--:-- Ben: Russia today is outproducing NATO on 155 munitions. And it took about two years. We have no strategic plan as a country for how we preserve supply in a catastrophic situation. Every war game we run, we run out of munitions and missiles in like six to seven days. And then it takes about two to three years to refill that battery. We shoot all our missiles in one week and then we have none for two years.
--:-- Ben: If the next conflict demands both brains and bulk, can the U.S. combine innovation with industrial scale fast enough to win? On this episode of The Ben and Mark Show, we're joined by Brian Schimpf, co-founder and CEO of Anduril, and Chris Power, founder and CEO of Hadrian. They break down what it really means to out-invent and out-adapt. Ukraine proved that mass production matters and that advanced weapons without industrial capacity don't deter. We also discussed the future of U.S. manufacturing, supply chains in rare earths, automation, offtake agreements, regulatory bottlenecks, and how China's anti-access systems shaped the fight over Taiwan. Let's get into it.
--:-- Mark: Brian, I believe Palmer has said in an interview that as we think about competition with China, it's going to be hard to match the scale from the get-go, but where we can continue to win is in innovation and R&D and that can make up for it. How would you edit that characterization? Or why don't you flesh out more of what that means?
--:-- Brian Schimpf: Well, I think the U.S. has taken this strategy over the last 20 or 30 years based on a very anomalous and incorrect Gulf War experience. The U.S., having prepared to fight on the most aggressive ally with the Soviet Union, had invested in all this technology. They believed that punching down on a third-world nation and winning was an indication that the strategy was vindicated. So they went down this conclusion that technical superiority is the only strategy, and that every marginal improvement we can make to these exquisite weapons, these aircraft, all these things, was going to be the dominant strategy.
--:-- Then Ukraine happens. The reality ended up being much more historically common. These wars are protracted. They are industrial in scale. If you were to bring it down to one variable, the amount of kilograms you can throw at each other turns out to probably be the single most important factor. Things have to be technically superior, but mass really does matter.
--:-- Technical superiority is a huge advantage. We are not going to just out-produce China on dumb ammunition. We clearly have to have an edge around our space capabilities, how we can sense, how we can confuse them, how sophisticated our weapons are. But the part that became so clear out of Ukraine was that industrial production has a deterrence factor all itself, and it is critical for winning a conflict. When most of the war games with China show that we would be out of key high-end munitions in about eight days, that is not a deterrence factor anymore, with no apparent ability to substantially increase production for a variety of reasons.
--:-- The technological superiority does matter, but the belief for a long time was that it was sufficient. A low number of really exquisite weapons is not going to win a conflict. I think Ukraine definitively showed that to the world.
--:-- Mark: What do you think? What are the factors that cause us to be so behind in manufacturing? And are they correctable?
--:-- Brian Schimpf: I will take a stab at this, and then Chris will have a lot more opinions on the matter. The erosion on these things happened quite quickly, and I think it takes decades to build back. We systematically outsourced all of the key scaled manufacturing things at a time when electronics and new generations of technologies became so prevalent. We just never built the skill sets in the U.S. for mass manufacturing at scale, with very few exceptions.
--:-- When we try to run anything down on the manufacturing side, looking for a great head of manufacturing, there are basically no tier-one execs we could find that were American-born. We ended up finding someone from Canada, but everyone I found who was great, thinking about this differently with a really novel approach, almost all were foreign-born. I think it is just not an aspirational job for young founder-mentality people for quite a long time to go into manufacturing. It was just not a sexy, exciting, high-growth space. The original theory was that low-cost labor was the primary strategy. Then China built up this very technical, sophisticated capability.
--:-- To get that back, it is going to take some time. The number of startups I have been seeing that are working on software to enable manufacturing, automation techniques, different approaches to composite manufacturing, wiring harnesses—it is massively ramped in the last three or four years, reflective of the drive to get these things reshored onto America.
--:-- What has worked so well in Silicon Valley is this diffusion process of knowledge, expertise, and people moving between jobs. You get this insane ramp in learning that happens incredibly quickly. That is gone. There is just no concentration of those people in the U.S. So you had this total erosion that happened very quickly. I think it will take a while to build back, but I think it will get there. I do not think it will start with making low-cost iPhones, but starting in defense production, starting in aerospace—areas where there is a national interest in keeping this in the U.S.—and then looking at mass-producibility there. We are finding a lot of great suppliers, not just in that space, folks like Chris, but also in commercial manufacturing, automotive manufacturing. There are a lot of good folks that do exist. It is just figuring out smart ways to tap into them and understanding on the engineering side, on the practices side, how you really do this at scale. That is a lost skill set in the U.S. It is going to take a lot of practice, trial and error, and iteration to get back to. We just seeded that. We said we are not going to do it. We should not be shocked that we do not have anyone as good at it. We have not even tried.
--:-- Mark: Yeah, yeah, yeah. Chris, being in that field and building a new manufacturing company, what are the issues in the U.S. with doing that here?
--:-- Chris Power: I think there are several, and I will break them up into three buckets. One is that most of the manufacturing we need for the things that are onshore in defense is extremely highly skilled. The Apple-McKinsey thesis was, let us offshore the easy parts of production first and keep the value here. That did not really work. It turns out that if you disconnect design from production, you end up with really bad design at the same time.
--:-- Apple trained something like 28 million Chinese people on really advanced manufacturing skills and invested, to quote Patrick McGee's book, like 50 to 60 billion throughout the CapEx. More than the Marshall Plan. It was like 10 chip sacks, basically. Two things happened. Like Brian said, all the jobs went away and therefore no one got into manufacturing. Why would you tell your son or daughter to enter a dying industry? Secondly, people forget that we did not have a defense industrial base in World War II. We just had the best industrial base that pivoted to defense when we needed it. To Brian's point, mass was the only thing that won. I do not even think our tanks were that good. We just kind of McDonald's'd the shit out of them.
--:-- Mark: That is a good phrase.
--:-- Chris Power: Everyone is operating in the mental model that we are still the America we were in World War II. From a productive capacity standpoint, we just are not. We were a data center company and then country, and then we gave the data center to the Chinese. The real problem here is two things. One is slack capacity. We used to be able to put defense spend on top of manufacturing hours that was commercial. All these businesses used to run 70% on commercial high volume with 30% of the business with defense. It was this load balancing effect that just does not exist anymore.
--:-- The real underlying problem is my thesis that basically everyone who is highly skilled in any manufacturing domain is basically 62. It is a skilled labor replacement problem. It is not just the CapEx. The CapEx is actually quite easy as long as you are good at fundraising and you can back it with contracts. It is really the skill. The only way to replace that skill in my mind is high levels of automation to make those jobs simpler so that a new workforce can ramp up in a month.
--:-- From a demand side, you have this demand equation. Even with Brian's products or other people's products in defense that are by number of unit manufactured more than other defense products, it is still nowhere near the kind of commercial scale you would really need to get some reps on. I think there are two ways to approach it strategically. Number one is subsidize a massive product and let the industrial base bootstrap itself around that. Our strategy is basically factory first, so the capacity exists, but it is this chicken and egg problem.
--:-- Brian, I am sure outside of your business, I know you are close to the drone space still. I do not think this applies to Anduril, but I think the Army still has this real problem of, "We really love this drone, we want 10,000 of them, but you also do not have the factory pre-existing, so we are not going to give you the demand signal." You either have to go top down or bottom up. I think we can get there. I just think people forget manufacturing is this real brute force equation where you have to pick like four to five companies and let them have at it at a scale that has never been seen before and let them figure it out. It is in the figuring it out that you get there. There is no real way you can plan around this. Our strategy is high levels of automation to get very leveraged very fast. But there is no easy way to do it.
--:-- Mark: So the way to scale it is really for the government to just place giant orders for things with U.S. manufacturing companies and say, "Look, if you can build this stuff, we are going to buy it all." Short of that, it sounds like it will be slow.
--:-- Brian Schimpf: We are building out a lot of our gigafactories well ahead of the demand because we know that we build them so flexibly that we can pivot that capacity to other products. We are not taking business risk, which is a very unique strategy. If you wanted to let the market correct for itself, someone would just have to make a purchase order of a million cars, force it to be all made onshore, and then let all these small suppliers and the skills come up underneath it economically. I do not think we have the time to do that, so we are just building them all.
--:-- The thesis has always been that in a time of crisis, manufacturing gets very hard. When manufacturing gets hard, people will be forced to partner. We are seeing that play out across the board right now. But if you did not have companies like ours aggressively doing that factory-first strategy, then you would just have to onshore a bunch of commercial volume. At the current manufacturing price points, with no automation, you have to subsidize it to have it here and let the industrial base come up.
--:-- In any near-term sense, the game is going to be high mix, low volume. You have to get good at that. It is very different than how manufacturing, when the U.S. was good at it, was really tailored, which was high rate, relatively fixed. The process to change, the process to iterate was not there. You could not change the mix of what you were making year to year super easily. This focus on just flexible factories at scale, built very cheaply, minimize CapEx, minimize tooling, all these things, and allow it through better automation, through better software, and a flexible workforce, and better design of the products to flex into that—that kind of works. I think that will get us there, where we can just lean into capacity and advance, believing that in aggregate, we do not know exactly what we are going to build, but we are going to build a lot of different things.
--:-- The other side of this is there are key supply chain issues which defense demand cannot solve. We have seen this with rare earths and magnets. China has strategically strangleholded that. Not only did they have the processing of those rare earths, but now they have exported control, so they do not allow the export of magnet-making technology. That is now a restricted national technology that they have banned from being exported from the country. It is a good strategy. It creates a real bottleneck there. Germanium, gallium—there are a lot of these different core materials where the U.S. has to find industrial policy strategies to create viable commercial alternatives.
--:-- The way they are structuring some of these is actually really clever. With one of these materials companies, they basically guaranteed minimum offtake and a price floor. That is a really good strategy. That way, China does not have as many abilities to do dumping strategies or subsidization strategies to undercut other competitors trying to break into the market. This is where I think tariffs applied on more of a national security industrial policy basis can be really advantageous. Create the time for American and allied industry to be able to get price competitive. It is just going to take a number of years. You have a big CapEx bell. You have a learning curve you have to go down to actually get back to the point that you have this capacity available, the expertise available, and it is efficient.
--:-- Ultimately, it does have to amount to real commercial industrial demand that is sustaining this and driving innovation into this industrial base. What ended up happening in the defense world is that it was the last spot where we insisted on manufacturing in the U.S. It became Galapagos Island, where everything evolved in a totally different way than everything else in the world. We are kind of shocked it does not look like anything else that is modern and efficient. It is like, well, what? Just totally different evolutionary pressures. The things we have tried to preserve look a lot closer to what they were in the 80s and 90s than what is going on in China today.
--:-- Chris Power: As of right now, if you look at the PlayStation 5 as a product, that is effectively 90% fully automated production line. Writ large across the board, people really miss this. I think it is strategic CCP comms. There is a little bit of manual stuff for things that robots cannot yet do. But on pure manufacturing autonomy, they are probably 20 years ahead of us.
--:-- Our version of this at Hadrian is that most of this is actually a software problem. People see a robot do a thing, but it is actually all of the machines are extremely dumb computers, basically. It is a coordination problem and an autonomy problem. The one thing we are still very, very good at in this country is software engineering. There was just no pressure to put software engineering automation in manufacturing because our version of automation commercially for the last 30 years was just give it to China. Defense did not need to do manufacturing automation innovation because it was so high priced and protected that there was no pressure.
--:-- I completely agree with Brian, but I think we forget that if you are still assuming that we are great at automation and software as a country, then if you dump enough and go hard and fast enough, you can actually catch up. That is effectively the macro bet that we are doing at Hadrian.
--:-- Mark: You guys alluded to this, but there is this famous book, *Freedom's Forge*, talking about the ramp up of U.S. industrial production during World War II and the repurposing of all these car factories. But as one of you guys said, a tank in 1940 was not like a tank today. An airplane or anything else. Put any of us in a plane or a car driven in 1940, and we would be absolutely shocked at how crude and primitive it was technologically. A tank in 1940 or a car has no chips. It has nothing that we would consider to be modern technologically. Today's weapon or vehicle is far more sophisticated, both in hardware and software. The typical car has like 500 chips an hour or something, so I can only imagine what a drone has.
--:-- The fact that the products themselves are so much more technologically sophisticated and have so many more components, both soft components and hard components—does that mean that the country that is best at technology over time is going to be the best producer of those things because they are so complicated? If all the software is written here, that gives us this huge leg up to build these things? Or is it actually the opposite? The supply chain is infinitely more complex. The kinds of parts in these things are all themselves extremely complicated. If they are all by default made in China, they just own so much of the supply chain that we are at a more or less permanent, or at least very long-term disadvantage. Which way does that dynamic tilt?
--:-- Brian Schimpf: Supply chain. Look at COVID. It was not the microprocessors that were the bottleneck. It was the power regulators. It was these analog components. It was all these things way down. When you even look at semiconductor manufacturing and everything going into it, China still controls upstream components of that in a substantial way. From the point of view of industrial independence in a conflict, who has more of it? China has definitively more industrial independence, the ability to continue production absent their choke points, which are quite a bit more minimal. Some key things on energy, which they are strategically trying to solve, and some raw materials, which they do not have as much natural sources of.
--:-- From the production perspective, it is quite clear to me that the supply chain constraints, if there was a conflict, if China wanted to cut it off, seemed pretty determined. The U.S. would have very little leverage or ability to respond in any timely way or with allies. When you get to who can build better, smarter weapons, have different offset strategies, and think in a more technologically sophisticated way about how technology influences warfighting, I think the U.S. has a significant advantage. We are much better at that. We have a more sophisticated military that is more practiced and understands how to infuse technology into warfighting in a better way. That is quite clear to me.
--:-- But then the question becomes, do you have enough? Can you actually produce it at scale? And can you afford to produce it? Under a conflict scenario, that is quite hard. The *Freedom's Forge* thing is quite interesting to me. We had about two years of Lend-Lease to retool American industry. We were not in the war, but we had about two years to retool industry, and that seemed to be about how long it took. Russia today is outproducing NATO on 155 munitions. And it took about two years. There is a time lag, even if you have the industrial capacity or the potential for industrial capacity, that takes to ramp and repurpose and retool these things.
--:-- I have heard various theories that in a conflict, the U.S. would band together, make all these amazing things, and repurpose. I believe it is probably true. But is it too late? And can we get the very basic components that are necessary for any modern technology? Both of those seem to be pretty clear answers to me. I do not think it could be fast enough. We have no strategic plan as a country for how we preserve supply in a catastrophic situation.
--:-- Mark: I think we have a mythological view. There is a book by this guy, Alexander Field, an economist who goes through the economics of the cutover of production in wartime and then back to peacetime in the 1940s. The picture he paints is, to your point, it was a two-year ramp up to cut over all the car factories to be plane and tank factories. It was a very short production time. Within two years, we basically made all of the weapons and all of the vehicles that basically won World War II. The ships—the production happened extremely quickly. Which it could, because these were not that complicated. They were still mostly bent metal and rubber. They did not have that many parts.
--:-- According to Alexander Field, the spin down of wartime production actually happened before the end of the war. By 1944, we all knew that the U.S. was going to win, so the factory started to be cut back over to making civilian products in preparation for peacetime. Brian, to your point, I think we have this mythology that when push comes to shove, this stuff just happens by magic. Because it happened by magic 80 years ago, it will happen by magic again. That seems very much not the case.
--:-- Brian Schimpf: I think we also miss one important point. Here is my guess on what we call parallel production. You take an existing defense product. How much could you mass-produce and how much is some secret sauce? My guess is that any given missile, about 15% of it is secret sauce or critical. The rest of it is just reproduction, not that hard. I think the thing people miss, Mark, is that CapEx or skills were kind of at a 10-to-1 ratio, and then you could pivot to defense. We do not even make the CapEx that goes into the factories in the U.S. anymore. It is all German, South Korean, Japanese, or whatever, and they are controlled products. If you wanted to go scale a gigafactory at a time of crisis, it is probably already too late. By the way, all those machines have rare earth magnets in them. Guess who controls that?
--:-- Mark: If you guys were in the White House or Congress, Brian, let us go back to your supply chain point. Part of what we talked about earlier was you could place a strategic bet on five companies and spin them up and have them get big. But check me if this is right. It is not just those five companies. It is the other 5,000 companies that are the supply chain for those five companies. What is the policy approach to that that passes the sniff test of something that would both work and be achievable over a five or ten-year period?
--:-- Brian Schimpf: Number one, you need data. There is very little data on what two or three levels down into the supply chain is actually on that bill of materials and what are the critical components. Again, the COVID example: we had no idea. Then we found out that the market was constrained on the thing no one would have ever guessed it would be constrained on. If you have data, you can start to look at interesting strategies. Stockpiling power regulators that are expensive is probably extremely cheap. A lot of these things. You do not have the industrial capacity, but you need to very quickly create mitigating resilience strategies. If you can get some visibility on these things, you can take strategies to either pay industry to stockpile. There are a lot of basic things you can do there.
--:-- The second thing that is pretty key is more of these MP material-style deals. Identify the key supply chain constraints way upstream so that we even have the possibility of being able to ramp in a crisis period. If we are looking at a crisis and work all the way back to the raw materials that we did not plan for, you are looking at five years of lag to reconstitute that supply chain. If you can start cutting that lag and working from the bottom up in terms of allies who already have capacity that you can get them to invest in with some sort of guarantees, or you can invest in U.S. capacity around these things, and then regulatory relaxation around some of the constraints of why it is hard to build in the U.S.—you can start to systematically reduce China's leverage, create competitive industry, and secure supply chain for national security. It is just going to take a degree of targeted focus and understanding of what those key bottlenecks are. I think there is an understanding of the obvious five or six, but I suspect it is 10 or 20 things that we need to have a more sophisticated and holistic policy around. It uses all of our economic levers.
--:-- If everyone else is subsidizing and playing a different game on cost of capital and how they want strategic industries to grow, why is the U.S. not doing it? The government is a terrible capital allocator historically, but low-cost loans, sharing risk with banks, guaranteed offtake—there are a lot of these strategies. The U.S. has the best capital market system. We could use a lightweight government backstopping to substantially create a different incentive structure in the U.S. and get these industries up to competitive very quickly.
--:-- Mark: Chris, Brian mentioned the regulatory inhibitors. What are the regulatory inhibitors that you face now in terms of what you are doing at Hadrian?
--:-- Chris Power: It's pretty extreme, but it's state by state. Oddly, a lot of the stuff that we would stand up in California, frankly, is just flat out banned from an environmental or permitting perspective. Our teams just frankly fight through it. But it is truly a permitting regulatory issue. Mining is obviously a big one. Even the basics, like some of the exquisite stuff can only be made in a certain state to feed the defense ecosystem because all the other states banned it. They're literally in facilities that got certified 50 years ago and you could not build a new one today because you would never get the permit to do so.
--:-- Mark: So it's stuff like that. I think the biggest unlock is certainly the government creating large offtake agreements. In the last six months, I would say there are two big realizations. One is that we kind of just have to pick seven to eight entrepreneurs that we really trust who are making a lot of money, but they're really just patriots. The second one is creating these large offtake agreements so that the CapEx and the commercial market can come behind it. The regulatory side is still a heavy environmental permitting nightmare.
--:-- Ben: Actually, speak about that because there's this weird tension in politics. You have entrepreneurs who can do it, but if there are two—Elon being the classic example—then all of a sudden the government doesn't want them to do it. They imagine that General Motors can take Elon's place. How is that playing out today? How should it change? Brian and Chris, I think you're both on the same page about this.
--:-- Brian Schimpf: My perspective without going to specifics is that there have been two realizations. Realization number one is that you're just going to have to instantiate six to eight massive companies. Instantiating 100 medium-sized players is just not going to have any effect because you need to get scale, you need talent aggregation, all these other things. In certain areas of the government where there are a lot of immature companies, they're going to spend the next year making eight bets and seeing how it plays out. In other areas, in the next two years, you're going to see people getting pulled into rooms and saying, "Cool, we picked you for this category, start running." That will come alongside some MP Materials-style deals and a lot of headbanging.
--:-- My opinion is we've reached the point where this has started to already happen in some of these critical industries. It's going to look like this: imagine you at Andreessen Fund a startup and Apple gives them a billion-dollar revenue deal, and Apple takes some warrants. That is a perfectly reasonable strategy. Otherwise, there is no other structural way to go this fast where the market's not playing it out. You're going to see a macro cultural return to, "Hey guys, there are 10 of you, go figure this out. You fix this, you fix this, you fix this," with a lot of strings and public embarrassment attached if there are failures. Spreading the capital around 100 different companies is not going to work. There will be 10 of you to experiment, and we're going to scale two. That's going to happen across every program if we're serious about it, and it's the right thing in this era to do.
--:-- Ben: Will there need to be some kind of federal override of the state-by-state regulatory regimes to make this work? Or can you work around it with enough money?
--:-- Brian Schimpf: You're going to have competition between states. We've seen this with running factory searches. You can really tell the difference between states who are trying to drive growth and manufacturing and how they think about enabling companies to be successful, versus states who take it for granted. They think, "Obviously you want to be here, so we're not going to go out of our way to help." You get that sense quite quickly. So there's a natural competitive function on the state side that's not overwhelming. Maybe California gets more efficient, but it's probably not going to get as efficient as Texas. Companies are rational. They know this and can make informed decisions about where to put their plants, trading off talent, labor access, and everything else.
--:-- I don't think you have to solve the federal override on the state level per se. There are things the federal government can do with federal EPA issues. States don't even have a lot of say in certain cases. There's this aspect where an area can be designated a non-attainment zone that is over-polluted on a certain pollutant. The EPA says any new construction requires remediating that pollutant by any means available—not cost-benefit, not due process. Dems and Republicans alike, when they get that designation, don't like it because it is a death knell to getting new manufacturing and new investment in that area.
--:-- I have been a lifelong Democrat, and I believe we need some environmental conservation pieces, but the balance has not gotten to the correct trade-off. There are ways to mitigate the harms and expedite these decisions. The time factor is often undervalued. In business, time is extremely valuable. If it takes me five years to build a data center, that's irrelevant. I need a one-year timeline. For a lot of these places, the view is, "I'll get you to the same outcome, it just might take five years," and they don't view that as a problem.
--:-- On a state-by-state basis, there's a ton of competition. The federal government can do a lot to expedite getting to yes. They have unique authorities, especially in defense, to push things on nuclear, to push the envelope on a lot of things where there is a lot of risk aversion. They have a huge opportunity to take the lead on showing what's possible. But we don't need to march into every state house and say we're taking over. There are a lot of easy states and a lot of hard states, and that's okay.
--:-- Chris Power: The talent is movable. We've seen this as we're launching in Arizona. Brian, I know you've got this perspective from Ohio as well. The way we think about it is we will probably do most of our advanced engineering in California because it's very hard to get the conflagration of software engineering and manufacturing in one building. But once you've worked out how to do welding, for example, and it's now a simple system, that can go anywhere. What we look for uniquely is the technician workforce. What we're also seeing is that people at Hadrian, as we expand to different states, they're like, "Great, I get to keep my software engineering or manufacturing engineering job, but I get to buy a house because now I get to go to Arizona or another state." It's less shiftable than people think, especially for the highly skilled jobs. But everything else is a grand adventure for most of Brian's and my teams: "Hell yeah, let's go two years in Texas, or two years in Arizona, or wherever," and follow the companies around.
--:-- Brian Schimpf: I completely agree. For a lot of these, particularly as you get to more hardware skill sets, manufacturing skill sets, getting them to come to California is a huge problem. How much does a house cost? Not that interested. It's actually easier when it's lower cost of living locations for a lot of these skill sets. But completely agree with Chris: a lot of the advanced software is still going to be in California. That technology is then relocatable with relatively little difficulty.
--:-- Ben: Why are we so much better at building data centers than factories? Colossus, the biggest data center in the world, was built in under five months. How do we explain this?
--:-- Chris Power: I think it's very easy. The financial markets understand data center offtake agreements. Brian says to me, "Chris, please build a data center and I'll give you this kilowatt hour of compute at X cost unless you screw it up." All the big capital allocators say, "Great, I'm going to let you put $10 billion in CapEx in the ground as long as you can credibly execute." This is how we build power plants as well. There's a price per minute, you put the CapEx in the ground, and there's a 30-year payoff period.
--:-- Because manufacturing is high mix, low volume, and defense spends money every year, you financially have a one-year agreement. You literally can't get the capital markets to underwrite it unless you're at sufficient scale, which is what we're doing now, where you can functionally underwrite across many different contract vehicles and have this 10-year CapEx spread. Then you can build more data centers if you're good at it. I am fully convinced it is literally just because Amazon wants a data center and it's a 30-year offtake agreement, so you can put the CapEx in the ground. Because defense spends contracts once a year, you can't have that long-term revenue agreement, and therefore you can't have this structured finance agreement underneath it for the CapEx. We've solved that by having an aggregated portfolio strategy. But my one answer is: if the Defense Department created offtake agreements for capability or capacity overnight, that would change everything. Data centers just have that naturally.
--:-- Brian Schimpf: The other part is we've been doing data centers for 30 or 40 years. We've just been doing it. It's a thing. USO knows how to do it. Is putting up the concrete walls and building the cold shell the hard part for any of these factories? No, that goes up really quick. Distribution centers are extremely efficient to build at this point. A cold shell concrete building is trivial. Those go up in less than 12 months, probably three or four if you want to.
--:-- Once you get to manufacturing, the question is: what are you putting inside of it? Do you actually understand the process? Have you built the technology to go in there? The nth copy that Chris is going to build of his plant will go up extremely fast if he can finance it and capitalize it. But the first copy took quite a while. For a lot of these manufacturing plants, they're the first one. You have all the unique regulatory approvals, the unique chemicals you have to bring in, the health and safety risks, and how to mitigate all those things. You're sort of N of one on a lot of these cases. But if you're on the tenth copy, it's going to go fast.
--:-- Ben: When people talk about China's success, some point to their industrial policy, their protectionism, their subsidization of certain industries. We've recently experimented with that a little bit. What can we learn or not learn from what they've done? How should we think about it here? Chris, do you want to take it first?
--:-- Chris Power: The most important thing for people to realize is that China's stuff is not cheap because they have low cost of labor. It's because they subsidize CapEx energy, which is the main cost of manufacturing. 95% of the cost of aluminum is just the kilowatt hour of the power price. They also do export subsidies, which is like a reverse tariff. If I sell a million dollars worth of stuff, I get a refund from the government. They did that for the last 30 years in part to drag the capability off the US because they knew they're very long-term thinkers. Unlike the US, we tend to operate only when we get punched in the face. It was a strategic subsidy.
--:-- The byline is: it's not economic protectionism. It's an unfair playing field because right now it's companies like Brian and me versus the CCP, not versus Chinese innovated companies. We're happy to go toe to toe on an even playing field, but it's not an even playing field, and it's not by 10%. It's 70 or 80% in some cases. If you want the US to be an exporter, you have to do tariffs or some economic policy that corrects for that structural imbalance. Then we'll fight over how efficient we can get in America and how many products we can produce.
--:-- That narrative is really wrong. The level of subsidies they do from the government into their national champions is crazy. It's one of the reasons why most commercial manufacturing is offshore because of that price function. If you want to compete, and we can, you have to create an even playing field. Tariffs are one mechanism, but you have to do something. Otherwise, the playing field is so uneven that no amount of technology can catch up to the China cost.
--:-- Brian Schimpf: China has a track record of laying out five-year plans, like "Made in China 2030." They prioritize targeted industrial areas, and that aligns all the way down to the provincial level around these subsidies, these export strategies, the finance structure of how these things are capitalized—zero percent loans, all these different things. There's a strategic framework of what are the key areas they want to control. Not everything they've succeeded on is in that category. They've also been commercially successful in a number of areas, but facilitated by a lot of things Chris was saying around lower cost of capital and export subsidy pieces.
--:-- There are a handful of these policy levers that are effective. If we end in a world where the government says, "We must make X company succeed," but we all know this company's bad, it's not going to work. It's not a competitive company. That's where US policy has landed: let's give grants to these industrial giants that everyone agrees are no longer competitive and try to save them. It's obviously not going to work. Why don't we create a scenario that allows next-generation companies that can actually compete on a level playing field to fight? That strategy has a lot more legs than handouts to a couple of zombie companies.
--:-- Ben: How would you structure something like that, given that the government has some corruption in it? If you're naming or handpicking the companies you're going to subsidize, particularly in the US system, there could be issues. How would you structure the subsidy so it would be merit-based?
--:-- Brian Schimpf: Get the government out of the underwriting process. Take loans. I want the banks, who are pretty effective underwriters, to actually execute the loans, but the government will backstop and help them on lowering the interest rate and absorbing half the default risk. The bank still has skin in the game. The company has skin in the game because if they default, they default, and the government is effectively a zero—they might actually make money off that deal.
--:-- In those scenarios, you can take advantage of the US capital market structure. It's not 10% better than anyone else. It's several thousand times better than anyone else. It is wildly better. If you lean into that and say, "We actually have great capital allocators that know how to underwrite companies," great, let's just juice the capital stack in a way that is aligned with national priorities around certain industries and capabilities.
--:-- Tariffs, export subsidies, export financing—a lot of these things also work. Those are not winner-specific. They are segment and strategy-specific targeted policies. Grants are, in my opinion, the worst possible form of this. Giving Intel an outrageous amount of money to build a facility that they didn't believe they had demand for—why would they not take the money to build a facility? They didn't even have skin in the game. It totally distorts their incentives to actually be an efficient, competitive company.
--:-- Keeping companies accountable, using the capital markets we already have and the allocators we already have—that feels pretty good. Lean into what we're good at, then counteract at a national level some of these unfair practices from other countries or areas where we want to incentivize growth. We're doing some protectionism because we need it for national interest. Those seem pretty reasonable.
--:-- Mark: It's ironic that we as a country sometimes get critiqued for being overly financialized or too much into financial engineering, but we've noted a number of examples on this episode where we can actually benefit from leaning into that strength and greater financial engineering in terms of aligning good outcomes for the markets with strategic priorities.
--:-- Chris Power: Where else could you finance $200 billion of data center investments with basically no real revenue to show for it? It's crazy. A government can't even do that. It is such an asset. Is it always right? Does it get bubbly? Certainly not worse than how most governments operate on these things. It has enabled the West to do such outrageous things, and the US to do such outrageous things that are capital-intensive in a way that nobody in history has ever seen. That is a huge asset. Does it get a little crazy with some of the really exotic instruments? Sure, but that's on the margin compared to the massive asset we've created.
--:-- Mark: You guys saw Oracle earlier this week. Oracle, in one day, added more market cap based on a single AI data center build deal. They added more market cap that day than most national stock markets in there.
--:-- Chris Power: This is actually my big takeaway from Eric. It is astonishing how good we are at things like that. The pessimistic view is the fall-off to the world that you guys are trying to fix. But the other view would be: imagine if we could be as sophisticated at manufacturing and industrial systems and military systems as we are in data centers. To Eric's point, it's not completely clear why they're that different when it comes down to it.
--:-- Ben: Guys, we brought up a couple examples, but I want to get more into what's misunderstood or not fully appreciated about our competition vis-a-vis China, whether it relates to manufacturing or security, defense—either the state of today or where things are going.
--:-- Brian Schimpf: America has a very hard challenge. We always fight an away game. We're always projecting power somewhere around the globe. That is a very hard challenge. In recent history, vis-a-vis Ukraine, and what would pragmatically happen in a Taiwan scenario, there's American support with the ally under attack being able to defend themselves. That's the backdrop of how the US thinks about these things.
--:-- That presents a very hard problem. In a Taiwan scenario, your ability to stage, resupply, and get things in is the critical question. China knew this, so they have invested systematically in technologies that push back the US and push our whole strategy way out. They've invested in space-based sensing. They can find our ships and aircraft way out, thousands of miles away. They've invested in the DF-26 missile, called the carrier killer missile, which can shoot about 1,200 miles out, making carriers and warships massively at risk. They've got very long-range anti-air missiles that can shoot down aircraft. They've created this impenetrable bubble, knowing that if they can break the ability of the US to get close, to stage, to resupply, that cripples the entirety of the US war apparatus.
--:-- Combine that with their production capacity. There's something like 250 times the shipbuilding capacity, massively outpacing on the weapons side, on drones. Category by category, they massively outpace on production. Even if their stuff is half as good, they've got so much more of it, and they have systematically built up a system that can target us at range, strike us at range, and hold everything at risk. That breaks the entire US war strategy.
--:-- The counter to this is that in a Taiwan situation, the geography is very hard to sustain the rate you would need to actually occupy. The goal is to defeat a Chinese occupation, not just an invasion. You have to sustain extensive maritime campaigns, and that is hard. There are very few places you can land. The calculus basically boils down to: is Xi going to be successful? He stated his intention clearly. He wants to reunify if necessary. He stated that he wants to have the military ready by 2027. The whole game is really just making it clear that their strategy, which in many ways broke the US war plan, has a counteracting strategy. Taiwan is prepared. It will be too hard to be successful. The odds are not in your favor, because the loss there is probably the thing he fears the most.
--:-- Ben: Chris, anything worth adding to that?
--:-- Chris Power: Brian and I talk about this every time we see each other over dinner. The one thing that always shocks me, and probably shocks Brian, is how uneducated the smart people in the country are about how bad this looks. The easiest example is: we run out of munitions, missiles, whatever, in every war game we run in like six to seven days. Then it takes about two to three years to refill that battery. Apart from the strategic layer, every single war, if it goes hot, both sides throw everything they've got at each other over a time period. If it gets protracted, how fast can you reconstitute? Brian's correct. It's not like we're 50% behind. We shoot all our missiles in one week, and then we have none for two years. That is how terrible the situation is.
--:-- One last point to Brian's point: because of the one-child policy, China's population is going to drop off very quickly at a certain period of time. This creates a forcing function where, in my opinion, the Taiwan scenario is more likely to happen because if they don't hit this window, they're not going to have a chance in 30 to 40 years. They have this huge population drop-off coming really fast down the pipe. There really is this Davidson window. Over the next 10 to 15 years, it's going to be this constant CCP-style pressure game of test, test, test, boil the frog, boil the frog.
--:-- The only thing we can do is focus on attritable systems. If you kill a carrier and they've got that range, there's nothing we can do about it. Counteract sensing in space. Mass produce those systems so that we basically blow everything in the seven days. We've got to have the manufacturing battery to be able to refill the clip extremely fast. It's going to be this pacing event. Unless it goes hot in 2028, it's going to be 10 to 15 years. As long as we can hold them off there, then their population drops off and we're good for another 300 years. That's how I view the equation.
--:-- Mark: 300 years is a long time.
--:-- Ben: It seems like on one hand, there's the extreme bearish view that people like Peter Zeihan have about demographic collapse, combined with their future challenges in energy and food. He thinks they're going to implode in the next few years. On the other hand, you have this extreme bullish approach: they make everything, they're smarter, et cetera. There seems to be a gap in a sober perspective that analyzes where they're great at but also acknowledges how much they're overleveraged, the state of their economy, the slowing growth, and accounts for demographic challenges as well.
--:-- Brian Schimpf: Are they ahead in manufacturing and defense production? Absolutely. To Brian's point, the buildings in China—if you knock open the concrete wall, is it just cardboard? Yes, most likely. That is their culture. But if you're at 1,001 in production, it doesn't really matter. The economy is under serious stress. But again, all these economic stresses make it more likely that they have to do something versus just not doing it because they're starting to get weak.
--:-- If we were 50% behind on production, it would be fine because half the missiles are not going to work since they build them like they build houses and apartment buildings. Half of them are concrete. But at 100 or 1,000 to 1, it doesn't matter. It's just mass.
--:-- Chris Power: Historically, China has been shockingly good at overcoming a lot of these challenges if they commit the state apparatus to it. A lot of people apply the Western view: prosperity for the population vis-a-vis economic growth is the only religion. It's the only driving principle. If you tank the economy, you would be voted out. That's the dominant factor. Xi is not that.
--:-- Chris Power: Like, his goal is preservation of CCP and his legacy. And all his actions are commensurate with those things. And so the economy is a supporting role to that goal. He has demonstrated that he does not care if he harms the economy if he believes it is in the interest of national security and preservation.
--:-- How he views allies and partners, it's always viewed through a different lens of Chinese power. I don't think a lot of people put the Western leader rational, growth-first decision-making against Xi's strategy. There's no evidence that's how he operates. All the evidence is actually that it's much more a preservation strategy and a legacy strategy.
--:-- When things are under threat, when things are under duress, he'll choose the thing that he thinks is most aligned to that goal. If that's dealing with the economy through preservation, great. Zero the debt and a whole bunch of things take a bath, he'd probably be fine with that too. It would keep him in a powerful position and preserves his legacy and preserves his CCP. So I think the calculus is just wildly different. When people try to apply Western rules to this, you get very incorrect conclusions that don't seem consistent with his actions or the results.
--:-- Ben: You guys are on the front lines of helping fix and improve our situation as it relates to manufacturing and national security. We're honored to be supporters. Brian, it's great.
--:-- Brian Schimpf: Thank you.
--:-- Ben: Thank you for coming to the podcast.
--:-- Brian Schimpf: Thanks, guys.
--:-- Chris Power: Thanks, guys.
--:-- Ben: Thanks for listening to the A16Z podcast. If you enjoyed the episode, let us know by leaving a review at ratethispodcast.com slash A16Z. We've got more great conversations coming your way. See you next time.
--:-- [Closing announcement]
--:-- As a reminder, the content here is for informational purposes only. It should not be taken as legal, business, tax, or investment advice, or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any A16Z fund. Please note that A16Z and its affiliates may also maintain investments in the companies discussed in this podcast. For more details, including a link to our investments, please see a16z.com forward slash disclosures. Bye.
