Live from RNDF: Substrate, Semiconductors, and the Brave New World in Chips

The work that we're doing is very ideologically driven. To go and do this, it is incredibly, incredibly hard. I think most people would believe that your sort of guaranteed to fail, but we see this as one of those things where we have to really as a nation throw everything against the wall at this point to try and solve and really take advantage of what is the opportunity of a mass proliferation of compute, of AI. And so we have a, I think, a very, very rare chance now to actually do that asymmetric bet, turn the tables once again. And so for us, it's not worth doing this for money. It's only worth doing if this is something that the United States desperately needs. Welcome! I'm Roger Zachime, host of Reaganism, and we have a really exciting episode today. We're here at a special location in Seamy Valley, California at the Reagan Library. We're hovering over our shoulders. Is Air Force One, where Ronald Reagan flew during his presidency today. We have a very special guest, James Proud, who is the CEO of a very exciting company called Substrate. And before that, he was a serial entrepreneur, startups, including Hello. And before that, he was one of the teal fellows, whereas I understand it, Peter Teal encouraged very smart young people in university or think about going to university to drop out university and just start to pursue their creativity and ideas involving any particular, particular technology or startup. James, you are a product of that world, and you've landed on Substrate. Welcome. Tell us about it. Thank you. It is quite a background to have Air Force One. I know. You're not allowed to touch the wingtip, but it's like so close. I know, you can do it. I know. It's incredible. So, thank you for having me again. And Substrate, really, our goal is to return the United States to what I think is its rightful place as the dominant leader in the production of advanced semiconductors. We historically, as a nation, invented all of these major technologies over the past decade. We slightly lost our way, but I think that now more than ever, it's important that how do we get back to that position in leadership? But you have not spent a career in semiconductors, right? And this is new for you, at least since 20-something. Yes. 20-something. 20-something. Yes. No, this is something that I've been thinking about for about seven or eight years at this point. And the company itself is four years old. And really what we've been doing is trying to find, is there a different technology pathway to get us back to that sort of leadership and fabrication, both low cost and low complexity? So, low cost, low complexity for chips. Correct. I think people who have been listening and watching this podcast understand semiconductors and their importance. But just tell us how difficult it is and what the current landscape looks like that makes it so complicated and so expensive. And while you think James Brown, you can disrupt the entire ecosystem of chips. Well, I don't know about entire, but tactically going after a few key areas. Look, when you look at a modern-day semiconductor fab, they're getting close to $30 billion a piece. That's a double in aircraft carrier. And if you plot this out towards the end of the decade, we're trending towards about $50 billion for these advanced fabs. So this is really going to make the wafers and the chips that they produce astronomically expensive. And they already are on the leading edge very, very expensive. And so what we've done is really sort of take a look at from sort of first principles, why is it so expensive to build a fab? What goes into that cost and the largest component of that cost is this key step in that fabrication process called lithography. This is where you take credible machines, very short wavelengths of light, and you shine that onto a mask and you shine that then onto the wafer itself and imprint that image. Like that is the key cost driver and also complexity driver. And this is the one area of technology where we've been told the United States is full and behind. We can't catch up. China can't catch up either. We've got to sort of leave it where it is. And we took the sort of slightly different view of we invented all of these technologies. We can probably do it again. Slightly different view. This is the single point of failure where, as I understand it, only one company in the world can actually deliver this. And you think you could do it differently? Yes, it does sound a little crazy. But when you trace back that history of advanced EUV lithography, it was actually three US national labs that invented this technology. In the late '80s and early '90s, it was Lawrence Berkeley, Lawrence Livermore and Sandia National Lab. And so American scientists are government funded labs, the best and brightest, but the government funded labs. Even better, the biggest funder of this was actually Intel. And if you actually look at outside of US national labs and US educational institutions, Intel and IBM are like the two companies who have really invented all of this in the private sector. And so Intel funded it predominantly. There was a consortium of companies, but Intel was the major one. And then US scientists and engineers really went through all of these core fundamental technologies. It then took another sort of 15 years of scaling this up. And that's when we lost it overseas in Europe was able to actually run ahead and commercialize that. We lost it overseas. Now we've had Chris Miller on the show, we know about chip wars, we know about how TSMC became the fab of choice and the US fell behind. But we're not talking about TSMC and the fabs right now in Taiwan. We're talking about SML in Europe. So tell us the story there as you understand it, why it left our shores. Yeah, so I think for both the fabrication and the fabs moving overseas, but also the lithography, these were two very smart companies and countries who made asymmetric bets. They bet on technologies which didn't yet seem like they could maybe scale, they maybe had issues. But the key thing about the US semiconductor industry, and I think this lesson applies for us as well, is that what really happens is you make a bet that takes five years to get to a point, and then it takes another five years for it to really play out. So a decade can go by and then suddenly overnight the tables have completely turned. And both with Europe supporting the scale up of this technology and Taiwan supporting the implementation of it, they made those bets and we sort of got to the point where we are today. I remember when we met each other, came to your headquarters in near and around San Francisco. Great to meet you, love walking in and my recollection. We're a house sort of place, lots of people working, you know, as you'd expect to start up, you know, everybody's sitting next to each other, meritocracy sort of feel and a big flag in the back. And that came through in space in our conversation. For you, this is about the red white and blue, this is about America and America's security. Tell us about the presence of the flag in substrate. I'm talking metaphorically about it. You can feel free to comment on the flag that you have in your spaces. We've got an even bigger flag. Is that okay? We're competing with card dealerships because they have pretty big flags. We need to get tech companies to have flags like card dealerships. We need to have like, I think a reinforced pole put on the roof so we can actually have a true quality. I might get in trouble. No, you should do it. San Francisco, I think you're leading with that. You should do it. Yeah, no, we're like, I'm very clear. I think most people would believe that your sort of guaranteed to fail. But we see this as one of those things where we have to really as a nation throw everything against the wall at this point to try and solve. And really take advantage of what is the opportunity of a mass proliferation of compute of AI. And so we have I think a very, very rare chance now to actually do that asymmetric bet. Turn the tables once again. And so for us, the only, it's not worth doing this for money. And we believe it does. I love that. But you're not going to get $100 million in venture capital just on patriotism alone. 100%. No, should you? And so talk to us about the challenge. What is attracting this sort of capital? Because, again, they're not going to bet it on patriotism. They're going to bet it because they see something here that could truly be transformative. I'm looking at an article that came out not too long ago. How to kill two monopolies with one tool. Well, that's provocative. So if you've got a story behind that, how you can disrupt TSMC, the fabs, and the lithography in Europe. Well, maybe I'll bet on that because the off chance it works. Wow, we've created a new new world. Is that the story in short? How you get 100 million bucks? I think this is the epitome of the American capital system and venture capital, where you make an asymmetric bet. You see potential upside that perhaps is sort of not being valued correctly at that moment. That's why you make the investment. Right. And hey, if this team can sort of put off the impossible, that's when you have the sort of outlier returns. And that sort of found us fun, Peter Till, sort of real mentality of how he approaches these things. And they're one of our major investors. So yeah, that's really the sort of bet. And as it works, then more capital comes along, but really credit to the VCs who saw this early. So there's a lot of coverage. In October, you sort of had this coming out and telling the world about what you're up to. When we had met, it was sort of below the radar, deliberately, very below the radar. And then you had the great debate about, okay, well, is this thing viable? Can this happen? And when you go public and you're saying something pretty profound like this is going to be no shortage of critics, right? And people questioning, what I liked is your response, at least as I saw in the press, to those who said, well, this is really hard. It's unlikely to happen. And instead of saying, no, we're going to get this done, you're like, no, it is hard. And it's going to take a lot to get there. And specifically, they were challenging, as I recall, the ability to scale, to move from the lab, to actually be something that can be scaled and compete globally with some of these other players. Yeah, and I think that, as you said, we spend a very long time not even having a website. Like, no public presence, and working through a lot of these very challenging technical problems, but also making sure that we fully understood them. And so the fact that we decided to go out there publicly and say, we exist, this is what our purpose is and what we're trying to achieve, that in itself came from a place of, we feel like we have a very good understanding of what needs to be done here and how to go and scale this. That doesn't mean that there's not the lot of work and the engineering to go and do. It's more, we actually understand what is involved here. And we feel confident that we can go and do that. Did that go as planned when we went public? You know, everybody was swirling talking what we were up to. We were happy with the outcome. I personally didn't really read the press, but from the reception and how our team saw it, and the people close to us, yeah, I think it was as well received. All right, so the critics have weighed in. Obviously the venture community came in, you know, not just with words, but indeed. There was a New York Times article, and I take it maybe, you don't read the press in October, when you guys were being profiled, and it took the vantage point of a meeting. They characterized you as an unassuming British-born American without a college degree, but your mom and dad be proud of that description. I think it's accurate. Okay, and you were sitting in Vice President JD Vance's office. That's what they were talking about. And you were explaining how what substrate was doing. Why was Vice President Vance interested in substrate? I don't want to put words into his mouth, but I think that if you look at his public, he did an incredible speech in March, I believe, at a Paris AI summit, where he spoke about the importance of the United States to really win this race. And if you believe you can have different sort of outcomes, there's the, hey, we're going to have artificial superintelligence. And if you believe that, then 100%, the United States needs to win that. But then if you just believe, hey, AI is a very powerful tool for good, but then also potentially for bad in the wrong hands, you want to make sure the United States wins that race and sets the norms and the rules there. And so I think that he rightly has seen, and the administration has seen, we can't just be content with making the models and designing the chips. We have to make the chips. We have to own the supply chain. We need to own the energy, build the transformers. We have to have this in our control. And there's not really many countries in the world who can sort of lay that down and say, this is important for us. We need to own this entire stack. But I think that the United States and certainly China are the sort of two nations who can go and do that. And so I think he and the administration rightly recognize, like, hey, there might be a chance, but if that's a chance, we need to pursue. Let's just say you succeed. And all of a sudden, TSMC, which is this hugely profitable fab, you know, company that, that, that, you know, the fab of choice for the Nvidia chips and everything on down. And, you know, ASML is no longer, you know, the place to go for lithography. It's substrate now. What does that look like? What happens to our friend, Taiwan? What does that, you know, what does that mean geopolitically if you can look out beyond the confines of you trying to just deliver on this technology? Yeah. I don't see, when I sort of think about what we're doing, I don't really see it in opposition to any country or any company. I see this more as the sort of natural progression of technology and really sort of the natural, the natural sort of progress that the United States makes in areas that are of national importance. And so I don't sort of look at it personally in that view. That's for, that's above my pay grade. That's for policymakers to sort of see. But what I do-- You'll let the Vice President worry about that. Not you. Yeah, that is-- I'm not elected in any way to think about that. Mine is just make the technology work. But really, I see that, look, we are success provides more optionality to the president, to the administration. And I see that that is of great importance, and we take that very, very seriously. And so when we say we believe that we can deliver an exquisite technology capability and just to put in perspective for advanced lithography. Right now, depending how you count, there's nine countries in the world who have nuclear weapons. Now there are two countries in the world capable of printing at the resolutions that define really advanced lithography. Of course, we know that the European group, but now it is the United States and it's through a startup. And so we take that very, very seriously and really believe that, hey, success gives us, as a nation, more optionality. Orrin Cass, who is a public intellectual, wrote a piece, often he has affiliated with a national conservative movement. And thinks broadly that we should have bigger government, or a government is a big government here to stay. And it's about what government does, as opposed to more Reagan-esque outlook, where we believe government should be limited just in terms of the constitution mandates the government. But with that disclaimer aside, Orrin wrote a fantastic article, National Affairs, a number of years ago, where he highlighted how innovation that is separated from manufacturing. Ultimately, results in loss innovation, what he was saying was, is that if you simply are designing something, and you have that IP, but the manufacturing is overseas, eventually design will be elsewhere. We'll lose that know-how. How does substrate answer that, as it relates to chips? Well, I think that, I would say I agree, if you lose the manufacturing, then even if you're designing the thing that gets made, if you're not actually able to make that thing as well, I don't think it really can work in the long run. Actually, taking a step back, Andy Grove, the famous Intel CEO, founder, really sort of forced behind that entire thing, he had an incredible op-ed in Bloomberg of July 2010, where he spoke about this exact thing. And even when through this, there's a line where he speaks about to solve this, he says, "I recommend a trade war, and if it ends up being a trade war, then we fight to win." And that was, I think, back then, suddenly an unpopular position, but really I think encapsulates how he saw that if you lose this, you are in a very, very bad situation, and you sort of need to do whatever it takes to regain that ability to own the sort of core means of production. Means of production. All right, well, okay, let's go back to what Americans probably care most about, which is how much is this going to cost? And from a kitchen table standpoint, will this mean goods can potentially be cheaper or don't have to increase in costs in the way that we've seen it, take us through the AI landscape and the huge amount of capital going into it. If substrate is successful, what does it mean for the data farms? What does it mean for all the things that we want to see out of AI, but are worried we don't have the capital, or even the energy, the ability to deliver on this? What would substrate contribute to solving that problem? Yeah, so, as you said, with the energy and these huge infrastructure projects, there's a lot that really goes into that cost, but when I think if you map out on that current trajectory that I spoke about, where fabs go from costing 30 to 50 billion towards the end of the decade, what actually happens to the wafers that they're producing? Well, they go from around $30,000 right now on the really leading edge to close to $100,000 a wafer. When that happens, the cost of those individual chips are going to get pretty prohibitively expensive. And so, when I sort of imagine, okay, what does this future look like of sort of spreading intelligence to be useful to the American people and American business, that requires that to be accessible in price. And so, for our piece, we really plan that, and aiming for, by the end of the decade, we're much closer to the $10,000 a wafer than the $100,000 a wafer. And sort of extrapolating that out a bit more, I really see that these models that we're using today, they're going to keep getting more powerful, but one of the things that they're going to be able to do, I think, relatively soon, is design chips better than we can. When that happens, you're going to see the ability for there to be a proliferation in the number and types of semiconductors. And really, we need to have a cost model that actually supports that proliferation. And that, I think, is the real key opportunity for the United States, because if we can lower the cost at the same time that we increase the sort of space of design and actually build out that fabrication that supports it, then that's how we get back into that, from just a sheer numbers perspective, that leadership position. Where will you build out? Where are you building out? As you exit the lab and attempt to scale, where is that going to happen? Yeah, we've not announced a site selection yet, but hope to do that relatively soon. But in my mind, I want to have multiple of these facilities dotted around the United States and not sort of picking anyone like Final Home. Reagan National Defense Survey just came out and showed that 65% of Americans support the U.S. government strictly limiting U.S. companies from selling the most advanced AI chips to China. And American people view this as something that is economic competition, national security, and the most advanced chips. Super majority Americans do not want sent sold to China. A question for you. Is there a substrate in the People's Republic of China? We believe, yes, we believe there's probably more than one. We spend a lot of time sort of looking at what China is doing and I won't go into too much related to that. But really, I think if you look at Huawei as a competitor, and I think this is a lot of the discussion that is going on now about, do we or don't we, what does Huawei and others do in response? And I would say that we as a country need to do whatever we can to make sure that we continue to leap ahead and push that gap as far as possible. But we've been very open in our views that Huawei and the likes will be very formidable competitors. And that the gap that currently exists between the leading edge chip reduction and where China may be today is probably going to close much, much, much more rapidly than people would expect if people expect that at all. And if that happens, then the position that we're in now might become very tenuous. And so we really need to sort of keep that lead as far as possible. And there's a debate on what you do with or without export restrictions, but at its core, build advanced technology. And keep that lead, keep that technology lead, and that's where we see coming, substrate coming in. Last one, and then we'll go to our lightning round. Oftentimes when I have a chance to engage with people like you, entrepreneurs, visionaries, people who are not going by inertia, but actually want to disrupt and do something entirely new, shift the paradigm. The key piece is not just a venture capital, although that's critical and important, it's a talent. James, as you go out there and you're building this, you need great people with you. Have you been able to find those people? Clearly you've found some people. What is their reaction? Is they're like, yes, I want to get after this problem. No one's actually giving me this opportunity. We've been doing this the same way for decades as a relate to lithography, and I've actually wanted to do this differently. What's been your experience like are you getting the types of people that can deliver this? Yeah, I think that you perfectly summed up the attitude. I think that we as Americans, we're very bad at playing catch-up. But if you say to an American scientist or engineer, what about leaping ahead? Oh, suddenly, yes. That's far more of a pin one in. They want it. And we've been able to build an incredible team of the best and the brightest this country has to offer. And fundamentally, this was a bet early on on American talent. Do we still have the best national labs, the best universities, the best private industry? And I think it's yes all around. And so if that's the case, if there is a technology pathway and we can bring everyone together moving in the right direction, we can go and do this. Lighting round. Here's where we take off the start-up guy founder cap and put on your Reagan cap. Give us your favorite Reagan speech quote. But we'll take all three, two or just one. What do you got? We win. They lose. Wow. That was a good one. Appropriate. Any day after that is a good place to end. James Proud. Thanks for being on the show. Thank you so much, Roger.