawfaremedia.org. This podcast is edited by Jen Patja with audio engineering by me, Peter Harrell. Our theme song is from Alibi Music. And as always, thank you for listening.