What's Real And What's Fake In Tech - EP 56 Peter Barrett

I think we're on a period of incredibly rapid change and there are exciting times ahead, but there is, you know, reason for caution.

Our guest today is indeed Peter Barrett. This is a, a special episode for me, at least, in that Peter is a general partner at Playground Global, which is a venture capital firm that was investing in hard tech before it was cool and has been kind enough to let Core Memory come to life inside of its, its confines. And so I, I I don't want to get too emotional and sappy, but I will say a very, very heartfelt and deep thank you to everyone at Playground who's let us set up the studio in here.

Of course, delighted to have you here.

As usual, you guys are chasing and supportive of people doing interesting things in the deep tech world. So this was a great home for us. So very happy. I'm going to do a quick background on you and then I just wanted, I wanted to describe what is outside our doors to people a little bit, but so. Peter and I have known each other for a long time, although this podcast, we've never had like a proper get-to-know-you conversation. We're always nerding out on things. And so this is a nice excuse for me to dive into a couple stories about you that I've always been curious about. But as people will find out from your accent, you're Australian. You born in Sydney, or at least grew up in Sydney. Got to Silicon Valley relatively early by way of your coding expertise, became a internet multimedia kind of guru of sorts, doing web TV, doing early video codecs for, for the Mac, um, and, and then becoming a distinguished engineer at Microsoft, and then, um, doing some startups along the way, which we're going to talk about, and then, and then becoming, uh, investor.

Yes, all of that's true.

Okay. And, and for people, so people know, Peter is interested in many things and goes deep on them. You do not have like a deep formal education. I don't think— I think you dropped outta college quite early.

Amateur enthusiast.

Yeah.

Like one year at Sydney University and do a lot of reading.

But can talk physics or, or gene editing with the best of them. And so, okay, that's the intro. We're gonna go all kinds of interesting directions today, I think. Why don't we— I don't want to do like a playground ad, but this is a very unique building. We're in hallowed ground in some ways. This used to be a Fry's. It was the Fry's in Palo Alto. For people who don't know about Fry's, go Google it. It was the best electronics store imaginable. I'm sure Steve Jobs was walking around these hallways at some point buying bits and pieces alongside Woz. Um, so yeah, but you guys took this building over and have— I mean, just you describe what is outside our doors to people.

Sure. So in a building that before it was Fry's was a cannery, right? When Silicon Valley made apricots instead of apples, uh, going back to the '20s. But, uh, Playground's an unusual deep tech venture firm, right? So we have, uh, a building here in Palo Alto with, um, about 380 people in it, uh, across about 35 companies, uh, which some of which are portfolio companies, some of which are companies that have interesting adjacencies. But within these walls, there are scientists and engineers who are creating absolutely novel results in their domains. And then we have a venture firm full of nerds, right? That all the GPs here at Playground are engineers. We've all built technologies. We understand the mechanism of the things we invest in. And we've got a team of scientists and engineers, you know, PhDs in, you know, bioengineering, nuclear engineering, AI, material science. And they work with our portfolio companies to develop technologies that almost certainly won't work, but will be consequential if they do. And as an engineer, here, it is a very fun day at work.

Yeah, yeah. And, and you guys have— you've got full-on massive hardware lab with CNC machines and laser cutters. There's an optics lab, an electronics lab, and, and the startups can hop in there and prototype and, and do things.

I mean, we, we've got, um, lots of different ways to collaborate with a portfolio on their technology. And so working side by side with them on building fixtures for quantum computers or using lasers to do novel things. I guarantee that this venture firm had more liquid helium in it than any other venture firm. And there are some stains on the floors from PsiQuantum's early days when they were here from 3 people through to about 90 people. But I, it's pretty unique in that the energy that you get from the portfolios can be turned back into roadmaps that allow us to invest in things that we understand, right? And We can look over the shoulders of these great scientists and engineers that are producing these novel results and sort of get a preview of what the future is going to look like. Uh, and that allows us to underwrite wild things like quantum computers and superconducting logic and, um, quantum sensing and various other things that would be impossible if we didn't have that context.

Yeah. Okay. We're going to get into a few of those in a second. When I, when at first some Peter background, when I, when I, one of these stories I knew well, because when I wrote my book on Elon, I'd talked to you about it, which is how we first met. However, if you, if you research yourself on the internet, there are, there's, there's 2 or 3 stories that come up all the time. I found some of the descriptions unsatisfying. It was the same sentence or 2. So I was gonna, we're gonna, for the record, we're gonna deep dive on a couple of these. One is that you're 19 years old, you write some kind of software application, you're in Sydney. I think you're still in Sydney at this point. And the NSA makes contact with you. This is, this is largely the thrust of what the internet says. I want, I want to know more about this.

All right.

What do we do at Core Memory? We cover innovative, fast-moving, forward-thinking companies, which is why Core Memory is sponsored by Brex, because Brex is the intelligent finance platform for many of these companies, 30,000 companies from startups to the world's largest corporations. Rely on Brex's technology for their finances. They've got smart corporate cards, high-yield business banking, and expense automation tools that are fantastic. I hate doing my expenses, and Brex's AIs and software run right through those expenses, figure out where we're spending money, and take care of so much stuff for you so you don't have to waste your time on it yourself. Go to brex.com/corememory to learn more and just, you know, get with the program, let's get going, let's get out of this archaic finance software and move toward the future, core memory and BREX.

Yeah, so I'd actually written while I was still in Sydney, an encryption program that used both the data encryption standard of the time and a couple of algorithms that I had developed that were faster, right? That sort of more like AES. Similar structure to what is now the streaming encryption standard. So wrote it while I was in Australia. When I moved to the US, the company I joined was marketing and selling that software, and it was showing up in, you know, in lands foreign, but it was a munition. It was because it implemented the data encryption standard, it was a munition. And so the NSA show up, And I'm 19 and there's, you know, a bunch of folks show up to the, to MAC, which where I was working, and they have reverse engineered my software and had a presentation on how the algorithm works and what changes I would need to make to the algorithm to make it exportable, right? And this instilled in me a lifelong appreciation for the sophistication of our intelligence communities, but was, it was kind of, uh, it focused my attention.

Do they make an appointment before they show up, or do they just appear?

No, no, no. So they would, they, they would apply. Uh, but just the amount of, uh, dedication and sophistication to be able to do that work and then have an opinion about how the algorithm needed to be degraded in order to make it safe to export, uh, was, was kind of remarkable. And it— it informs, you know, it's really, really hard to stop the other guys catching up, right? So the cryptographic knowledge was leaking out of the US, not in the form of, you know, naive algorithms that I would write, but in the form of books and various other technologies. And so since then, I think we've taken a different view as to how we stay ahead in cryptography and how we stay ahead in quantum. And, but he did underscore that there are a lot of very accomplished people out in the world who are spending a lot of time and energy keeping us, trying to keep us safe and really appreciate the sophistication of those efforts.

How were you doing that at such a young age? What did your parents do? How did you get into computing?

Yeah, so my dad worked for the New South Wales Health Department. I grew up with a lot of, so he was, you know, building their systems. So I had a lot of punch cards lying around when I was a kid. Early interest in computers and electronics, principally through software piracy, was kind of how I got into it. But have been, you know, writing code now for, what, 45 years, which is distressing, a distressingly long time. And just fell in love with, you know, having agency, getting computers to do something for you. And still write code for the original machines, right? I still write 6502 code,, for fun. Um, and, uh, just, you know, as a kid, just spent an awful lot of time reading about computers, reading about hardware and, and constructing things and made my own circuit boards and loved, just loved, uh, summoning some utility out of those parts and that software.

And you, I mean, you come to Silicon Valley so early. If I remember right from my reading, it was a trade show or something that gave you a taste of the Valley for the first time.

Yeah. Came, sort of came to Macworld. And had started a company in Australia sort of reselling Mac stuff and met a company that was only a handful of people that were making hard drives. I eventually rewrote the software for those hard drives. And then the next year when I came into Macworld, I never went home.

Okay.

And stayed with that company until it went public. Have built a number of other, you know, other startups, but got to work with Steve Blank, right? Who was a VP of marketing at the first company I joined. As a mentor. And that was a fascinating, that was a fascinating journey.

Everyone should put in their notes to go look at his Secret History of Silicon Valley video, Steve Blake. It's great.

Yeah. And then, you know, met my partner Bruce Leak, who's here at Playground in the '80s, working first on TrueColor QuickDraw and Color QuickDraw and then QuickTime with the video codecs and so on. So yeah, just fell, fell into lots of the things that made Silicon Valley exciting in the, in the '80s.

Blake, when you drop out of school and come here, year at 19, your parents, I mean, did they have thoughts on this?

Um, yeah.

I mean, I, it wasn't like the, the, the whole dropout tech story was not, that wasn't a quite a thing.

Well, it's, it's very common for Australians to take a gap year.

Yeah.

And either right after school or just sort of just not show up for college for a year. Um, and you know, normally they show up confidently as ski coaches in Vermont.

I was gonna say, I'm usually a bartender in Greece.

Yeah. And they don't like no domain expertise, but it was like, You know, there's this sort of license to go do that. Um, but I just never, I just never came home.

Okay. Okay. And you, you, you kind of get immersed in this, like you said, the Mac world and, and doing multimedia, pushing video to its limits on, on these early computers. And, and then, um, Rocket Science Games was before Microsoft or after?

Uh, that was, that was before Microsoft. Okay.

So you guys, you guys end up at this, this company. I mean, now in retrospect, it was which was full of all-stars. You were doing incredibly ambitious things. I mean, my sense has always been, I mean, there's this Wired cover story. This was a big deal. Um, my sense is that you guys were pushing video games to their limits at, at the time.

Yeah. So there was a sense that, you know, there was a intersection between storytelling and the new capability of actually being able to mix. Interactive video and traditional gaming mechanisms into a new medium. That was the CD-ROM era. And all of a sudden you could put a couple of hours of video and be able to structure these, these stories. And so we had amazing designers, amazing storytellers. Many of them have gone on to produce some of the most recognizable art direction and, and creatures, um, that we all know and love. Um, One of them I think is an absolute unsung hero, a guy by the name of Ron Cobb, who designed the DeLorean from Back to the Future. He designed many of the creatures in the Star Wars scenes. He was art director on Conan the Barbarian, on Alien, and watching interactions between, you know, Ron Cobb and Elon, I think had a lot of influence on how he thinks about moon bases and rockets and things. But there was just many, many people of different disciplines, of engineers, of artists, of of people who were just curious about the storytelling evolving. And we made some very beautiful but not very fun games. But some of them, you know, I think we had a couple of things that really touched people's lives and, you know, became notable. But we did spend a bunch of money and dug a crater for our investors. But, you know, Tony Fadell hung out there a lot and did some work with us. A lot of Academy Awards and a lot of recognition for the visual storytelling that we were trying to capture. But a creative success, but a commercial failure.

All right. Well, you did, you fast forwarded to one bit of that because you brought up Elon and I apologize because I know you've told this story to me and I know you've had to tell it to some other people and I know he wasn't there that long, but it is kind of a treat. I mean, that was his first He'd had jobs in Canada, like, you know, working in furnaces and things like that. But this was his first tech job, as I recall, and definitely his first Silicon Valley internship. And like, sort of, you couldn't plan this a different way. I mean, he's at Rocket Science Games doing this internship. So he's what, he's like 18, 19, and he's in a summer away from Queens or Penn. And is popping by. Yeah.

And he's really bright and he's working at a supercapacitor thing during the day and rocket science at night. But very good, you know, knew how to write code, very humble, very smart, very curious. You know, fond memories of interaction with him and Ron Cobb, who had designed this Mars mission as part of a joint Soviet-American, they wanted to create a movie that described a Mars mission.

Like this wasn't for a game. This was—

No, no. So this was work that Ron had done with NASA prior to the collapse of the Berlin Wall. And there, if you go and look at some of those drawings and some of the scenarios, they resonate with Starship and some of the Mars ambitions. And Anyway.

Wait, wait. So I don't think you told me this story when I was doing the book. So Ron, he was laying out some kind of future that—

He was doing design for this ambition of building a, creating a movie about a joint US-Soviet Mars mission to capture public imagination of the effort to be able to draw funds to make such a mission a reality.

Yeah. Well, that has a, That Elon's— before SpaceX was SpaceX, he really wanted to do almost a publicity stunt on Mars with this greenhouse growing a— yeah, I mean, yeah, you can see some seeds, seeds of all this. Yeah. And, and anything you— I'm sure you guys all sat around and you knew he would, he would become the richest person in the world.

Yeah, it was. Well, it was clear he was going to be an entrepreneur, right? It was clear that he was going to start a company. And I introduced him to the guys who funded Zip2 and, and I've lost track of him. I don't know what he's up to these days, but he seemed, he seemed to have a lot of potential. But, you know, bright, sweet kid at that point. Yeah.

I'll fill you in. The story's got some good and some bad, right? Quite dramatic on both ends.

Yeah. No, we, look, I think all technology has the ability to do great good and grave harm. And I think Elon might be the physical embodiment of innovation. In that regard.

Yeah.

But, you know, he is an extraordinary figure. And that, you know, every time you write him off, he snatches a 23-story building out of the sky with chopsticks, right?

So I wasn't covering tech during Rocket Science's heyday, but from everything I read, it just reminded me of you guys. It was the hot shit of the moment. And I think you raised $35 million or thereabouts. Okay. So, and how old are you when it goes tits up?

Uh, yeah, mid-20s, late. Yeah.

So like mid-20s, you've got the coolest company in Silicon Valley is just gone. You do cool stuff, but it's, it's gone under. Yep. And that doesn't, it seems to have no effect on your, you know, reluctance to keep pursuing things. I, I, yeah, it would just feel like that might be a, I know this is, this is sort of normal for Silicon Valley in some ways, but yeah, for a 25-year-old kid to kind of go through that, it, it, it didn't make you reticent or reluctant? Nope. No.

And because we leapt right back into, you know, many of the folks who were at Rocket Science, a lot of them went to Industrial Light Magic, right, to LucasArts, to, uh, and then the technical folks, right, went off and spun up WebTV, right? And I just hung out and wrote the original browser for them. And eventually, as that grew into something rather extraordinary, right, and rolled into Microsoft, found myself back in that fairly quickly.

Yeah.

The look, the, yeah, you cry yourself to sleep a couple of nights, but the whole point of this part of the world, right, is that it's hard. And unless you are, you know, it's a power law, whether it's investing in venture or whether it's personally being in a startup. If you're not failing most of the time, you're not trying hard enough, right? And power laws apply to human endeavors. They apply to hard technology and they're the substrate of venture.

You end up at Microsoft for 7 or 8 years. 13 years. All right. I undercounted. When you're in, like, curious how we feel about WebTV in hindsight. I remember at the time it was kind of like amazing when you would go to a hotel or something and this would pop up and it was quite different to anything else that was out there. The internet being what it was and the interfaces, it was always a pain in the ass to type the URL in with some remote control that you had. I, now when I think about it, it was just seems like very advanced in many, many, many ways compared to what other people were doing. Is, how do you feel about it? Is this like a lost opportunity or?

Look, I think So the amazing thing about WebTV, right, when you think about the, the ambition of what they actually achieved, right, and this is Bruce and Steve Perlman and Phil Goldman, um, they could make something, that thing had 2 megabytes of memory, right? And that included the memory you needed to use for the screen. Everything was folded in half and folded in half and folded in half, right? They had multiple OEMs building the thing. It worked. And that service ran for many, many years. And it was really, really well executed in a very, very short amount of time. An amazing group, really amazing group of people. But it was the progenitor of what Danger did with the Sidekick, right? So Danger is really sort of a mobile replaying of the WebTV service and integrated beautiful hardware with beautiful user interfaces. And if you look at what Danger was, that was the progenitor of iPhone and iTunes and everything else. So I think you can, you know, go back to those roots of really, really good design, really, really efficient engineering, scrappiness, and all of the things that Silicon Valley is really, really good at creating and short timeframes of relatively small amounts of money. Uh, and so yeah, it was great to be, um, at the origin of that, but also see the diaspora that's come out of WebTV and come out of those, come out of Danger and come out of, um, that initial work was really exciting.

Is that what you primarily worked on at Microsoft?

Well, so Microsoft, we did the, it was very TV focused, right? So TV division built the first IPTV system that was used in, I don't know, probably about 40 countries at the limit. Um, But a lot on kind of how television evolves and is evolving. And there are some really interesting lessons in what happened then and what's happening now as to what the agentic web is gonna wind up looking like. But got to spend a lot of time with Bill and Steve that I admire greatly. And they were very, you know, they were very good to me in that process and got to be around a bunch of really, really smart people. But it's really comfortable to have a, you know, a well-paying job in a big organization like that. And it just felt like I was not— eventually was— had an itch to, to do something else.

Yeah. And the— it is funny, there, there's a— you don't have to work too hard to get to a universe where Microsoft is— has Netflix, or— yeah, it was always amazing to me that they didn't end up owning the living room more, um, especially between WebTV and the Xbox and even things like Kinect, which I, I remember doing the first story on that. Yeah. That actually felt really good. And I was like, oh my God, they're, they're doing really fascinating things. And then I used to go, so they used to invite journalists once a year for a day or two. They would invite like 12 of us and, um, You know, you'd be with Ballmer or whoever was in charge at the time and they would take you into this, um, fake living room or, yeah, it was like their demo room and they would show you the future. And I have to say, I mean, I was always quite— both Microsoft and Intel, right as we're heading into the era of the iPhone and things are about to change, they like they kind of got everywhere first and they had pretty good products and they were so close to nailing all these things that they knew exactly where the future was going to be and they just never quite stuck the execution for the next big thing.

Yeah, they missed mobile. They, they missed mobile. But one, one thing they did do, like the hard pivot into Explorer was pretty dramatic, right? That go back to your desks, erase your hard drives. There is this thing called the internet. We're going for it. Right. And the willingness to do that. And obviously, you know, Sacha's done an incredible job with AI. So there's, you know, there are some lessons learned, but missing mobile was the big—

That's what I mean though. Even when you have these like pocket PC devices way before the iPhone existed.

Yeah, but they didn't understand the difference between leverage and burden, right? The leverage was, hey, we have all these enterprise things that we can make versions that go onto these mobile devices. So they have to have keyboards, they have to have these characteristics that meet our leverage, but it was just a burden. It just completely misunderstood what mobile was going to be for.

Yeah.

And you know, they were asleep at the switch on that stuff.

Do you have any particularly favorite Ballmer or Gates memory?

Look, there's a funny one on Steve. The day I said I was leaving and he just wanted to check in and see whether I sure wanted to leave or not. And he was so focused and so energetic. He had this giant thing of Coke on the table. Full of ice. And we're having this animated conversation. He crosses his legs and he kicks the thing of ice all across the thing. He's like, don't worry about that. And I'm like the entire time the ice is melting and everything else. But look, the brilliant people cared a lot about doing useful things. Obviously, you know, great admirer of what they've both done in philanthropy sense. But But just smart people that cared about doing useful things.

Did you ever go to his room at the Ruth's Chris Steakhouse?

No, no.

But so I did an interview with him there once. I mean, there's a back room where they must have banquets and things, so it's quite large. And, and, but he wants some privacy and he would interview job candidates and have meetings in, in the steakhouse. And yeah, I was doing a big feature story and it is just so bizarre cuz you end up— in this table in the middle of this large room, and it's just you and Steve, and you have an unusually large number of people catering to your every whim, and they already know what he wants to eat and everything. And then I always thought the caricature that Ballmer played, the public persona, which is like part of his nature, so it is authentic. But I always found it a shame that that's all the public saw, because when I was sitting with him at that table, Um, and then we ended up getting along. So I had subsequent interviews, just like an incredibly smart, thoughtful guy. Um, like his math mind was exceptional. He was rattle off all these numbers and, and way more dynamic and interested in more things than, than what the public got to see. And this was like when Microsoft was struggling and, and I, I, he would just get pilloried and I, I, I, I don't know why it's not really my job to like force people to care or something about this, but I always found it a little sad that he, people, I don't think really get him.

Yeah. No, like both, there is a lot of breadth and depth in Steve, obviously in Bill as well, that people don't get exposure to. And especially now we live in a weird world where people's first, you know, the first bit of their understanding of those folks is so far from the mark. It's a missed opportunity.

Yeah. Okay. So you, so you're there, you're comfortable, and then you decide you're gonna, you're gonna change it up.

Yeah. And there's, you know, lots of different, lots of different ways that I explored mechanism and various other things pretty quickly came to this framing of getting to work across a cross-section of technologies, uh, and a cross-section of really, really interesting, talented people that it sucks being, you know, uh, not having people challenging you or not having smarter people than you around. Um, it's in every day that goes by that that's, that's the case. It's again, a wasted opportunity. And so the whole framing of venture is so much more exciting than doing it one at a time. Yeah. Right. So I could hop between various, you know, projects and attempted retirements and angel investing, but you want to do it at an industrial scale and being surrounded by people that are way smarter than you that are doing the most important stuff. And that, um, you know, a decade ago when we came to form Playground, there was just a recognition that Silicon Valley had got lost in abstractions, right? You know, software may be eating the world, but you can't eat software. Right. We, we live in a physical environment. If you want to build a quantum computer or a cancer drug or superconducting logic or nuclear reactors or whatever it is, you have to involve the physical layer. Uh, and so the combination of returning to venture capital roots and to doing it right as these watershed technologies are emerging, right as AI and quantum and these other technologies are about to go through this phase change, it just seemed like an unmissable opportunity.

Hey, since day one, the Core Memory Podcast has been supported by the fine people at E1 Ventures. They are a young and ambitious VC firm in Silicon Valley investing in young and ambitious companies and people. Thank you so much to E1 Ventures for all your support. This is, this is where we're going to change gears a little bit because I do want to, I want to mine your mind, um, on these different areas because Every time I run into you in the hallway, you can just go so deep on so many different things. Um, there is, that is funny. I mean, that was, I feel like I was writing about hard tech stuff back when you guys were investing and now everybody's doing it. Um, so I lied right before we get into our, our journey across Techland. Do you feel, how does Playground, what is your edge in this world in which deep Tech is suddenly sexy and there's quite a few companies trying to invest in all this stuff.

Yeah. Look, I think our superpower is being able to underwrite technical risk and retire it, right? We'll write the check when somebody comes in spouting nonsense that seems like science fiction, that to the right eyes of the, like the nuclear engineers on the team or the, you know, the bioengineering experts, they can see that this is a testable hypothesis that we can underwrite. And get an answer for a deterministic amount of time and money. You know, when we invested in PsiQuantum, there were 3 people, they couldn't get arrested, right? That there are, there are things that lie between the improbable and the impossible, but the right side of impossible that can be turned into commercial and societal outcomes for sometimes billions of dollars, sometimes tens of millions of dollars. And I, our edge is being able to have a platform where we can know which is which, right? And Silicon Valley is not immune to investing in things that violate, you know, physics. And no amount of entrepreneurial zeal is going to overcome physics. Uh, in enterprise software or other, other kinds of things, you can pivot a brilliant entrepreneur from one technology to another. Deep tech, you have to find the domain experts who can build the world's most powerful lasers or build quantum computers. And guide them through that, that journey of, you know, developing the technology, retiring that technical risk, you know, attracting the eye-watering amounts of capital necessary to turn it into a reality. And 10 years in, we're getting kind of good at it, I think, um, and have made some of the most influential technical bets, um, and feel emboldened to do more of that.

This could be wrong because I've never been a VC and I have no idea what the timeframe is for these types of things, but I could imagine a scenario where when you guys started and, and you're really pushing the edge of these inventors and they've been struggling to get funding, you have a bit more time to do your research and really dig in on what they're doing and, and make sure you understand it. Is it different now that there are more VCs investing in hard tech and you're there's this pressure to maybe just cut a check because it's a bit of a race and you've gotta get in and now you can't do all the due diligence that you would like to do.

Yeah. So I hate hot deals, right? That we are, because we're investing off roadmaps, we're often the first to a new category, right? And there's nobody else around the table. And that, you know, yeah, last fund, we're the largest shareholder in all but one of our companies. And that company, we're about 50/50. Right. So we haven't seen a lot of time pressure. Now there's certainly time pressure in, you know, LLM madness or SMR this and fusion that. And we just not participate in any of that sort of stuff. And occasionally we'll, you know, have a returning CEO who we love to death and we'll overpay for them because they are great and will always return value. But we operate off roadmaps. And because, you know, if you're building a quantum computer inside these walls, you can look over their shoulders and figure out what the Fortune 50 want to do with that quantum computer. So then we bet on quantum algorithms and quantum applications and quantum verticals and quantum sensing and all of the nexus of opportunity that comes from having this stuff built in your backyard and seeing things 5 years ahead of anybody else. And so that's a flywheel that is accelerating now as we fill roadmaps and we see, okay, that's going to be a thing. So we don't need to worry about those, the things around it. We just, we're just looking for one of those. And if we don't find one of those, we'll go make it.

All right. Well, let's do, let's do quantum computing first. I think I did the first story ever on PsiQuantum, maybe I want to say 6 years ago, roughly. And then I just revisited them for a couple of days, um, just a few months ago. And then we put out a video on our YouTube channel, like and subscribe. Uh, and, uh, and so it was a bit of a refresher course. I have a couple data points coming out of that. One was that I, I understood it better. It's starting to feel more real, whether it's PsiQuantum or some of the other startups. You know, when I first started covering this, it was rare to make a qubit. If you could make 2, that was great. It was a massive effort to get to 4 and 8, and now everybody can make qubits on, in some level. PsiQuantum wants to make a million, so quite a bit more than some of the other players. But after I wrote that piece, I saw you and Jensen sparring online. Jensen was at least in the moment skeptical about, about quantum computing. You kind of rebutted him, I think, in MIT Tech Review. Yep. George Hotz pinned me down at a, uh, at a conference. He said, I saw your video. Quantum computing is absolute bullshit. You can't tell me one single thing that you can actually do on this. And then more recently, Scott Aaronson, who's a professor at the University of Texas and an unquestioned expert on quantum computing. He's kind of dedicated part of his life to it, but is also, I would say, skeptical in some ways. He's always been quite conservative. He was writing just a couple of weeks ago. He thinks it's more real lately than he had expected, although even when you read his stuff, it's kind of like, well, in a few years we're going to— I do believe now we're going to see some calculation, you know, it's, it's still a bit reserved. And then you guys are, you're, you're maybe one of the most pro-quantum computing people alive. So yeah, what is, there's all these data points. I feel like obvious progress, still skepticism. The one thing I hear the most, I feel like, is the George Hotz thing, which is, I know you guys can point to like a couple applications you could run on this thing, but really, what are these things actually? We don't need this. This is not that useful.

Yeah. And look, so you've name-checked a bunch of very, very clever people, right? And, you know, I think Jensen, of course, invested in the last round of PsiQuantum.

Right. After we did. Yeah.

And I think that that was really from conversations with, you know, Pete Shadbolt and the PsiQuantum team walking him through not only the quantum machine they're building and why that's interesting, but the degree to which it will achieve things that AI can never achieve, right? And many of our fondest wishes for AI are really quantum wishes, not AI wishes. But also sees, you know, PsiQuantum has built the world's best silicon photonics platform in the world, right? So for AI, for classical applications, you could value that company at twice its current, you know, capitalization based on what you use what you do with their silicon photonics for advanced AI, advanced switching.

Can we explain that for a second? So, I mean, you're talking about using that as an interconnect inside of the data center to transfer information optically between racks of computing systems.

Yeah. And so they are building a photonic quantum computer, which is uniquely scalable because they have built this advanced silicon photonics process, which holds essentially every world record in low loss, in switching fidelity, in switching speed, in single photon detection and sources. And all of that stuff will replace electrons in the data center at scale. And it is so much further than state-of-the-art. You know, the switching coefficient's about 40 times better than what people normally fabricate transceivers. Optical switches out of. Um, so yeah, so I think Jensen, you know, I think Jensen is, is more calibrated now.

Um, but are you kind of saying he's views this as like a data infrastructure, data center infrastructure play as opposed to believing in quantum computing?

I think he believes in both and I don't want to put words in his mouth. He, he can speak for himself, but I think he's realized that first of all, useful systems are very close and, uh, and that there are this exhaust heat of technology, which is also incredibly valuable that can be pressed into service. But what are they for? Like, why do we care? And there is one application that our intelligence community cares about, obviously, which is being able to read the things they've been collecting for 30 years, right? That's a real application. It's an important application. It's important the US win that. But to me, the most interesting ones are stem from the fact that we don't know how to do chemistry, right? That we design very little, that our industries are animated by discoveries, right? The catalysts are drugs. There's a reason why they call it drug discovery because we can't simulate the quantum nature of nature with a classical machine. And the behavior of many prosaic things are beyond our comprehension or understanding. And You know, that example with the LK-99 superconductor a couple of summers ago.

That was fun.

Right? And so we took our best computational models and our smartest chemists, and half of them says it was a superconductor, half of them says it wasn't. We just had— it was no better than chance. Because the highly correlated quantum systems are beyond the capability of a classical machine the size of the moon, right? And no AI is going to help either because we can't simulate it, which means we can't build a training set for it. So AI is not going to solve this understanding of quantum materials and give us design agency over this incredibly fertile set of possibilities. Like, we're building fusion reactors using Type-II superconductors. No working model how they work. No idea, right? The iron nitride magnets, which are a crystal of iron and nitrogen, very common elements, produce a magnetic field far beyond any theory that we understand. Right. And we're stumbling around in the dark. And so if, you know, George might say, well, it's only good for one thing. If it's only good for chemistry, right, that's a trillion-dollar opportunity that changes everything about medicine and industry, right? That's enough. That could do other things, but there are some really interesting, you know, things for exploring physics and exploring plasmas, exploring, you know, beyond material systems. So It's not just chemistry, but if we just solve chemistry, right, to allow us to design things and not discover things and get agency over the physical world, that's enough, right? And we are tantalizingly close to results that first of all, reveal new things about how those systems work, and then ultimately have machines that are big enough to just change that world of discovery to a world of design. And that's a couple of years away, right? This is not a decade journey.

What's the— when you say AI can't get us there, you know, the last month to 6 weeks I've been bombarded by people using AI models in biology to simulate the, or to kind of play through the process of a embryo differentiating into many cells, people designing antibodies. And so they're taking these kind of existing datasets and, and, um, yeah, pressing fast forward on them and complex systems. It's, it's biology. Why would something like chemistry be different to that?

Yeah. And so there are great applications for AI in biology, right? We, we have many examples in the portfolio, right? Atomic AI that built the first RNA structure from sequence and behavior from sequence models. Um, so, but, but there are a class of things about highly correlated systems, right? Like a type 2 superconductor, um, where there's just no foothold to get started, right? We, we don't understand those, the dynamics are, you know, the, the density functional theory tools that we use to do it. Just produce the wrong answer. Like, we can't even, you can't even simulate a silicon wafer, right? The silicon crystal seems like the simplest possible thing on earth. Um, you know, the DFT calculations, ground states are just wrong. So, uh, it's a, it's a specialized kind of, kind of knowledge that's at the root of these things. AI is incredibly useful for science in many, many, many different ways. But the fundamentals of understanding I take 4 prosaic elements, I mix them together in a crystal, and they start doing things that cannot be understood or explained by conventional methods. Um, and, you know, I liken this, you know, when you think about, okay, say we have agency, say we can design these things, right? And now instead of dealing with the elements, I can deal with prosaic things that I can mix together in crystals. The design space is essentially infinite, right? It's like waking up one day with a million new elements. On the periodic table that all do useful things. And we've discovered some of them, you know, perovskites for solar panels, superconductors, iron nitride magnets, right? We have indications of them, but getting your head around what it means to be able to design in that space is a little bit like, you know, imagining a world without metal, right? We could have stone houses and wooden plows. We can have books and poetry and movable type, but we wouldn't have electricity. We wouldn't understand electromagnetism. We'd have no working knowledge of it, right? We, you know, we'd have a culture, but we wouldn't have social media. We wouldn't have AI. We wouldn't have motors or Teslas or any of that sort of stuff. We do live in a world without metal, right? The world of quantum materials is so unfathomable to us that we just can't get our heads around what it will mean to have agency over nature and be able to design those materials and design in that space, as opposed to just stumble across across some tiny percentage of what's possible.

Well, okay. And so if you're trying to understand superconductors on this level, what is the— if we don't— if we're so struggling to approach it, what was the software that we're— how do you even begin to lay that even onto a quantum computer?

So yeah, I mean, you want to, you want to be able to design algorithms that can map the dynamics of the quantum things inside the materials to you know, qubits that are inside the machine, and see how those systems evolve over time, sort of understand the physics of how those, you know, how Cooper pairs are moving through Type I, which is well understood. You know, what's the equivalent in Type II superconductors? And those algorithms exist and those methodologies exist. And we're right at the threshold, even on the crappy quantum computers we have today, we're right at the threshold of being able to simulate those systems beyond classical capability. And I think there will be papers published in the next 6 months that show things about science and physics, which were invisible in classical systems that are visible now in even early quantum machines. Now, there's certainly an entire class of things you need millions of qubits to do, and the SAI guys are doing that. But the algorithms to represent those, uh, those systems, to represent those ground states, to represent those time dynamics. Those algorithms are getting better 1,000 times faster than the hardware is, right? So the hardware is getting big really fast, Psi is 10,000 times the size of state-of-the-art. The algorithms are accelerating, right? It's not this asymptotic, you know, approach towards some minimum capability. Like software is getting better really fast and the hardware is getting better really fast. And those paths cross in the next, arguably they have crossed already for some examples, but this is not something that's going to be dragging out. We get useful machines and we get quantum advantage at kind of breakneck speed. And what's evident is things that we thought were multi-generations of the Sine machine away are now coming into focus as near-term possibilities. And you see that even in the cryptographic application,, right? That you see orders of magnitude improvement in the runtime for, you know, factoring. Um, and that is a continual, you know, that will continue to happen as more and more people have access to these algorithms and these machines.

In the, in the 9 months or so that I've been in this office, I think you and I and most people in Silicon Valley have gone through ups and downs with AI and enthusiasm. I remember running into you one day. Not long after some of the new OpenAI reasoning stuff had come out, and you're like, man, this is changing the way I read all the papers. And, and then a couple months pass and everyone's like, oh, it's all hit a wall. And then we come out of that, and then it's hit a wall again. Where, where are you currently on your AI?

Yeah, so I— look, I think it is an incredibly powerful technology, even in its early nascent form. But we're right at the beginning, right? That we've sort of pushed LLMs into places where they've like, we, we've asked them to do things they really can't, right? And understanding that they're incredibly useful, right? And they can't write code, but if you're writing code, not using one, you're wasting your time, right? That they're really, really, really good at, at, you know, go and do a very, very sophisticated search on a bunch of stuff. And just gather the things that I can be in dialogue about to, you know, come up to speed on things incredibly quickly. Amazingly powerful there. But, you know, they're obviously not, they're a leaf node. They're not on the path to AGI, but are incredibly useful, right? And I, you know, yes, we've got maybe too many undifferentiated foundation models that all kind of behave kind of the same way. But doesn't mean that they aren't gonna have a huge impact on the economy, but we're just scratching the surface.

I feel like the world's so divided on this coding thing. The, was it Anthropic last week? They're like doing so much of our code. They, they would say, you know, it's nice now, it can do pretty low-level work and you don't have to double-check it. And, and, but then, so they were kind of pumping the brakes, but then they're also like, oh, 100%, this is gonna put me out of a job in a, in a year. And then I'll talk to some companies and they're like, oh, you're crazy if you're not using this throughout your whole company. We're replacing all kinds of people. Then I'll talk to others and they're like, no, still total crap. And, and I'm, I don't understand how there's this, this seems like there should be some kind of consensus on the level.

Yeah, it's, I, I mean, the, the, if you wanna bang out boilerplate, if you wanna do unit tests, if you wanna make, you know, they like, I, I still write, when I'm writing code for machines from the '80s, I love having all of the affordances that AI bring you, but they're not, it is not capable of doing what humans do in design and implementation of things without, you know, going off the reservation, right? The experience of you've banged some stuff out, The AI has piled in and added a bunch of stuff and it makes a bug, right? And you're like, you screwed that up there, can you fix that? And they'll fix that one, then a bug will appear somebody else. It's that sort of where you're pushing the springs down on a mattress, right? And then it just starts, you roll off the end of context and starts forgetting things and it behaves, it doesn't behave like a dumb assistant coder that you're training. It behaves like something else. And it's just maddening at that point when you realize I've got, it got me 90% of the way in 10% of the time. And then the other 90% I spent just trying to convince it to please use, enumerate these things the same way that we've talked about 500 times. And it just becomes, I code for pleasure. So the velocity is really exciting, but the minute I am, you know, trying to come up with the right incantations to get the fucking thing just to not reproduce that the same bug over and over again. It's maddening. And so, yeah.

You seem to have thoughts about this agentic AI future.

Yeah.

They sounded possibly dystopian. I don't know where you were going with that.

Yeah. So, well, it's interesting, right? So agents aren't humans, right? And it's amazing to see MCP and these other things evolve without any consideration of attestation or security or any of those kinds of things. So that's gonna be interesting. But look at the dynamics of what's happened to broadcast television that I spent so much time in, right? The average demographic of people watching broadcast, they're people in their 50s and 60s, median age, right? And programming is reflecting that and all of the flight of any of the interesting advertising demographic has left for digital. And we've got Netflix, right? So television is great and broadcast was the technology that you had in hand. It's dead. And now we have a different form of much higher quality subscription-based media. And then we have this digital thing over in the, over in mobile and the web.

My children have no idea that broadcast television exists.

Well, the idea that we would watch something when it was on.

Yeah.

Right. So I built the first DVR way back when it started, you know, pre-Tivo, pre— what was the other one? Replay TV or whatever it was. Um, yeah. And like, I haven't watched an ad since on television since last century.

Yeah.

Right. And. And the idea that you would sit, gather as a civilization and all watch, 30% of you would watch the same thing on a time. Yeah, it's completely alien. Now we've had this web thing that's advertising-based and click-throughs and everything else. And now we're in a world where agents are doing the searching. There's zero clicks on any of that stuff, and that agents aren't subject to vagaries of brand or of influencers or this, that, and the other thing. And more and more, we are going to be delegating what we want and what we're looking for to the agents. And they're going to be swayed by very, very different things, right, than humans are. And it's going to evolve insanely fast, both in terms of malfeasance and hacking and, you know, all the dystopian things. But I don't think banner ads make it.

So like, how do you get an agent's attention? Yeah.

How do you get an agent's attention? And what's it indexing off? And I think it is concision will be one of the primary metrics, right? Concision and results. And, you know, but how does that, how do you get paid for that? How do you measure that success? How do you curry favor with the agents? How do you attract the agent's attention, right? And there's been a couple of papers on this of, you know, we're now going from this, you know, attention framework where you are the product and we are just actively wasting human time to, you know, get enough impressions in front of you to something which is weirdly about how do I pull the wool over agenti guys to get them to buy my product or service. I think it's going to be a wild year next year to see how these inputs change companies that are based around, fundamentally around advertising models, which—

So it's like AIs trying to woo each other with— Yeah.

And so what are they talking about? What language are they wooing it in? Like what's the ontologies that you woo? How do you target? How do you buy those wooings? Yeah. Right. I get the sense it's going to move really fast. Yeah, Microsoft not having a huge advertising base, you know, maybe isolated from some of this, but what happens to social media? What happens to, you know, what happens to Google? Yeah. And they've done an incredible job, right? Jeff Dean and team have done an amazing, amazing job in how they've embraced that. But I'm curious to know when they're having these kinds of conversations where they think it's going to go.

It looks like— that's fascinating. Um, it looks like both OpenAI and Anthropic will probably try to go public. I have long been curious how the two of them remain independent in this, this very expensive world that they play in. What do you see for their future? It's really hard for me to do the math and figure out how they are not subsumed by one of the existing giants. Yeah.

So probably above my pay grade. I think OpenAI, look, that if all of these companies are copies of the same, fundamentally the same thing that have high revenue, high costs, at sort of no significant margins. There will obviously be consolidation at some point. I think we are post-scaling very focused on research, and I think these companies win or lose based on how they are progressing the state of the art and what degrees of freedom exist to make these models do better things. But I think there's too much money into too many undifferentiated Undifferentiated foundation models that at least at pre-training stage all do the same thing, right? All the same inputs, all the same, roughly the same architectures. You're getting some differentiated in the RL post-training stuff, but yeah, there's too, too much money in something that is high cost, low margin. And so yeah, there'll be consolidation. But Google, they're hard to beat.

Seems strong. Yeah., you know, seem like they lost, lost the plot for a moment and have recovered and have endless reserves unless somebody can—

yeah. And it's always amazing when a company that size can move on a dime and rethink this. Now it may turn out that collapses the advertising model, but, um, we'll see.

As you know, I'm obsessed with hardware and manufacturing. I have a slight patriotic streak to me in that I'm encouraged that we're trying to make stuff in the US again. So I generally support these things. I like to cover them. I have to confess, there is like a part of me that now feels this has gotten overhyped. It's, it's, there's a level of BS to some of this. I wonder if this is sustainable. I look at the last 20 years, SpaceX and Tesla, even if you hate Elon, stand out as the unquestioned manufacturing giant stories. It's hard for me to pinpoint another potential one of those, like maybe Anduril, but there's a things that need to be done. Um, yeah, how excited should I be when I hear people talking about the reindustrialization movement? Clearly, I think the US, if it really wants to participate in robotics and the future more broadly, has to do something to be more competitive against China.

Yeah, I think that's right. I think we need globally resilient supply chains, and I think America needs to lead in things advanced semiconductors and things like quantum, you know, there are, there are degrees of freedom there that, with things like, you know, we have this company X-Light that recently got an LOI from the Department of Commerce that can double the productivity of semiconductor fabs by using the world's most powerful lasers to create semiconductors at unprecedented rates, but also continue to push Moore's Law and make smaller and smaller features, right? And it's,, national lab technology, right? And national labs are a national treasure that we are, we have Aladdin's caves full of what the reindustrialization could look like. Uh, and we need to do better at translating those out into, into commercial, uh, framings. Uh, but yes, advanced semiconductors and what comes after advanced semiconductors, right? That we know that we're, we're not at the end of Moore's Law. Right? We are, we are right in the opening innings for how we use matter to do computation, not just quantum, but classical as well. And that it's much easier to lead than prevent the other guy from catching up, right? China is being very efficient in deploying, uh, you know, public funds, uh, nation-state funds into innovation. We need to get better at that, but we have the best scientists and engineers on the planet. And we have these underutilized resources and we have to, you know, I think we need to continue to lead or at least restart that effort to lead in things that will animate civilization moving forward. Right. And so a lot of our efforts are in around building resilient supply chains in, you know, in the UK, in Australia, in Europe, in the US, and building technologies which are fundamentally useful in commerce., but just are naturally strategic because, you know, quantum computers are important for doing things and semiconductors are important for doing things. And, and, um, you know, we don't do anything, any, any purely defense-related stuff, but everything we do in the portfolio is, has strategic utility, I guess.

This is so fun for me. I get to go to all these different areas and, and, uh, robotics, especially humanoid robotics. I think, as you know, I've gotten into these robot fights and, uh, tons of money going into these models for these, these robotic systems and a number of humanoid robotics companies and robotics companies of different stripes. You guys have invested in this area historically. What is— I am again having trouble believing this one's for sure real on any kind of short-ish timeframe, which I'd put it like under 5 years.

Yeah. And so look, the minute you put a human form factor on a robot, you're projecting a lot onto that. It's, it's a human like me. It thinks like me, has dexterity like me, has perception like me. And none of those things are true. Uh, and that the more overly anthropomorphic the robot is, the less likely it is to provide utility or ROI or safety cases or any of that kind of stuff.

Stuff.

And that I love humanoid robots. Agility Robotics is, you know, we met those guys when they were a handful of people. They're the only robot that actually shows up to work in scale every day. It doesn't look much like a human because humans are not evolved, right? Turns out if you have roughly the same number of thumbs as fingers and you have a lot of them, you can do things that humans can't do, right? And the idea, so Imagine you're on the floor of Boeing, right? And there's 40,000 carts with stuff on them. Are we going to replace that with 120,000 humans carrying boxes? That doesn't make any sense, right? So are we going to make, for a forklift, are we just going to have a giant humanoid, right? So humanoids don't generalize. They can't. The claim is that you can just build a human form factor and can do anything, right? We still need a forklift. You still need, to, you know, humans can't, humanoid robots can't fly and they can't go underwater, right? They don't generalize in that way. And the ADA guarantee is that if you move around a factory on wheels, then you can still, you could still do useful work. So then the argument goes to, well, how's my grandma going to find her contact lens? And no humanoid has anywhere near the dexterity, safety, or perception to lift up a couch and go fish grandma's you know, contact lens out and put it back in her eye. That's—

God forbid.

It's inevitable, right? But not imminent. And I think it's a human projection onto these things that infer capabilities they simply don't have. And there's all this other interesting stuff going on with automation, right? Where, you know, Rodney Brooks's company, right? Rodney's the most sophisticated uh, you know, the most storied roboticist that survived.

Does he have a new company?

Oh, he has a company. Well, Robust AI.

Right.

So, but when DHL deploy those robots, you know, they nearly double the productivity of the workforce. The workforce love them because they see them as an extension of themselves. And, you know, the robots are easy to use. They're, they're safe. Um, and they have none of the complexity or none of the limitations that humanoids do. Um, but you know, there'll be agility robots, there'll be robust robots, there'll be things that don't look like robots. There'll be other forms of automation and they work in concert to provide, you know, create abundance, which is great. But we, it's another example of whether it's metaverse or NFTs or overly shiny metal puppets that we fixate on, um, that we tend to go an inch deep and then not have the deeper conversation about, okay, is this really useful? Like fusion, there was like, why, why are we really doing these things?

Right. Okay. Yes. I want to, okay. Before I go to kind of your, you know, the bullshit ideas, we will, before we started chatting, we were talking about rapid progress coming, the world, world changing. Do you have scenario? Do you have a takeoff scenario or this is different to that? This is just, you think the world is about to get topsy-turvy very quickly.

Well, I think the world, we're obviously living in challenging times, right? The rise of tyranny, the people demonstrating that you don't need democracy to be very productive, right? To create wealth and create abundance. Democracy is one way of doing it. It's not the only way of doing it. I do think we are at the threshold of really extraordinary things that can make, you know, reduce these tensions and provide a framework where we'll accelerate into much more comfortable circumstances for all. An advent of, you know, quantum computing, AI, that new kinds of semiconductors, right? We're just scratching the surface in all of them.

In what, 5 years, 10 years, 15 years?

In the next, I mean, it's really the next 5 years, right?

Right.

So there are, there are so many things coming into place and we see it in the funnel for the fund, but, um, we will just have an incredibly new set of tools, right? And we're not in this Malthusian zero-sum game where we have to fight over it, right? Like doubling the productivity of photosynthesis or are, you know, being able to design things instead of discover them, that this incredible explosion in the capabilities of quantum algorithms, quantum sensing, right, where we can make MRI machines 100,000 times more sensitive and see cancer metabolism for the first time, that I think we've got good answers to climate change. I think we have good answers to building resilient supply chains around the world. And there's plenty of capital, right? I mean, there's the amount of money that is pent. And we see this, I think, being wasted largely in some of the LLM investments. But it's not like we don't have the money to solve these challenges. And we've got tools of unprecedented power that are right at our doorstep. So I think we're entering a period of very positive, very positive change if we lean into science and engineering and maintain that interest.

Is the quantum sensing stuff, is that stealth or is that public?

No, it's public. So there's a company that is, they're shipping preclinical hardware and It turns out when cancer eats glucose or a metabolite, it eats it differently than normal cells. And you can use these quantum contrast agents, which are really just take the sugar, align the spins of the nuclei, and those aligned spins are very visible to ordinary MRI machines. And so for the first time in a minute, you can see cancers that are otherwise invisible, um, that aren't, which don't have stuff circulating in your bloodstream, that don't have more morphological variances that can be seen with an MRI. They, you light up like a Christmas tree. And so that will be the standard of care for cancer detection and neoadjuvant treatment. But it started off as a quantum computing company, right? It's one of those examples of the roadmap pointed one direction, the physics revealed something else.

Is this a portfolio company?

Yeah, yeah. The company's called Envision Imaging.

Okay.

They're in Ulm, Germany, but they come from, I think there's 40 different countries on the payroll. Uh, and so it's like a Manhattan Project for these quantum contrast agents, but they are— there are now 9 cancer centers around the world that have these machines, uh, and their clinical systems will start shipping next year.

Okay, all right, your top 3 could be bullshit ideas or things that people are spending a lot of money on that they shouldn't.

Yeah, okay, 3. I only get 3?

Well, you can have as many as you want. I don't want to All right.

I'll be quick. And look, so I mean, fusion, fusion comes up a lot and the civilization will operate on fusion, 100% guaranteed. Right. But that's a long way. It's hard and it's a long way off. Right. Uh, the physics of it don't allow, there's no free lunch. Right. And these systems need to be big. All the physics demands scale. So you need billions of dollars to come up with a new answer. And a surprising number of answers have already been collected over the last last 100 years, right? And we, we achieve fusion before fission, and it's really hard. I mean, the only, the only, uh, one that really works uses gravimetric confinement, and we orbit it, right? And it works great. Uh, and nuclear weapons work great, but, um, harnessing it as a tool— I love the— I love what the Helion guys, their ambition is— this really, really, really hard. And I think things like arguments for it, it's like, well, it's clean and it's this, that, and the other thing. It's like, well, so is fission. It's, it's, we've gone to the world's most difficult physics to avoid an education and engagement on science problem. Right. And similarly with SMRs, they're, which are small modular reactors, small modular reactors, which are they're trying to solve a financial problem essentially, that big reactors are too expensive to build. And so you want to build little ones. But the trouble is there are all these power laws and scale laws with those. And so the cost per megawatt is much higher with SMRs, and it's not clear you can build them any faster, especially if they're first-of-a-kind design. And we should just be building known reactor designs that are very big, that are very safe, that can be built in less than 4 years.

Reactors.

Those reactors exist and we should be building those. But more importantly, the reactor fleet that we have in the US can be upgraded to produce SMR-like additional amounts of power without changing the fuel cycle or the mechanism of the reactor, simply by using modern steam generators and operating these things to work. You know, it's like you've got a Datsun from the '70s, and it's not operating at its rated horsepower, well, you can go and, you know, you can operate and put new seals on it and use better fuel and, and better injectors, and you can get another 20 or 30% out of it. We can— there's 10 gigawatts just sitting there in the US fleet where within a handful of years we can be generating SMR-like amounts of power out of the existing reactors. And then beyond that, if we just build many of the same proven designs, you can add 150 gigawatts to the existing fleet without changing the fence lines, right? Without, without changing anything about the transmission or cooling, but just add new units to the existing plants. And so I see the fusion stuff. It's important science. We should be doing it. It's not imminent and not something we can invest in. I think the SMR stuff is largely unwarranted innovation. And we should be building big modular reactors that we know work, that we know can be built for a fixed amount of time and money.

Wait, so we should do fusion and we will. We're just not smart. You, you think we need new tools to arrive at some of the stuff we've been talking about to do it smarter and better.

So I think the way that you— I think the way fusion happens, candidly, is with nation-state scale things like ITER, like JET or EAST or ITER, right? These things require concentration of capital because these things favor scale. Right. And you don't, you'll get nowhere giving 100 companies a billion dollars. But if you concentrate those bets on a few modalities and have $20, $30, $40 billion programs, we'll get there. Right.

Well, and then even what you said on existing plants and fission makes sense. However, it's been a totally intractable problem. I mean, we can't build anything.

So it's like, even if you're, you know, well, the most recent reactors are Vogel, right? But they were first— they were first of their kind reactors. There are other reactors that are, you know, the ABWRs. I mean, AP1000s now, we built a couple, we'll get, we'll get good as well.

No, I just mean the public sentiment hasn't—

Yeah. So that's so— but the public sentiment's not grounded in reality.

I get that. But I'm just saying it almost seems like if you put some new eye candy and you're like, well, we didn't do that, we're going to do this, people get excited or something by something. Well, that's true.

Maybe that's true. That like when, like people realize, well, your fusion reactor is producing neutrons, right? You still have biological shields. You still have a whole bunch of, there's many of the same things. And the waste problem for fission versus fusion differs by degree and not kind, right? That the fact that you could store, you know, Yucca Mountain is not a problem. Waste is a consideration, but not a problem. Yes, it's radioactive for a long time, It's only dangerous after 300 years if you eat it, right? If you don't, if you remember not to do that, waste is an entirely tractable problem with existing, very safe, very large, 90% plus capacity utilization capacity of the technologies. It's just an obvious thing as opposed to continuing to burn coal and burning fossil fuels still kills millions of people every year, right? So we We have— nobody's ever died from nuclear waste, right? We have good reactors that are safe, that we know how to build. And, uh, I think it is critical that we lean into every safe, scalable, uh, form of energy and love wind and solar. It's perfect in northwestern Australia for turning iron ore into iron. That's great. There are places where the wind don't blow and the sun don't shine. We need nuclear reactors there. Um, so everything would be part of the solution, but we know how to do it. Like, and we've been doing it for an awfully long time, and we don't have to find new physics and eye-watering amounts of venture capital to, to do it.

All right. Bullshit item number 2.

Well, we gotta talk about, let's talk about, uh, space data centers, right?

Okay. 'Cause Elon and Jeff Bezos, they, I mean, and Eric Schmidt, right? You know, okay. They all have rocket companies, but they all have rocket companies.

And they're all brilliant. And I, so I mean, the problem, the problem with putting things in space, apart from the fact it's in space. And so from serviceability, all that sort of stuff. And yes, you get, it's a good place to do solar, um, modular, what it costs to put it up there. But what you're really building are radiators with a few solar panels and a few GPUs on them. Uh, and in a harsh radiation environment, is it impossible? Absolutely not. You could do such a thing, but it's in no way practical using the current technology. There is a technology which is uniquely suited to do it, right? And that we are building power-hungry systems with, with silicon, which has very mature understanding of the physics of that. There are other ways of doing computation that don't use semiconductors and don't use transistors., and they're emerging in quantum, but Josephson junctions are a good substitute to transistors, but they use 5 orders of magnitude less power to switch. So you can build a megawatt data center that's about this big if you can keep it at 4 degrees Kelvin, right? Which happens to be the temperature of space. Um, Josephson junctions allow this incredible density because they don't generate, you know, 5 orders of magnitude is a lot different in the amount of heat you produce. And I think any scale electronics in space are going to be based on that superconducting, um, classical Josephson junction logic.

What is a Josephson junction?

It's this, uh, so it's, uh, transistor has, uh, 3 connections. Josephson junction have 2. And the basic, the basic idea is that it, it's a superconducting path that will superconduct until you reach a certain threshold of current, and then all of a sudden it will lose its superconductivity and spit out this quanta of flux, right? And so the technology, the computing technology is generally known as single flux quantum. There are variants of the logic, but they are, those 5 orders of magnitude are precisely what you need to move beyond having to start many, many nuclear power stations to power AI, it gets you in a regime where you spend a couple of orders of magnitude to keep it cold. But having built warehouse-sized cryogenic plants for PsiQuantum, that doesn't seem that scary anymore. Like keeping things— like cryostats are pretty simple. It's 4 degrees, which is not that cold. And just the incredible densities you can achieve without nearly needing to, you know, build giant liquid cooling systems that if you take the heat out before rather than afterwards, you can make these things pretty small. And especially when you're in space, which you're generally, you know, you're in the tens of Kelvin because of the way aircraft, those spacecraft heat up and stuff, but you're much closer to having something that you could deploy at scale. I suspect that that's what Elon is planning on using. I think we, we, we have the leading, you know, uh, SFQ company that will have product in the next couple of years that demonstrate this. But these devices have been around for years, right? And they've been used in signals intelligence work, uh, and for DSP-like applications, but run at, you know, there, there were devices that were built in the '90s that ran at 700 gigahertz, right? Which is when our laptops sort of tap out at 4 or 5, it's a, it's a big it's a big discontinuity.

So you don't think these guys are totally fooling themselves? You think they're betting on future technology or you think they have fooled themselves a little bit?

I think when you are, well, when you are Jensen and you have a lot of value created in AI, you want to be able to manage the spin of the quantum thing, which may replace some of those core scenarios, right? So very, very clever people can be wrong if it serves their commercial interests. Uh, and, you know, I think that they very quickly will be aware of the real opportunity, right? Which is not only the fact that you can make these things small, make lots of logic, 10,000 times more resilient to radiation. It can survive prompt nuclear events, which is of interest for, you know, both commercial and defense applications. And it's just obvious that that's how any significant amount of computation is going to get into space. The math on having a 4-kilometer-wide object in a sun-synchronous orbit that, you know, can't maneuver, that has as much area dedicated to, you know, radiation, heat pumps, it's just the physics doesn't pencil out. It's, but 3 orders of magnitude difference or 4 orders of magnitude difference. Yeah, that'll work.

All right. Um, I only have a few minutes left with you. There's 2 things I want to do. Um, one is let's make some friends. How do we fix Australia? A country with vast resources, um, many, many wonderful attributes. On my visits, lacking some entrepreneurial spirit and ambition.

Well, that's actually quite, yeah. Look, Australia is a very, it is a very lucky country. It has the highest density of quantum experts in the world. There's a reason why we built that machine in, or we're building that machine in Queensland. So leads the world in science and engineering publications, or is world-leading in those. That has the economic complexity between Gabon and, you know, Pakistan. It's because they've become wealthy digging things out of the ground and selling them to China, right? So coal from the East Coast, iron ore from the West Coast, wool, right? It's a, it's a mining state, but there is tremendous capital, tremendous talent, and I think real opportunity to really change the way supply chains work. We have a company We're building a solar farm, which is 35 times the size of Manhattan and have another company that has a process to turn iron ore into iron and do so in Australia, right near a port where fully a quarter of the world's iron ore moves through that port. But we'll put iron onto boats instead of iron ore, and that will go to Korea, Japan, the US instead of going to China. And that goes for silicon and a whole bunch of metals. It's true of critical minerals. Nearly all the mineral wealth in Australia is, is all the mineral wealth in the world is in Australia. We haven't bothered to dig up most of it, and it's the best renewable site. So 70% naturally firm renewables, right? You've got that thing in sun synchronous orbit. There's a place in the Pilbara where the sun always shines, and when the sun doesn't shine, the wind blows, and it's ridiculously firm. If you have a process that matches that, right? An industrial process that can handle firming, you don't need batteries, and renewables are dirt cheap for turning iron into iron ore, iron ore into iron and rocks into polysilicon and all that other stuff. The other thing is I think Australia can lead in advanced semiconductors. I think that's a good place to build fabs and advanced packaging, great place for fabulous semiconductors. So we just gotta figure out how to get, again, how do we best use public or semi-public capital sources to create innovation? And so where, you know, I grew up there, my mom and dad need to know what I do for a living. So we're going to be putting a lot of energy into how do you translate all of this wealth into economic complexity, economic growth and vitality in that, in that place. Because I think Australia has a leading role to play in quantum, in all these fields. In biology, we just did an amazing phase 1 clinical trial with, uh, uh, in Australia with, um, this amazing Professor Dr. Georgina Long.

With Strand.

With Strand, right? Yeah. You met with Jake. Yeah. That I think she will have a role in globally curing cancer with, in collaboration with the Strands, but also other work that they're doing. So yeah, Australia is an important area of focus for us. And I think it's globally unique in its role in changing the way we make metal in a way that's much more economic for Australia, but also much more sustainable.

I forgot to state this part. I'm half Australian before I say what comes next, and this comes from a place of love, but almost, it's almost too nice to live there. Very complacent people. Tall poppy syndrome runs pretty high. Many, many of the smartest people run off to places like Silicon Valley, just like you. Yep. Yeah. I mean, how, you know, is this future that you're describing, this is just thrust upon them or they find this inside of themselves to do this locally?

No, I think, look, there's a lot of enthusiasm, right? So we spend a lot of time working with the superannuation funds, with the Australian government. And there are the fact that they gave Sai $940 million between the Queensland government, the federal government is a real watershed. And there are really well thought through special investment funds to do this. There's just not the machinery of Silicon Valley there yet, right? There aren't that many exemplars. There aren't that many startups there. There is a good but relatively small venture community there. And so we can build that stuff up over time. But what's utterly unique is they have capital, natural, and intellectual resources that are all sitting there side by side. And they just need to be catalyzed to turn into some real opportunity. And the same with the UK, same with Europe, that there is a, you know, the Contrast agent company is based in Germany. They benefit greatly from new innovation around non-dilutive capital and acceleration of crossing, chasm crossing things for deep tech. So we're trying to catalyze all of those in places that make sense. 'Cause I think liberal democracies have to lean into staying ahead and we could do it. The degrees of freedom to do it are incredibly important. But if we don't, our adversaries are going to, you know, they are executing extremely well and we have to figure out how to deploy money to do useful things that have huge returns well beyond what a, you know, a superannuation fund is going to get out of building another tunnel.

Hey, all right. Last question. It's, it's your, you get a choice. Um, sort of the best idea that the least number of people know about that you've run across, just kind of an amazing idea that, that people aren't aware of, or the thing that you believe that you kind of think the least, least people agree with, your most, your most contrarian idea.

Oh, so you do either or? Yeah. I can take the first one. Look, I think Uh, we are entering a period of extraordinary agency over nature and, and our physical world. Right. And I think one thing that exemplifies that is the question of why plants are green, right? Which is they evolved in a weird evolutionary niche. Or chlorophyll did 3 billion years ago, accident of history that it's green, right? That's where all the good photons are. So, um, there was some other organism eating all the green photons and when we decided that plants should be green and not black, right? They should be black if you want them to be energy efficient. The metabolism that was created to turn those photons into sugar was created in a different atmosphere, in a world that no longer exists. Uh, then we can go fix that. Like, we can go back into plants and double their productivity or more by making very selective edits to that metabolism and the way plants work. So we can double the productivity of corn and soy. We can make trees grow much faster. Um, that we are not at the end of anything, right? We're living right at the beginning of history and things like agency over biology, agency over materials, agency over these new kinds of computation give us this absolutely unprecedented set of tools, uh, to create abundance and reduce suffering.. And I think that it's tough sometimes where we see, you know, shitty technology like the metaverse or NFT sucking up all the air. But with the right storytelling, the right capital, the right science, we can do anything. And there's never been a period where we need to do it like there is right now. So very optimistic about the future and very optimistic about the small role that Playground has, but also excited about the role of what you do, which is about getting people to engage with these ideas and these opportunities, because they're important. Um, and correctly done, it, you know, it works for everybody.

Wow, it's nice to end on this optimistic note. And, uh, you know, I just want to reiterate a bit what I said at the beginning. Um, huge thanks first of all for you coming on the show, and then just massive thanks. I, I don't know if people fully get— I don't think there's many companies that would basically one email, let me show up and start a company and have a podcast studio and then surround me with all interesting people that I can badger as I desire and tap into. As you can tell, Peter's brilliant and one of many brilliant people here. And so, no, I mean this deeply. It's been better than I ever could have hoped. I sort of unreasonably lucky to have just kind of stumbled into this in some ways. So I think Elon brought us together too. So there you go. So thank you. Thank you so much for letting me crash here the last year. And I'm glad people got to hear some of, you know, how you approach the world.

Delighted and looking forward to the next couple of decades together.

The Memory Podcast is hosted by me, Ashlee Vance. It is produced by David Nicholson and me. Our theme song is by James Mercer and Jon Sortland, and the show is edited by the Jon Sortland. Thanks as always to Brex and E1 Ventures for making this possible. Please visit our our Substack, YouTube, and podcast channels to get more of what Core Memory makes. Thanks, y'all.