Hello, data nerds. My guest today is Pat Gelsinger. Pat is the chairman of Glue, which builds tech products for faith-based communities. Pat previously served as the CEO of Intel, which is one of the world's most important companies and a company I truly admire. Also spent 30 years at Intel, became the first ever CTO. He was also the CEO of VMware for nearly a decade. And he's a general partner at Playground Global and executive chairman at Xlight, a semiconductor startup developing free electron lasers. Pat, welcome to Summation.
Thank you, Orin. Great to be with you today.
In like 1990, which wasn't that long ago, the US made about 40% of the world's chips. Today, it's like roughly 10%. Why did the US lose its edge during that time?
Yeah. And, you know, in some regards, there was never a vote of Congress to get rid of that terrible semiconductor industry. To say that there was ever a decision, you know, and to some degree, I'll say, you know, as I like to frame it, there were decisions in Taiwan, China, Korea, right, to get this industry, right? You know, there were substantial and you know, the chip war, you know, Chris Miller talks about something like that. There was long-term industrial policies put in places in other parts of the world, right? In the US, well, that's capital intensive. You know, we can make better margins on software, right? You know, why should I commit, you know, $10 billion to a FAB project where you don't get cash flow for five years, you don't turn, you know, a positive profit for eight years. That's hard, right? Great. You know, so all of a sudden you saw our supply chains starting to migrate to where the industrial policies were reinforcing the value of long-term capital heavy manufacturing industries and then three decades later what happened, right? You know, the entire industry has migrated to Asia. Then we wake up and then it's sort of like, oh my, we care about these things. Then COVID happens and chips act happens, we got to do something. Well, we weren't doing something for 30 years, right? And all of our Wall Street, you know, near-termism that, you know, focus on higher profits, you know, leveraging other people to do the hard manufacturing work.
Okay, we woke up. If we want to go from like 10% back to 30 or 40%, like is there some sort of realistic timeline? Is there some sort of like realistic plan for us to get there?
Yeah, you know, I think it, you know, it took three decades for our supply chains to setvent there, right? So to think that they're going to come back quickly is just naive. It was always this, hey, you know, if we sort of take a decade long view, can I make a difference in a decade? And the answer to that in my mind is yes, right? But you have to take a decade long view. You need meaningful pools of capital, right? You need consistent industrial capital, right? You know, investments, you know, I need, you know, big dollars, you know, going into this. I need real policy support, you know, and ultimately I need to create economic advantage to doing this in the US.
Do you think you need government capital or do you just need the government willingness to say you're allowed to build plants, you're allowed, you know, more like deregulation or kind of change regulations around it?
Yeah. You know, and I think it is so hard to move supply chains hard, right? I mean, just so hard, right? That I think without the government being involved, right? You know, to say that the private industry is going to fix this. Okay. To me, that just seems naive, right? You know, yeah, private industry allowed this to migrate away, right? So, why would you think that private industry can bring this back without substantial government involvement? I think that's just naive, wishful thinking. As black swans have shown, right, you know, as people like Ray Dalio talk about it extensively, you know, markets do not, you know, price in long-term risk, right? So the markets will not get us here, right? You need industrial policy, right, with real economic policy to make this happen. So when we started chips act, it was clearly, right, you know, hey, we need long-term R&D, check, right? We need real shift in capital pools, check. You know, the thing that we didn't do yet in my mind, and you know, this administration has been emphasizing tariffs and others to create economic value and advantage to being in the US, but you need all of those to come together. And you need to stay consistent on that, you know, for a decade or more because it takes you that long.
We're a few years now in like the chips act. Like, is it working?
You know, it's having some benefit because clearly, you know, we've seen big expansions of TSMC in Arizona, Intel's investment, you know, Samsung building, etc. You know, so clearly we have more fab starts occurring in the US than we were before. You know, next would be some of the memory work we've seen Micron, you know, big expansion announcements and, you know, they're underway, you know, in New York. So I'll say, you know, it seems to have some of the right characteristics, but we still have a long way to go. Now, I'm still, you know, critical that we have no leading edge TSMC in the US, no leading edge TSMC R&D in the US. We still have way too much dependency, right, on many of the, you know, long-term R&D pieces are not in the US. So, we still have a ways to go, but I'd say some of the early indications are more positive than they were in the prior decade. But we have a long way to go, right? I also argue, hey, I think tariffs are great, you know, a factor to drive the economic value back to the US, you know, but they need to be very consistent over a long period of time and big enough to make it valuable for the users. You know, most of the, you know, purchasers of fabs are, you know, the people who are buying wafers who are the fabulous companies today, the Nvidia, the Apple, you know, the Qualcomm, the AMD, etc. They need to see real economic demand for them to be back in the US.
Yeah. And the tariffs are tough, too. It's hard to figure out because you all these semiconductors, they're made up of all these other components by definition. Like so many of those components are global. And so if you're you have to figure out, okay, like you have to really target it very very specifically. We're put tariff here, but we might not put this tariff on this other component because that's actually important to bring semiconductors to the US.
Yeah, it's very intricate. You know, supply chains are very complicated, right? You know, where's the packaging, where's the testing, where's the substrates, you know, right? You know, not one. So, you know, and that's why I say you need good industrial policy over a long period of time with real economic incentives.
And you've also called for a US sovereign wealth fund, some ways to counter China's tech investment. Walk us through that idea.
Yeah, and obviously, hey, the sovereign wealth fund, you know, we got Singapore, we got Australia, we got China, we got the Norwegians, etc. So, the idea of a sovereign wealth fund, okay, you know, I mean, you know, for the most part, you know, the US has chosen to debt finance, you know, many of the things where other countries are choosing to investment finance many of the things. So I just say in that regard, you know, and I think most people are starting to wake up and saying, boy, you know, I prefer, you know, my retirement going into an investment fund as opposed to a debt fund, right? So the idea, right, of a sovereign wealth fund is something that I think the time is long overdue in the US. We see what some of the things like the superannuation fund in Australia, you know, they now have trillions of dollars that they get to start pouring into their national agenda. So, first I'd say the idea for the US is long overdue, right, in that respect. But then you peel the second phase of it and you sort of say, well, what are we going to invest in, right? If we have a sovereign wealth fund, you know, how do we decide what to invest in, right? And obviously the risk of that is, you know, be influenced by geopolitics and all of that kind of stuff becomes too influential here. But, you know, I'd say, you know, boy, if we would look at it today, if you and I were, you know, would get the 10 most important, you know, academic and policy, you know, technological policyminded folks in the US, you know, we're going to agree pretty quickly, right? You know.
I mean because I'd be a little worried if we do agree.
Yeah, absolutely. Because we're going to say, you know, hey, I need long-term minerals in the US, you know, rare earth minerals and mining, you know, right, to enable it. Okay, got it. You know, second is I need semiconductors. It's the underlayment of everything, right, that we do. And third, I need energy. Like, if you have a wealth fund, the goal is to make a good IRR on the wealth fund. And, you know, some of these things are like, okay, we need the government to do this, so maybe it's actually not a good IRR, but it's still good for the long-term economy, good term stability of the country or something, right?
Yeah. And a sovereign wealth fund. Clearly, you're overlaying. I want good return, but I also want to be targeting of things that we as a sovereign entity can take care about.
Because we could do that today without a sovereign wealth fund. But we're not, right? You know, those actions take extraordinary efforts. And remember how much shoe leather was burned to get chips act across the line in the bipartisan way, super hard. A sovereign wealth fund could be a mechanism. Not saying it's the only one. I'm not saying Congress shouldn't do things like the chips act, but boy, I think sovereign wealth fund, right, you know, with a good policy overlay, how we decide, and today, you know, if we picked those three areas and said, hey, we want a good IRR but we want to make sure that you're doing those three things, semiconductors, materials, right, and energy. Those are the national priorities for the next decade. We can get consensus on that, you know, and let a good investment team go to work. I think it's going to enhance the power of the nation. Not detrimental in any way. Most sovereign wealth funds in the world, they take their capital and they invest outside their country. The Saudi sovereign wealth fund, the Abu Dhabi sovereign wealth fund, the even like the Norway sovereign wealth fund is mostly taking that and putting it into America. We're getting that money from their sovereign wealth funds, right? And by the way, look at what was done in the tariff negotiations with Japan and Korea.
Yeah, exactly. We're essentially sucking their sovereign wealth fund putting your money into us.
Yeah. And by the way, you know, hey, my opinion, you know, I want to make those three industries great, you know, let's make, you know, semiconductors, you know, energy, right? Minerals, rare earths, right? You know, those are foundational, right? To the nation. Make them good. I think sovereign wealth fund is one of the tools to do that. I think tariff policy is one of the tools to do that. You know, like I think acts of Congress, right, with industrial policy is one of the tools to do that, you know, and frankly given where we are on those topics. I think we need all three.
Now, China is reportedly building like an EUV prototype with, you know, former ASML engineers that they brought over and stuff. How have you, how do you assess their progress in advanced lithography? Because that that seems to be one of the big things that's holding them back. Like they can't buy the ASML stuff or they can't buy maybe some of the stuff that you're working on.
Yeah. Well, you know, the first thing is, you know, remember, you know, we're not letting the good stuff get to China. You know, by international agreements, they've got to innovate on their side, right? Yeah. So, they have a huge gap. You know, they have radically overinvested in their semiconductor industry, right? They have chosen to aggressively capitalize that industry. And I think, you know, by most, you know, if you look at some of the historical equipment purchase indexes, you know, they've been buying equipment at more than 2x the rate that their market share in the world would justify. So essentially they are stockpiling, you know, deposition machines and, you know, etchers and wafer handlers, you know, all this kind of stuff. So they are, you know, over capitalizing that industry, you know, as China has done in other industries, but they can't get the most advanced stuff. So every evidence is that they're investing heavily to have their own EUV because the world is not letting ASML's EUV machines get there. But they're also needing and for that they're also looking at leap aheads, right, like free electron lasers as new light sources. You know, one of our companies, one of my companies, Xlight, you know, that we're working heavily. So, you know, I do believe that they're saying, you know, boy, you know, catching up with what they've done and have perfected, that's going to be hard. Can we, you know, are there leap ahead opportunities? So, yeah, I think they're pouring, you know, enormous amounts of energy here because largely, you know, they're sort of stuck at about 10 nanometers. Intel proves you can do seven nanometers without EUV. It's just hard, right? But the rest of the world is going from three to two to one and a half, you know, so that gap is, you know, about a decade, right, of lithographic capacity and it's getting bigger because they don't have the EUV. So to some degree, you know, their motivation to get to the other side of EUV is increasing because the gap between their capabilities and where the rest of the world is is increasing in the absence of that because they're never going much below 7 nanometers without something like EUV. So, of course, they're looking to leap ahead. And hey, there's lots of things that would say that free electron lasers is the next light source after LPP, which is used in the EUV machines of today, is a big investment area for them. Part of the reason I'm so motivated to move forward with Xlight.
And if you think about Xlight, like why is it viable now as opposed to a decade ago? Like what has changed?
Yeah. You know, and there were attempts, right, to use a free electron laser, you know, a decade ago, decade and a half ago. ASML even looked at it, right? They tried it. Yeah, yeah, yeah. And essentially, you know, there's been a couple of breakthroughs that Xlight has done, you know, to be able to create independent light sources of different wavelengths, energy levels, you know, in the free electron laser as opposed to pushing that separation of light of a single beam of light, you know, that is then essentially, you know, separated through mirrors and lenses because now you have the whole, you know, thermal load hitting that first mirror, that first point of separation. Xlight has found a breakthrough that allows that to actually occur inside, right, of the free electron laser and the undulators associated with it and stuff. So they, you know, that was the problem that, you know, led ASML say, oh, we can't make this work. And if by the way, if I was, you know, looking at it from their view, right, this is a little thing, you know, if you're looking at it, you know, from I'm getting a laser and now how do I carve it out versus I'm creating a laser that's optimized for this. It's a, you know, the breakthrough that Xlight, you know, has now clearly shown that we know how to make this work. That's what we're off to do with Xlight. And, you know, we made LPP work, right? You know, we're dropping droplets of tin, hitting it with lasers, collecting light, you know, it's just like an engineering marvel when you look at one of these things. I mean, it is really, you know, amazing, you know, but it's only ever, you know, it's a byproduct of the physics of tin. It's only ever going to be 13.5 nanometers. You know, we believe we can take Xlight and free electron lasers to two nanometers of wavelength. You know, it really is pretty magical what we see as possible.
If you go to Taiwan, like 100% of people will know what TSMC is. You could ask like the random cook and they'll know. It's just it's the it's like the key company in Taiwan. Everyone talks about it. I was in Amsterdam recently getting a tour of the Rijksmuseum by a very smart PhD, great tour guide who gave us a tour. And then, you know, at some point there's like a room and it was sponsored by ASML. So, I started telling my kids what ASML is. And the tour guide had never heard of ASML. She's Dutch. She had never heard of it. And I think it just it hasn't pervaded in it's it's not at the same level in Holland as TSMC is in Taiwan. Like, why has it not captured the imagination of the population there?
Yeah. Well, one, I find that surprising, but you know, remember, you know, TSMC, Taiwan, right, the whole situation with the Taiwan Straits, China, right, you know, everybody in Taiwan knows what the Silicon Shield is.
Yeah, good point. Okay, it's in the news every day. Yeah. No one's invading the Netherlands right now probably.
Yeah, you know, so this is front of mind. And the portion of the economy that TSMC represents in the little island of Taiwan is a larger percentage of the total economy than ASML is in the Netherlands. So you have, I'll say, you know, the geopolitical importance, of course, you know, everybody knows, right, what the Taiwan straits are and the situation is, you know, every single, you know, they probably, you know, like at two years old they're already learning that, you know, so like that. So I think that is just such, right, you know, a central aspect of the entire economy, society, you know, of Taiwan that, you know, to me it's unprecedented to think that a little, you know, Taiwan is what, you know, 26 million people, right? You know, it's not that big of a population. And I'll just say, you know, give you a little, you know, I was shocked by this fact that US now imports more from Taiwan than we do from China.
Yes. What? You know, this to me, you know, this is that's insane. If that if that is if that's true, if if it's even like 20% of China, that seems crazy to me. Like you're saying, well, you're saying more than 100% of China.
You know, and just to give you the numbers, right? You can go do some research on this yourself, right? That because of tariffs, because of this administration, some of the actions they've taken, import from China has decayed from, you know, 500 plus billion to less than 300 billion. And in that same period of time, the emphasis on supply chains, AI chips, the acceleration of TSMC's import, you know, we have gone over the last decade from 20 plus billion to over 300 billion from Taiwan. Taiwan is now a bigger importer to the US than China is as of the last quarter.
Stunning. Holy mackerel. That is crazy. I had I I was not aware of that actually. Yeah, that's that's stunning right at that point.
Now, of course, you could say, hey, some of that, you know, it's stuff that was done in China that's now being funneled through Taiwan. You know, I'm sure there's some nonsense. I'm sure there's some nonsense in those numbers and how it's reported, you know, given some of the, you know, tariff and other things like that. But even if it's close, that is such a dramatic geopolitical trade statement, right? That we are now more dependent on Taiwan as an importer than China. Wow. Right. And it's also a statement, you know, explosion of AI chips, the explosion of wafers and so on like that. That just shows.
And by the way, I am literally fact-checking you as you're talking using a few AIs. And you're at least according to these AIs, they agree with you. Like I am completely shocked. Like I had no idea that this was true.
Everybody needs to see the Summation show because they don't realize how current, important, and factual we are together. Orin, isn't this great?
Yeah, this is crazy. I just this alone is like the biggest stat that like I'm going to like start telling everyone. I had no idea. It's interesting because I've never heard this said before. So I would have thought that this would have been more in the news. Like that is that's an insane stat. So does that basically mean that outside of like Canada and Mexico it's like the number one country we import from?
You know, I think it now, you know, sits in the top five.
Yeah. Wow. Wow. That's yeah.
Yeah. In a nation of 20, you know, again, this is less than 30 million people, a small island.
So, it really shows you how important TSMC is like that cuz a lot of that is probably TSMC, right?
Now, let me give you a next factoid, right? That, you know, our minds were just blown by realizing we import more from Taiwan than China. Wow. Now, the next factoid is Taiwan runs on imported LNG. The reserves of Taiwan are less than 3 weeks. So a blockade of the Taiwan straits for more than 3 weeks is out the island. If a fab goes down, right, it takes months to restart it, right? You know, very long supply chain. So literally, you know, we've had multiple blockades over the Taiwan Straits over the last 10 years, right? The longest of which, if my memory is correct, is 13 days, right? So essentially more than 3 weeks and you become, right, entirely just, you know, undermining the semiconductor supply chains. A study by BCG last year said, right, you know, if the supply chain of Taiwan to the semiconductor industry goes offline, right, it has twice the economic impact of the Great Depression.
And why don't they stock 6 weeks? Like why it's not that hard? Like why have they only done that?
Hey, you know, I mean, it's like just in time supply, right? And you have to build new storage tanks, all those kind of things, right? To me, you know, I'd say if you want to have a resilient Taiwan, the best thing that you could do is build more energy resilience. You know, politically they were building a new nuclear reactor, right? They stopped it. It's essentially built, you know, not activated. You know, to me, I'm more interested in them, you know, buying energy infrastructure and stores than them buying defense capacity.
Yeah. Interesting. So, anyway, those are your two factoids of the day, Orin. So, that's super interesting. You're also like you're investing in deep tech and real deep tech, not just like, you know, someone writing some software and stuff. Like how do you when you're doing that, how do you underwrite the technology risk? Like how do you actually make that investment?
I joined Playground a year ago. Part of the reason I joined is because it's, you know, it's investing, you know, we're venture capitalists, we're out to make a return, you know, give a good payday to our limited partners, you know, but fundamentally all of our investment people at Playground are deep tech folk, you know, we're PhDs, right? We're professors. Most of the venture capital firms have forgotten how to do hard, right? They're over here doing the next AI-driven dating app, right? And I'm sure that's important, but, you know, hey, we do hard.
The thing is like if you're putting in hundreds of millions of dollars in like more traditional venture capital, like you're doing it at a point you already know it's working and now you're fueling the fire. In your case, like you, how do you know it's the right bet? Like where, you know, how is this different? Signals or it's the science or you really have to dive into, you know, understand the science of something.
Yeah, we are going deep on the science.
It's almost more like you're almost more like biotech or something like you have to go super deep, right?
Yeah. You know, we have energy projects, we have semiconductor projects, you know, quantum computing, you know, new thermal, you know, superconducting, you know, those are the things and biotech, right? You know, robotics efforts as well. You know, those are the areas that we're in, you know, because we think that our differentiator is given that we're a bunch of geeks making investments, you know, that we're better able to assess the technology risk than we are the market risk.
You know, if I create a superconducting, right, you know, capability that's a high, you know, that's going to, it's like biotech, you know, like this drug would be worth a gazillion dollars if it worked. The hard part is just getting it to work, right?
Yeah. Right. And we can make those judgments better than other venture firms, right? You know, because we go deep in the diligence early in that light, you know, that's hey, you know, if we make this quantum effect work, right, in photonics, okay, we can do a lot of them. You know, if we can make superconducting work, we can do.
You're not taking any market risk at all. You're taking 100% on the tech risk, right?
Yeah, that's what we do, right? And by the way, you know, I've been a deep tech guy for my entire career, you know. So, at this phase of my career, Orin, it's sort of like, yeah, this fits me. Yeah. And, you know, I get to leverage also the unique network that I've built. Imagine if we're creating a new power delivery technology. Everybody who consumes power delivery is in my Rolodex, if we know what that is. Yeah, I know those folks. We can accelerate, you know, and, you know, benefit from, you know, my decades of experience in the industry and bring unique and powerful ways.
Now, if I had to rank order the best managers of all time, ever in history, I would almost certainly put Andy Grove in the top five and maybe even the top one. You actually worked with him. What did you learn from him?
Well, you know, and just maybe a fun story for your listeners here. So, I am in charge of the 386 tape out. Okay, so this is 1980, right? You know, or 1983, we're assembling the 386, right? You know, putting the whole database together and so on. I'm the young engineer who's doing that work, right? You know, getting the database assembly and I'm giving an update to the executive staff of Intel. Gordon Moore, Robert Noyce, Andy Grove.
