Pablos Holman | On Creating Technology That Actually Matters

Pablos Holman: If you add up every software company—that includes Facebook and Microsoft and Apple and Google and Salesforce—even throw them all together, their combined revenue is about $2 trillion a year. Global GDP is over a hundred trillion a year. So the tech industry that you read about every day, that seems so massive and infused in everything, is doing 2% of what humans rely on. Most big problems in the world, you can trace them back to energy. You could solve clean water if you could solve energy. Solving clean water solves health, but you can't solve clean water until you solve energy to make the water clean. So, you end up just following this trail of problems until you end up at energy almost all the time. And you can see that now AI is like, "Oh, great. We need chips." And they went crazy building chips. They're like, "Oh, how are we going to power these chips?" And so now the whole tech industry has woken up. And it's very important because philanthropy failed to solve clean water. Charity failed to solve clean water. And if you don't solve that problem, you won't solve the health problems that exist.

Beatrice: Today I'm very excited to be talking to Pablos Holman. So Pablos is an inventor, a hacker, self-described, I think, and someone who has worked on everything from lasers to stop malaria to new kinds of nuclear reactors. And you now have this new book out called Deep Future. And so I'm very excited to talk to you about that today.

Pablos Holman: Yeah, there's the book. Did I send it to you? I have to send you one.

Beatrice: You sent it to me.

Pablos Holman: Oh, I did send you one. Yeah. Okay. Right. That's right. Cool.

Beatrice: Maybe we just start with the book. Tell us, why did you write this book and what is it about?

Pablos Holman: I think I wrote the book out of desperation. I'm not trying to be an author. Hopefully, it's the only book I have to write. I'm trying to help people see a little bit of what I see, sort of a different perspective on what technology really means. And I sort of feel like we live in this world that we think is full of technology but really it's just been full of software.

And so the difference is important because we got so drunk on software that we really haven't been able to advance other kinds of technologies with the same sort of fervor in the last couple of decades or so, and we're missing out. There's just so many other technologies that could make a huge difference and that could help us solve bigger problems, and I'm excited about those things. So the book is called Deep Future but the subtitle is Creating Technology That Matters, and I think if people think about their relationship with technology, a lot of times they're frustrated and disappointed that it's making their life a little better but not in the way that they wanted. And so when you think about the biggest problems in the world, they tend to be fundamental, meaning you've got to solve them first if you want things to go well. Those are things like food and water—clean water—and energy and construction and manufacturing and mining and shipping. These are things that every human on Earth relies on, and none of them are going to be solved with just iPhone apps. And so to me, deep tech is about going after these other things. And we have that possibility. And it's an exciting time because the world is sort of waking up to that. And a lot of the shifts are happening to make it possible to work on these things now. And I think it's a very exciting time.

Beatrice: Yeah. And I think throughout this conversation, we should dive into this and maybe explain to people what deep tech is and so on. But just before we do that, I think it would be nice for people to know a little bit more about you and who you are. And I know that's something that also runs through the book a lot is actually your story as well. And I think what I came across when I looked you up first was this, I think, 25 million views TED talk of your work as a hacker. So maybe tell us a bit about who you are and also tell us about this hacker mindset that you have.

Pablos Holman: Yeah, so I grew up in Alaska and I got one of the first computers in the state, basically one of the first computers you could have at home. And I was a kid and nobody had really ever seen a computer, at least where I came from, until they saw mine. And so it was this weird world where computers weren't very useful yet, but I was excited about them. I could see that someday they'd have more memory and someday they'd be faster and someday they would be useful. And I wanted to use computers to do things to help people in the world. And nobody believed me. I mean, everybody thought I was a crazy kid. And I remember distinctly having the computer, which was an Apple II, so one of the first couple thousand computers Apple ever made. And then I also had a skateboard, which most people thought the computer was a bigger waste of time. And so it just was this really strange environment. And because it was so new, there wasn't anyone to learn from. I had to kind of just figure it all out myself, which is excruciating. You can pick up an iPad and sort of figure out what to do with it. But in those days, computers didn't help you. You had to figure it out yourself.

So I ended up on an unusual track, a very early start with computers, and I just learned more and more about them. And it turned out, Revenge of the Nerds, I was right. Computers actually did get useful and more powerful and they went everywhere, and we all know that at this point. But the whole time I just sort of felt like I had this superpower. I had something no one else had. I could do things that no one else could do and I wanted to show them and share that and grow it and make it better. And so I ended up out of desperation when I was young just trying to get people excited about something they didn't care about with the computers. And you could say in some sense I'm still doing the same thing.

Eventually, I worked on computers for a long time, advancing them through the 80s and 90s, putting them into different industries for the first time a lot of times. And then by the 90s we're building the internet and trying to get people on that. And then I worked on a string of weird projects. I was working on cryptocurrency in the late 90s and AI. By 2000, we were trying to make AI that could read the news and trade stocks. And then in 2001, I went and helped start Blue Origin, which is a spaceship company. And Blue Origin was a way for me to start using computers to advance other kinds of technology. And that's really where things shifted for me. And really since then, I've been working on using computers—because that is the most powerful toolkit we have—but to help us advance other technologies that could make things even better. And I think that Silicon Valley at the same time sort of got drunk on software, and they just wanted to use computers to make money, and that's cool. You need to do that and you need to use computers in different businesses and things, and that's a lot of what has been going on. And using them to buy more [...]. But it's kept our brightest nerds from working on solving other problems. And so to me, that's how I got on this track. And you could say maybe I got to a lot of the technologies a little early. I think if you ask a lot of people working in tech, most of them would rather be working on something much cooler than spam filters or whatever. And so it's an exciting time where I think they're finally going to get to do it.

Beatrice: Yeah, we have a lot of those nerds at Foresight, I think. And so could you expand a bit on why do you think that they can finally do it now, or what is changing?

Pablos Holman: What's changed is that I think all the easy stuff has been done. We kind of had to apply computers to every business—to accounting and to law firms and to supply chain management or whatever—and then we had to sort of redo it for the web and then we had to redo it for mobile and now we're redoing it for AI. But basically, it's all been done. We might make those things a little better, but computers are doing them.

And so the opportunities were there, but if you add up all of the software in the world, like every software company that includes Facebook and Microsoft and Apple and Google and Salesforce, even throw them all together, their combined revenue is about $2 trillion a year. And global GDP is over a hundred trillion dollars a year. So the tech industry that you read about every day that seems so massive and infused in everything is doing 2% of what humans rely on. And I think it's just a very important thing to understand, that thing to calibrate. When you look at the other 98%, that's those things I mentioned before. That's hundred-year-old industries that are moving physical stuff around that you can't live without.

And now I think the machinery of startups and the machinery of venture capital has proven itself. It's worked as a way of doing new things, right? And we're good at doing new things with that. We've just been aiming it at what I think of as not very ambitious targets. So if you look at a a SaaS investor or even a Y Combinator deck or something, the first question is always, "What's the TAM, total addressable market? How big is the market?" And you might be really excited if you have a market for Zoom or Slack or something that's measured in billions or tens of billions of dollars a year. I said software adds up to two trillion. We have a company making cargo ships. The shipping industry is two trillion alone. It's as big as tech. Now, it's a shitty industry that burns five out of six dollars on fuel, but hasn't changed in a century really. And we have a team that's making a cargo ship that is sailing with computer-controlled sails, so it doesn't need fuel. And that's not that hard to do anymore. I mean, computers could certainly control sails. It's a lot easier than self-driving cars, and sailing across an ocean has been proven for several centuries. So, I think we're at this amazing point where if you just look at that and say, "Why isn't that happening?" The shipping industry isn't going to do it. They're not robotics nerds. But if you're making robots, I mean, okay, you could make them vacuum a floor or something, but why not go after these big old industries where you could make a huge difference? So, that's what I think is possible and that's kind of how I think about it.

Beatrice: Yeah. Let's maybe dive into the specifics of it a bit more. So you mentioned already deep tech. I think it would be good just to talk a bit briefly and say how you define deep tech.

Pablos Holman: Yeah. I think it's not that important to define, but I would say for me deep tech is essentially just everything but software. It's all the things that Silicon Valley has been ignoring, but are actually new technologies based on new advancements in science and technology, new inventions, new things, new ways of doing things. And those are technologies that could help us. For me, the other part of it is they are technologies that could help us solve big problems. And by big problems, I mean unsolved problems. So things that we're not going after right now. And that could be going after diseases. It could be diseases that the healthcare industry or biotech isn't able to focus on right now because they're not a big enough market. It could be infectious diseases that we solved in rich countries, like malaria, but still take a lot of lives all over the world. That's a case where in a lot of these cases, version-one tech was enough to solve the problem for rich people but it wasn't enough to solve the problem globally.

And there's this very important, I think, very important thing for humans to do, which is just step up and recognize, like, we made 8 billion people and they're us and they're our neighbors and we need to take care of them and we need to provide for them. And in the case of infectious disease, malaria is one that I worked on a bunch. We had malaria in the US and we were able to wipe it out just by spraying chemicals that kill everything. That's what DDT is for. And it worked well enough to solve the problem for Americans. And it worked in southern Europe. I don't think Sweden ever had a big problem with malaria. You probably have a mosquito problem, but not a malaria problem. And so what I want to do is figure out, all right, that technology of spraying chemicals wasn't good enough to solve the problem in some place like Africa, but what if we could invent better ways of attacking that disease and containing it and getting it under control? Then, well you could save... I mean we are doing that now and it's working and we're getting malaria down, but it was a million lives a year when I started working on it. Now we're probably a little over half a million lives a year. And that's what's possible. Half of those are kids under five years old. They never had a chance.

So what I mean by deep tech is take on a bigger problem. And one of the things that Silicon Valley got wrong, or I think is limited in its value, is this maxim of "scratch an itch." Scratch an itch is one of the first things you learn in Silicon Valley. And what it means is find some problem that you have. You're the first customer. It's a way to do budget market research. If you have this problem, probably your friends have that problem, and therefore maybe there's a market to go after. But the biggest problems aren't problems you have. If you're a YC dropout in San Francisco, you have a problem that your iPhone battery doesn't last all day, but you don't have a problem with malaria. And maybe malaria is not the biggest market and you're like, "Obviously I wouldn't try to solve that because I can't get paid to do it." All right, fine. But shipping is a big market, bigger than your entire software industry, and you're not going after that. So I'm trying to show examples like that where I think deep tech is going to bring much bigger opportunities to the founders, to the talent that we have in the tech industry, and to the investors that have been attracted to the tech industry.

Beatrice: Yeah. So what do you think is holding that back now? Is it always because, I guess, that's the case that you're making, that you can actually make money on a lot of these really big problems as well? Or do you think that for something like malaria, it's never going to get solved by industry or something like that, we need nonprofit?

Pablos Holman: So a couple of things. One, I think if you think about that shipping company I described, right now it's deep tech. It's crazy tech. It's nerds in a garage trying to make a boat that can drive itself. That's going well, but you could call it some crazy project that Pablos is interested in. But the day that first ship sails and we deliver a cargo container, it's not tech anymore. It's just a better ship. So then the question is, do we sell that to Maersk or do we go build the next Maersk?

And one of the things that Silicon Valley did is develop the Uber playbook. Uber replaced an industry with a startup. And that's a very important thing to understand because these industries aren't improving themselves. They're cinched down. They're trying to keep things the way they are. The taxi industry wasn't going to make an iPhone app. How hard would that have been? But they weren't going to try. They were going to try and keep things the way they are. That's what's possible with the systems that have been created in the tech industry.

And then I think the other part of it is the folks who are in the tech industry have to do these things. For all this disrupting other industries, everybody working in shipping started their career at chapter 37. They don't have any idea what it was like to start that industry. They don't have any idea what it was like to do a new thing. They started their job to do last year's thing maybe 1% better or exactly the same. And so those are not the people who are going to reinvent these things.

The cool thing about software folks is they're the ones who in one lifetime have had that experience of having an idea and building it and shipping it and scaling it and selling it, and maybe doing that two or three times. They know how to do new things. So, we need them to build. And if you go look at what's happening in the few deep tech cases we have in Silicon Valley, that's what's going on. It's software nerds. Look at Boom Aerospace as a great example. It's software nerds building a supersonic jet. That is awesome. But why they can do that is because the mythology in Silicon Valley is that hardware is too hard. That's why it's called hardware, right? And they think it's 10 times harder. But you know, you look at something like Boom, they design that jet in software, they test it in software, they crash it in software, they do that thousands of times and iterate in software before they go build a jet. And they're going to need to crash some jets, but they aren't going to need to crash nearly as many as Boeing or Lockheed did to get where they are.

So I think it's a new world where one of the other major things that has changed is our computational capacity is so powerful now that we can model everything in the world in order to run those simulations of what the jet should be. And that is what SpaceX is doing with rockets. That is what Tesla is doing with cars. It's all designed in software. A Tesla, a Cybertruck, is in GitHub, right? This is how these people work. And so you want to learn from that, but thinking about what are the other big problems in the world that could be solved in that model—and that's why I'm so lit up, because I'm seeing that happen more and more.

Beatrice: And I know you dive into a few specific ones in the book. Do you want to tell us which technologies you are most excited about or see as the most tractable?

Pablos Holman: Oh man, there's a lot of things. So one of them... so for me, when you work on these kinds of problems—this doesn't really happen if you're making Snapchat or something—but if you work on most big problems in the world, you kind of trace them back to energy. For example, you could solve clean water if you could solve energy. And solving clean water solves health, but you can't solve clean water until you solve energy to make the water clean. So, you end up just following this trail of problems until you end up at energy almost all the time. And you can see that now AI is like, "Oh, great. We need chips." And they went crazy building chips. They're like, "Oh, how are we going to power these chips?" And so now the whole tech industry is woken up to the need for energy.

So I think that's exciting and it's very important because if you go back to your question about malaria, can capitalism solve malaria? Philanthropy failed to solve clean water, right? We've been trying. Charity failed to solve clean water. It doesn't matter how much money you have, nobody has enough money to just go solve that problem. And if you don't solve that problem, you won't solve the health problems that exist globally.

But now that hyperscalers, which are the biggest industry in the world—in our lifetime, the biggest industry was oil, right? That's energy. But the oil industry is old. It's century-old businesses. They're cinched down like I was talking about with shipping. They try not to change. Probably Congress was staffed by Chevron and Shell for most of our lifetime. But now it's the hyperscalers. The hyperscalers are the biggest industry and they need a lot of clean, cheap power to power all the chips they bought from Nvidia. So what's happening right now is they're staffing Congress. Congress last year passed a super-bipartisan act called ADVANCE to open up building nuclear reactors in the US. And then last month a couple of executive orders were signed to free up building nuclear reactors in the US. That is very important because in our lifetime, nuclear reactors have basically been outlawed. When I was a kid, the world sort of conflated nuclear reactors with nuclear bombs and then we just outlawed the wrong one. If you had done it the other way around, you never would have heard of global warming. That's what's possible. But we're still living on the wrong fork in history. And we're still not being honest with ourselves that we could have prevented a lot of the problems that we're really pissed off about now, a lot of the climate change stuff in particular.

But the other one is if you had clean, cheap energy like a nuclear reactor can provide, you could get clean water for free. If you get the clean water, then you solve the health and disease problems, at least partially if not wholly. And so, ironically, it might be the hyperscalers that save us. And I think a lot of times this is what people miss. They want to have this story in their heads that business is evil and big business is even more evil. And while there's certainly some bad actors and things that are going wrong and things that need to be cleaned up and improved for sure, philanthropy doesn't scale. What scales is business. And so if you can find a way to take the thing that you care about solving and map and attach it to a business that can go forward, that can sustain itself, then you have a much better chance of solving the problem.

Like, we had this project, not crazy, but when I was working at the Intellectual Ventures lab, Bill Gates was supporting that lab in a lot of ways. And he asked us to help figure out what to do about these smallholder farms. People probably don't even know what that means. There are half a billion farms in the world that we call them smallholder farms. It's like a family with two goats and a couple of chickens and a cow, right? Like that might be a big one. So they're all over the world and these people rely on milking the cow. They milk the cow twice a day. They milk the cow into a jerry can, literally like a used can made for gas. They're rinsed out. The milk is going in there. They take it to market in the morning. They milk the cows in the morning, they take the milk to market. They sell it. That's their income. Then what happens is they come home and at night they milk the cow again, but that milk is going to spoil overnight. There's no refrigerators on these farms. There's no power. There's no energy. So the milk spoils. So if we could figure out a way to preserve that milk overnight, maybe we could help these folks double their income. That's the idea.

So anyway, that's the problem we were trying to solve. And so we had a lab full of PhDs in physics and chemistry and biology and we went crazy on this. And we invented this insanely cool microfluidic ultrapasteurizer manufactured in a semiconductor fab. And it was the most high-tech thing for pasteurizing milk ever invented. And we could never make it. Every time we took it to the field—it worked fine in the lab—we would take it to the field and it would immediately foul up and it was impossible to clean. It never could work. And then finally, after we iterated on this and we went to Africa a bunch of times to try stuff out, what we ended up with is this milk jug. We invented a better milk jug. It's made of food-grade plastic. It's big. It's got a huge opening on the top so you can milk right into it from the cow. It's got this black funnel so when you're milking you can see if there's any bacterial growth in the milk. So it goes straight from the cow into the container, and so it's sterile. It's easy to clean because it's big. You can get your hand in there and clean it out. You can't do that with a jerry can. So it's the dumbest, simplest thing we ever invented: a milk jug. And then you screw the lid on and because it's clean and it keeps the milk clean even without refrigeration, it'll last overnight, and they can take that milk to market and double their income.

So, this is called the Maziwa can. It's real. We invented that thing. It's in the market. It's all over the world. And how we did that—look, even Bill Gates doesn't have enough money to just send milk cans all over the world forever. So, what we do is we take those molds, we go work with a local businessman in that region, in every region. They produce the cans locally, and they ship them and sell them, and they make some money selling the cans. And they can keep going long after we're all dead. So that's the idea. You contextualize the solution in a business that can go. So that's a long story, but you get the idea. I'm trying to show that a lot of times if you can take the problem you're trying to solve and map it to a business, you can find a way to make it actually scale up and have the impact you're looking for.

Beatrice: Yeah, I think it's a great example.

Pablos Holman: Sorry, I'm not good at short versions of stories. That's why it's a book.

Beatrice: Yeah, and I thought it was really interesting. Are there any other... we can trace it all back to energy, you said, which is really interesting. Staying on topic, how do you actually choose? Are there... do you have a filtering lens in terms of which crazy ideas you think are actually worth betting on, but maybe people are dismissing too quickly?

Pablos Holman: It would be idiotic to invest in a crazy idea, but it might be genius to invest in all of them. So I sort of index on crazy ideas and by doing a lot of them, we'll find something that works. And so we typically are the first investor for mad scientists, right at that point where they're kind of coming out of the lab and into a startup. And so I see a lot of these things and we invest in a lot of them. I'd say on the most practical side—so we've done a bunch in energy—so the most practical side I'd say is we invested in Deep Fision, which is a nuclear reactor that will fit through a manhole and we bury it a mile deep in a borehole. So this is a nuclear reactor that is unquestionably safe. It's a mile from anyone's backyard under 10 billion tons of rock. Nothing can go wrong, but if it did, there'd be no radioactivity at the surface. So you just fill the hole with dirt and forget about it.

And so this is a way to not only make nuclear reactors safer, but it's small enough that you could build it in a factory. And one of the big problems with nuclear reactors is we've built them all so big they cost billions of dollars. And every one of them is like this bespoke art project. We have 92 reactors in the US. None of them have any interchangeable parts. Everyone has their own engineering team. They're all one-of-one, which is the most expensive way to make anything. A Deep Fision reactor is about the size of a Toyota. You can make it in a factory like a Toyota. Make thousands of them. And that buys you a couple of things. One, you can iterate on something small. You can make little tweaks and make it better here and there. And you can't do that if you're making billion-dollar machines. Iteration is what makes Silicon Valley so powerful. That's why software won: the iteration cycle is so fast and cheap that you can just try things for practically no cost. So rapid iteration is one of the other very important fundamentals of Silicon Valley. So if you could bring that to reactors, then they could get better and cheaper and you get economies of scale when you're making more than one of something. And so anyway, this thing will get made in a gigafactory and make thousands and thousands of reactors and we can deploy them all over the world. And that's what's coming.

And so that's really cool because you could put that thing under a data center or a neighborhood or a skyscraper or whatever and get the energy you need right there. It's totally clean. You don't need storage systems. You don't need transmission lines. You just put the energy where you need it. And that's very exciting. We're at a point now where we can design totally safe, efficient nuclear reactors, and we weren't allowed to build them. The regulatory environment was just too extreme. And so now that's all changed. The NRC has completely been overhauled. They regulate nuclear in the US. Most other countries, whether they want to or not, follow the US on nuclear. And so that's what's happening. And there's a hundred other nuclear reactor projects. A lot of them are also cool. I hope they all succeed. I'm not... I think Deep Fision is the best one I found, but there are a lot of other cool ideas out there and we kind of just need to do them all.

Sweden is awesome. I got to brief the Swedish Parliament last year and they've been doing great things with nuclear now and that's coming along. So I'm really excited about the future of nuclear in Sweden and hopefully they'll really set an example for the rest of Europe. France has been doing a good job. I won't name other countries like Germany. They're not doing a good job, but they'll come around. And so these are things that are just very important. So nuclear reactors, I believe, are the safest, cleanest, most scalable thing we could do right now. It doesn't need any breakthroughs. We've proven all the tech. We've been doing it before. All we got to do is get organized and do it. So to me, that's very exciting.

There are what are called Gen 4 nuclear reactors, which would be even better, meaning they have a more efficient fuel cycle. They get more of the energy out of the nuclear waste or they can be powered by nuclear waste. I worked on one of those a long time ago called TerraPower and they're still working on it, but they were a little too early and couldn't get it approved. But hopefully soon they'll be able to build those things too. So lots is coming in nuclear reactors.

And then I work on the crazy stuff. So we have a team building, putting solar farms up in space where they get sun 24 hours a day instead of just half the day. That sounds like science fiction but is actually quite practical. We have all the tech to do that. The only hard part was launch cost. Thanks to SpaceX, launch cost is down about 40x from where it was in the Space Shuttle days and continues to go down. So, it's getting cheap enough that you'll be able to store your old sports equipment in space instead of your garage before long. So, it's coming along. The company's called Vertis Solus and I think they'll have the first commercial array up in probably about four years. So, that's what's possible. And then you just beam the energy down to earth using radio waves that go right through clouds in the middle of the night, and you could send that energy to the North Pole if you wanted or to Timbuktu. It can go anywhere. So that's coming. We also have been investing in a whole raft of even crazier ideas, but we'll save those for when they're a little less wild.

Beatrice: Yeah. I guess one thing that we have to discuss, I think no podcast episode is complete without it these days, is just AI and what role AI will play. I think in your book, you seem a little bit skeptical of a lot of the AGI narratives but enthusiastic about other possibilities. What role do you see AI playing in building hopeful futures?

Pablos Holman: AI, like a lot of these terms, lacks nuance. It's become a catch-all. I think we just rebranded machine learning as AI a couple of years ago, and now we're rebranding it as AGI this year with whatever OpenAI is shipping. And then next year it's going to be ASI. It's lost a lot of its meaning. But overall, the track we're on, I think, has a couple of things you could say.

So one is, what we call AI—it's so hard for me to use that term—but aside from just language models, it's great that computers can chat with you now, but that's not the point, right? The point is these models I described, the model for a supersonic jet, right? That's not a language model. But what it's showing is that we have such a vast computational capacity that we can create models for all kinds of things. And I think there might be more important things to make models for than just chatting.

So for example, in the book I wrote about one that we worked on starting in 2007, whatever, 18 years ago, to develop computational models for the spread of infectious disease. Part of the reason we've been able to make a dent in malaria and definitely the reason we've been able to eradicate polio once and for all again is because we're able to make these models of how the diseases spread and then use them to optimize the vaccination campaigns, the ring vaccination campaigns, to contain them. That's been very successful. We've been doing that for 15, 18 years now, probably 15 years. That team should get a Nobel Prize. I mean, they have been fundamentally advancing the ability for humans to go solve an intractable problem. And that's why, in the first Ebola outbreak, 12,000 lives were lost. And then a couple of years later in the second Ebola outbreak, 12 lives were lost because in part we were able to use these computational models to contain that disease before it spread. That's what would have been possible with COVID, but we weren't trying, right?

So I believe very strongly in the potential for these computational models to help us make better decisions. That's what they're for. That's what they do. They can take in more data than any human can handle. They can run algorithms that are miles long. They can analyze all that data. They can help us figure out and show us with simulations: these are your possible futures. They can show us this is what's possible. If you do this, you end up there. If you do that, you end up there. It's still up to us to decide which future we want to go for. And I think this is why I'm so excited about Foresight Institute and the things that you guys are doing because there's a chance here to embrace these models as tools to help us see our possible futures and pursue the one that we want, right? I think we still need to take responsibility for choosing which future we want to go after. Humans to date have mostly just been fantasizing about what future they want, even if it's not possible.

And so I think of these computational models as like Google Maps. It's a Google Map to the future. It shows you a blue line, this is the best route, but here are a couple of gray lines that show you other options if you want to do something slightly different. And we can do that now for businesses. We can do it for the rockets and the Teslas and the hypersonic jets and all those things. But what I believe is coming is we'll be able to do it for governments. We'll be able to do it for societies. We'll be able to run models and say, "Okay, we would like our life to be a little better in these ways. What can we shift? What can we change? What direction can we go?"

And so that's the potential that I see. I'm very excited about it. And I think the folks fixated on LLMs, they want to get there, but they're kind of psyching themselves out with a lot of imaginary stuff. The thing you learn when you're working on inventions is that you kind of have to put them in two buckets. You've got one bucket which needs a miracle and the other bucket that doesn't need a miracle. Nuclear reactors don't need a miracle. Fusion might. Quantum computers need a miracle, maybe several, but classical computing doesn't. And so it doesn't matter how much you want it. Miracles don't happen on a schedule. And AI still kind of needs a miracle. There's a lot we can do with what we have, and we've been advancing on that and making it better. There's a lot of places we can apply it. And I think that's all got great potential, but people are, I think, disingenuously extrapolating from where we are to this magical thinking. And so to me, I don't have a lot of time for that. I think it's sort of irresponsible and I think the job should be to figure out how to wield the tools that we have. And that is growing and it's important. I'm glad that we're throwing lots of computation at these things. I think the potential is vast, but the conversation has certainly gotten off the rails.

Beatrice: What do you think about the potential of AI in terms of advancing science and technology? So advancing R&D.

Pablos Holman: Yeah, I mean I think the potential is very high, in part because we haven't been doing a real good job of those things. I'd say the limiting factor is probably just human psychology. I don't know. We've scaled up investing in science. We always need more. We want more because the potential is high. It's scientific research. You invest in it speculatively and that's very important to do, and especially in the US and in Europe. We've invested a lot in scientific discovery and we want more. But the problem is we aren't getting our money's worth. We haven't been doing a good job of focusing that on things that matter. We haven't been getting the results that we should be getting for the investment. And so I think we have to revisit how we do that.

The nice thing about AI is you could kind of do both. You could do whatever dumb [...] people think they want to do and you could do the important stuff too. And so I think scaling up... I think the nice thing about being able again—to me it's more about computation than AI, but whatever—we're applying a massive load of computation to these things that we couldn't do before. Just to level-set, when I wrote this book, which was last... I don't know, I think I finished it in November, December, something. There was, if you add up all the supercomputers in the world, their combined computational capacity was about seven exaflops, which is a lot. Nvidia is now shipping a cabinet the size of your refrigerator that's one exaflop. Thousands of them. In less than a year, we went from an exaflop world to a zettaflop world. And there's no end in sight.

And this computation is different. Most of the computation in the world last year, the preponderance of computation, was still in people's pockets waiting for a Tinder notification. That's all that was going on, right? Not doing useful stuff. Now all those computers, every Nvidia chip made, gets put in a rack in a data center connected to every other one in the world, essentially, or at least thousands, if not tens or hundreds of thousands of them. So these things run 24/7. They're computing 24/7. That computation is not wasted. It's being used. So we are living in a vastly more computationally intensive world. And I think that is very important to understand because that's what gets you to a point where, okay, now what if we could model every neutron in a reactor core? What if we could model every neuron in your brain, every cell in your body? Those are things that were unfathomable before. And even now to some extent they might be, but you know, how many orders of magnitude would it take for that to be true? We went three orders of magnitude in the last year. How long will it take for us to go three more? At some point here, we actually can do all those things.

And once you can do those things, then the scientific method flips. Then your relationship between causation and correlation flips. Then instead of hypothesis-driven science, you just reverse-engineer control groups. Once you have all the data, you can just... and that's what big data is about. That's what we're doing now. We just get all the data and then we go looking through it to see what we find. And the computers are so powerful, they can look through so much data, they can find things that we never would have found or hypothesized in the first place. That's what's going on with drug discovery. That's what's going on with designing molecules. That's what's going on with hypersonic jets and stuff. So that's why I'm excited about it, and I think that it will advance R&D in ways that are very important, that we haven't been able to do very well for the last couple of decades. Sorry for rambling again. I'm just trying to make sure I explain things without leaving people behind.

Beatrice: There's a lot to say, I think, about it. So thanks for going deep. Is there anything... I think it's always interesting to also think about what we're maybe getting wrong. So, is there anything in terms of progress narratives or something that you think is kind of overrated right now or that you think people are really bullish about? I mean, the LLMs is clear, but is there anything else that we're maybe missing?

Pablos Holman: Yeah, I mean I think like I said, I think "scratch an itch" is overrated. I think that product-market fit is overrated. What's happening in venture is you're taking a mix of technical risk and market risk. So market risk is, will people pay for your app? Technical risk is, can you build it? But any app you could draw on a napkin, we can probably build. So in software, you're not really taking much technical risk, but you're taking this sort of infinite market risk because you don't know if someone's going to make a Zoom killer next week. In deep tech, we're doing the opposite most of the time. The day that first ship sails, I don't have any technical risk anymore. And I just have this vast industrial market. There's no market risk. If you can move containers around, you can get paid to do that for as many as you can find. So that's not really the problem. And I think that investors are sort of waking up to that now and it's helping us course-correct.

So there's a mythology there that's off. I think the mythology around the genius founder who invents some technology and then figures out how to patent it and figures out how to raise money and then figures out how to take it public and choose an HR plan and all that—we've been telling ourselves that story and it's a lie. What always wins is teams. And teams have a quarterback or a captain or a frontman or somebody everybody fixates on, but it's always teams. And a lot of startups in Silicon Valley, they're starting with... if you just look at what VCs are doing, by their own admission they'll tell you they're just looking for hustlers. They're looking for the best entrepreneurs. They're looking for somebody who's going to make money at all costs. And that's great, we need those people. But the problem is we're building these companies out of our best entrepreneurs with whatever they could invent. So, our big win in Silicon Valley is disrupting the taxi industry. What if I can get Travis from Uber and arm him with a nuclear reactor? Now, maybe we're saving the world. Like, I know that scares people, but what we need to do is find a way to get our best entrepreneurs paired up with the best technologies, our best inventions. And we're not trying very hard to do that right now. I think that's one of the big problems. To me, that's the biggest problem that I have because I'm on the other side. I have all these inventors and they're only starting a company because it's the next logical step for the thing they worked on their whole life. They didn't set out to be a business person. And so I need them to pair up with entrepreneurs. So I need the best entrepreneurs to come and let me arm you with a CTO and an invention and IP and a technology that can change the world. Let me arm you. That's what I want. So send me entrepreneurs. I'm ready to arm them up.

So those are the mythologies that I think are off and that matter. And then we could sort of talk about what's wrong with the AI conversation as well. But to me, those are the drivers that we need to change in Silicon Valley.

Beatrice: Do you see a lot of movement on this now? Like, are people actually... do you see more deep tech investment and do you see more interesting stuff happening? And do you have some players that you would recommend people look at?

Pablos Holman: What I think is that... I was in Silicon Valley in the 90s and it was pretty awesome and it was all nerds. And then these opportunists started showing up that had MBAs, and in some ways it started to suck for the nerds, but it turned out to be very important because it rounded out these teams. The nerds weren't figuring out how to scale the business. They weren't figuring out how to get enough customers. They weren't figuring out how to adequately finance things. And so they needed that help.

And I'm thinking deep tech is at that stage now. We're in the late 90s. We're armed. We have all these technologies. We have all these superpowers. But until the story around some successes permeates, we can't attract all the commercial animals, and we need them on the team. So I think the stories are there. You can see it in SpaceX and Tesla. It would be great if we had more diverse examples. I think this is another problem we have right now, is people are so pissed off about Elon that they're failing to learn what they should. He's showing us how to build modern industries. Why do we only have Elon examples? We need a thousand Elons is what we need. Maybe they don't all need X accounts, but we need a thousand. We need a lot more people who understand that we're going to rebuild industries with our software toolkit. And that's what he did to automotive, that's what he did to space launch, and that's what he's done to communications for that matter with Starlink.

So these are the ways to build. And the cool thing is a lot of engineers now have cycled through SpaceX and Tesla and learned what that looks like, the actual pace you have to move at, the actual way you have to build. So, I'm excited about that. And I think we're at a point where now you can... and we see that a lot. I see a lot of space startups started by former SpaceX engineers, and I don't know if we need that many space startups, but we definitely need the people who have learned how to build in that context and we need them to come build all the companies that we have coming. And so to me, that's another thing that people are blinded by. They're blinded by how pissed off they are at Elon for one thing or another. And it's keeping them from taking the good stuff. I think a good way to operate is just use your shopping instinct. Look around the world for the things that are working. Steal the best ideas. Copy. That's why Silicon Valley worked. You could rapidly steal the best ideas from each other and grow. And that's what we need to do. Steal the best ideas from Elon, show us how to do it better.

Beatrice: Perfect. One thing that I just want to make sure to ask you also, because this is the Existential Hope podcast, when you think about a vision of the future that makes you excited, could you just share with us a few elements of what that contains? What does the future that you find deeply exciting look like?

Pablos Holman: Yeah, I'll give you one example that sticks with me. I remember when my daughter was in school. She went to public school in Seattle, and a really good public school. You couldn't complain. Those public schools are probably as good as they get. But we would still complain anyway. And we would complain because she was in a class with 30 students and one teacher. And she wasn't the best kid, but she wasn't the worst kid. So, she's just kind of in the middle. And it was heartbreaking to see her come home not having learned anything, not being challenged, just another day at school. It's like babysitting. She's not living up to her potential. The school wasn't able to challenge her, wasn't able to meet her at her level. And we complain about that and some years get the student-teacher ratio to 28-to-1 instead of 30.

But meanwhile, at the exact same time—this is 15 years ago—I'm seeing the headlines in the tech industry were how self-driving trucks were going to put truck drivers out of business. And this is 15 years ago. And here we are in 2025, and nobody's lost their truck-driving job to a self-driving truck yet. It's getting closer, but it takes time. And so what I thought at the time is, man, it would be great to take one of those displaced self-driving truck drivers and give them one student. And if you had a one-to-one student-teacher ratio, I mean, you don't even really need a very good teacher in that case. If you have a personal relationship with somebody, you know them, you know where they're at, what would be a little too hard or a little too easy. You can put a challenge right in front of them. That's what a good teacher would do.

And we're at this point where I think we're just not being honest with ourselves that we have volition in this. We are building the future. We can choose what future we want. And we aren't being honest with ourselves. We say we care about kids, but we're spending all our money on people to drive trucks full of Happy Meal toys. If we actually cared about our kids, why aren't we just aiming ourselves at that?

So, I look forward to a world where our economy has shifted to the point where, yeah, the robots are doing all the dumb, the dangerous, the boring things nobody wants to do anyway. They're just doing them because they have to get a paycheck. And I look forward to a world where we're paying people to teach our kids, to take care of our elderly parents, to help. Like, I could use some help. I would like a world where we get to focus on building that sense of community, sense of family, sense of support.

I think a human is wired to feel fulfilled, to feel happiness, when we are feeling needed by other people. When we can see how other people need us and how we're useful, then people thrive. And I can connect a lot of dots. Like, I put a little money here, I help this guy out, he builds this ship, and then we do this and we do that and then we kill Maersk, and then 100 years from now the world doesn't burn fuel to ship anymore. To me, I can do that and I feel satisfied knowing that what I'm doing is going to help people 100 years from now. I'm fine with that. I mean, hopefully 10 or 20, but most people can't connect those dots.

You could see nurses maybe feel overwhelmed, but they certainly feel a sense of how they're useful because they get to see how they help somebody every day. And some people are in that position. So, if we could put more people in positions where they could see how their work is directly contributing and positive to someone else's life, that's going to be a better society. And if you look at these... we have all these arguments about whether people are being nostalgic for a world that they think was better 100 years ago when everyone was dying of dysentery and [...]. The world wasn't better, but we have a nostalgia for it because something was better. And I think the something that was better is that everybody had to work a lot harder, for sure, but they could all feel how they were needed. When you're a five-year-old kid born on a farm and you have to go help with the harvest, you feel needed. I mean, we call that child exploitation. We wouldn't let kids do that. Instead, we send our kids to schools where all the work they do, they watch get marked up and thrown in the trash. What could be more demotivating? You are not needed. You're a kid in school. No one needs you. You're a liability. You're an accessory. Of course they have mental illness and all kinds of depression problems because they don't see that the world does. My kid knows that I don't need her. She knows I could have had a Lamborghini instead. She's just a fashion accessory. That's not good for them. They need to feel needed. We all do.

And so, look, you've got to be very compassionate with people who lose their jobs or have a hard time with transitions. I get that. But on the other side of this, we should end up in a world where—going back, let me just say the last thing I'll say here—if you go back to energy the way we were talking about, if you look around the world, what are people fighting over? What are these wars about? They're fighting over access to energy, control of resources, mostly energy, right? All those wars in the Middle East, that was just geopolitical [...] so we could get the oil, right? So if we provide through leadership—if we're the country that's good at doing new things and we show, "Hey, here's how to make unlimited, clean, cheap energy with solar panels in space, with nuclear reactors, with even fusion or whatever"—if we come up with these things and we make them work for the world and everybody got as much energy as an American gets, then we live in a world where there's nothing to fight over anymore. We can fight over the next season of Love Island or something. We don't have to be fighting over access to oil in the ground. So, I think we are on the precipice of a much more beautiful and peaceful world. I know the transition is hard for people and it's difficult to embrace it, but I think people should. There's a lot of cool things you can do right now to help make the future happen faster and smoother. And that's what I'm excited about.

Beatrice: It reminds me of a concept from Eric Drexler, the co-founder of Foresight. He has the concept of Paretopia. Because this way if everyone has more energy and just in general there's not a lack of resources, everyone gets better all the time. There's just less incentive to argue with people about what they're having because you're also getting better and better.

Pablos Holman: And even if we do still persist with this kind of competitive mindset or tribalistic thinking or something, we should be competing on making better societies. And you could see even in our lifetime, it's sort of sad to see a lot of them have been damaged by, I don't know if it's just consumerism going around the world or whatever, but it's one of the exciting things about traveling around the world is you see... I love going to Sweden. Sweden is really good at some things we suck at. I love going to Japan. They're really good at things that we suck at. They have some beautiful things. There's some super [...] up [...] about Japan and things that I couldn't live like. But I love seeing that they've figured out some things we haven't. And there's some things that Americans are good at, but mostly things that we suck at. And you go to... I used to go to Ethiopia a bit and these are some wonderful people, very gentle. I remember being in the biggest open-air market in all of Africa, surrounded by thousands of Ethiopians, and it's quiet. I'm the noise here, the noisy American. A thousand Ethiopians and they're just quiet. I'm in New York City. I can't walk out my door without being bombarded by just noisy [...] everywhere.

Different societies are good at different things. I love that diversity. And I would love to be in a world where you could wake up in the morning and instead of thinking about getting to inbox zero, you'd be thinking about, "Man, who can I go see today? How can we make our community better? How can I get everybody together? How can we do something enriching and fulfilling?" I wanted to be enriching as a parent. I wanted to be the guy that played Legos with my kid all the time, or at least more than I did. I was busy working, and not just working, but doing dumb [...] I had to, I don't know, moving stuff around, cleaning stuff up, ordering more stuff, getting rid of the stuff I bought last week. I'm busy with stuff that robots could do and I could go play Legos with my kid. I think that would be cool.

So anyway, I'm rambling, but you get the idea. I think there's a lot of real potential for societies to focus on making themselves better once they're not... I mean, right now I think we could be doing it, but we're very busy watching Netflix season 14. And we could be using some of that time we spend on entertainment, some of that time we spend on screen time, to go teach a kid. Pick one. Look at your screen time on your phone, divide it by 10, and use one-tenth of it to go hang out with your niece or nephew or neighbor kid or anyone. If everybody just did that, we'd be fine. But so this is again, we want to blame the computers and blame the technology, but the truth is we haven't gotten honest with ourselves that it's up to us. We have the choice to make and we have the opportunity right under our noses to do what we think we value the most. And you vote with your time.

Beatrice: I think that's a great note to end on, and I just want to also leave people with your book as a recommendation. I think it was a really good read. It's called Deep Future: Creating Technology That Matters. It's available on Amazon. And yeah, it really I think makes the case that we could and should be aiming higher with the technologies that we develop.

Pablos Holman: Yeah, thank you. It's a fun book and I really tried to give people as much as I could about the problems, the technologies that could help us solve them, how we can bring them to life, how they can become a part of that. So yeah, I hope people... So far, people really love it, but I haven't found anybody who doesn't so far. So hopefully... Yeah. And I'll start doing podcasts in the afternoon when I'm less hangry and more caffeinated. So thanks for listening to me rant.

Beatrice: Thank you, Pablos.

Pablos Holman: Yeah. All right. Ciao.