Josh Wolfe of Lux Capital on Turning Sci-Fi into Sci-Fact | 2025 Upfront Summit

Alright, Josh Wolfe, thank you so much for being here.

Good to be with you, Nick.

Let's just get right into it. You've recently said that we once looked to science fiction for escape, and now we look to it for instruction. That our scientists and storytellers seem locked in a symbiotic exchange of ideas. We're sitting in the storytelling capital of the world. Tell us about the narrowing divide between sci-fi and sci-fact.

Yeah, so my riff is always that either our scientists are becoming more creative because they're compounding with the technologies and the instruments that they have, or our sci-fi authors are becoming less creative. I don't actually think that's necessarily true, but so many entrepreneurs, engineers, founders are inspired by the canon of science fiction, and whether that's Star Wars, Star Trek, graphic novels, comic books. And we actually go into the archives of that and try to trove and pull from it. If you take a few of our recent companies, it was literally X-Men, which everybody's seen either the movies or read the graphic novels or the comic books. When you have mutants that are literally shooting lasers out of their eyes or conjuring fire from their fingers and levitating invisibly, And we basically said, look, in reality there's 8 billion people on Earth, and if there's a 1 in a billion chance of some crazy phenotype, some actual trait that might be in a person, then there should be 8 people walking around that, for example, can drink alcohol without getting liver disease, or eat fatty foods without gaining weight, or walk in the Himalayas at the highest altitudes and not need oxygen, or plunge the depths of the ocean and be able to hold their breath for long periods of time, these sort of rare variants. And so we ended up starting a company called Variant Bio, and tried to find these outlier people that are in outlier parts of the world, with outlier traits. And most of the people that were ever sequenced by 23andMe or Genentech Roche were basically what we would call pale male stale white Europeans. But most of the really interesting genetics are co-sanguineous, people that are sort of genetic islands. And we started this company, I hired away the head of the Bill and Melinda Gates Foundation investment program, who was putting out $2 billion a year, he's like a mensch on a mission, and he took the helm, raised $150 million, and now has 17 partnerships all over the world, from the Maoris in New Zealand to the Faroe Islands, East Africa, Uganda, Pakistan, and 4 drugs going into the clinic. So totally inspired by sci-fi.

Wow. Have they figured out how to make all of us sleep only 4 hours and get more out of our days?

Sleep is actually one of them, and another one is people that live in a weird part of South America that have a genetic condition that raises their body temperature, so their basal metabolic rate. They evolved because the precipitous drops in temperature, and so the idea there is could you create some fat fat-burning drug that raises your metabolism while you're sleeping, and you burn sleep while you sleep. Most of it today is very practical. It's focused on two main indications for the four drugs, for liver and kidney. But that's, you know, one that literally came out of X-Men. And we like finding stuff like that. There's another one where we actually spun it out of Google, where I was obsessed with some graphic novels that were talking about digitizing smell. So if you think about, you know, our phones have eyes and ears, right? They've got cameras that can see and record and play back, and you've got microphones that can hear and play back, but you can't record smell. You think about Proust in "In Search of Lost Time." He bites into the madeleine and it harkens back to his childhood and all the memories that flood and flow and then the narrative story unravels. We should be able to walk into a café and smell, you know, what is that smell and be able to record and share it. You should be able to smell something and share the first shampoo that harkens back to your first kiss and the smell of your grandparents' home and a great meal you had or a terrible thing that's disgusting. And so we found this group out of Google and spun them out. It's called Osmo and they are digitizing smell. Straight out of science fiction.

I think another interesting part of your guys' investment process is the idea of risk gaming. You guys introduced this as like a novel approach to scenario plan decision making. Can you tell us a little bit more about that and how it plays into your investment process?

The idea of risk gaming is sort of like war gaming, but it was like, let's think about low probability, high magnitude consequences in pretty much everything. So from geopolitics to capital markets, policy, science, breakthroughs in technology, and we conjure these ideas and scenarios, and it's really our partner Danny Crichton, who used to run TechCrunch and then came to run content for us, he's a brilliant guy, polymath, And then we invite incredibly intelligent people, people who might be in policy, people who might be in geopolitics, ex-military, 4-star generals, brilliant technologists and inventors, and we assign them different roles. So you might have the former head of SOCOM that is actually playing a startup CEO, has no experience ever doing that, and a startup CEO that's playing like procurement for the Department of Defense. And we conjure all kinds of scenarios. It might have to do with the climate impact on shipbuilding on the Eastern Seaboard of the US. It might have to do with AI and elections. We have a new one that we just did yesterday on biotech. and, uh, some of the cutting-edge technologies that are happening globally, and whether you have open source or closed source AI, and how that's gonna affect the progress in biology. It helps us identify brilliant people, first of all. It helps us convene people, and they're very quick icebreakers for these people. And then it helps us conjure scenarios so that we can end up deploying capital along some of these themes. And so it's a really interesting, unique thing that we do and, um, puts together all kinds of fascinating minds.

Awesome. Some of the most experienced VCs have remarked that the pace of innovation in AI is unlike anything they've seen in their careers. I know you guys are doing a lot of work out there. How are all, how are you guys thinking about AI and what are you guys focused on these days?

I would say, you know, everybody's focused on AI. And so I'm always trying to think what's the consensus? What's the variant perception? What's the thing that other people aren't thinking of? A big thing was actually a hint that in a scientific paper that Apple of all people published about a year and maybe 15 months ago that talked about small models on device. And I think that you're gonna be a big shift from what Jensen and NVIDIA would like you to believe that you need large 100,000 H100 clusters to be able to do both training and inference. And you're starting to see a shift away for small models So that's number one, small models on device where it's mostly gonna be maybe 50% of your inference will be from your cached emails, chats, WhatsApp, photos, videos, health data, et cetera. And then 50% you're going to the cloud, whether it's with Perplexity or ChatGPT or any other web accessible thing that might be displacing Google, including Google itself. The second thing I would say, which is really important, is the memory players. I had an insight back in 2016 that I got very lucky about, which was NVIDIA, because we were invested in a company called Zoox that Amazon ended up buying. That was their self-driving cars. And they were using these Nvidia chips before other people were, and it was sort of an insight that, hey, you're gonna take something from traditional GPUs that are being used for PS4 and Xbox and you're gonna repurpose it for compute and logic. And I feel the same way right now about the memory players. Most people think that memory is commodity, DRAM, SRAM, and NAND. The 3 major memory players, you've got SK Hynix, you've got Samsung, both Korean, and then you've got Micron in the US. SK Hynix in particular is coupled with their high-bandwidth memory to pretty much every NVIDIA cluster that exists, but they're starting, particularly on the small mobile side, to play a very big role. So I think that they're gonna be absolutely ascendant to become a power player. And I think that shift from CPU to GPU to the memory players, and memory being used for inference is gonna be a big deal. The third thing I'd say is, you know, I always talk about this 5-year psychological bias that everybody wants to be invested today where they should have been 5 years ago. And your, our job is to try to think about where do people wanna be in 2, 3, 4, 5 years. I think the next wave for AI, and the thing that we've been investing over the past 2 years, having done a lot in text, video, image, audio, et cetera, is in three-dimensional, the physical world. And so that really is in biology and in robotics. In biology, we spun something out of Meta called Evolutionary Scale, which is developing foundation models for biology to be able to go from a prompt to a protein, actually making drugs. Physical intelligence, taking a team from Stanford and Google and being able to think not about the humanoid robots, of which I harbor a lot of skepticism, but the brains, the embodied intelligence, regardless of what the, the physical instantiation of that is gonna be. And I personally think it's gonna be way more like, I don't know, the cantina scene at Star Wars. Like, you know, as Darwin would have said, like endless forms most beautiful, super weird, not bipedal, two arms, you know, two legs. But the thing that you look at this and you're like, holy shit, like how did it do that? And it's just, you know, 7 different arms moving in weird constructs. And that to me is the future of robotics.

I wanna go back to the humanoid thing. I can't let you off the hook there. I think it was Vinod this morning said he expects billions of bipedal robots to be roaming the Earth. I think it was 2040. What do you make of that, and do you believe in the humanoid thing?

Well, I think if it's Elon and he's doing his Optimus robots, which those were, by the way, teleoperated, right? They weren't actually autonomously programmed and whatnot, except for the choreographed dance. He'll do it, and he'll have them in factories, and they'll be in, and so I think there's without a doubt a market for that. But Unitree is the Chinese company that's gonna dominate this. And they sort of took the Boston Dynamics baton and have run with it. They'll be far cheaper and whatnot. But I think that the real value is gonna come whenever you talk to the humanoid companies, and there's a dozen of them here domestically. It's like, we're gonna help with home work tasks for the aging demographics. And I just don't buy that. People can do dishes and fold your clothes better than that, but I think it's gonna be much more sophisticated tasks again, that break the mold. Even if you apply one of these mental models, like when radio first came, it was basically broadcasting plays, right? We just did what we did before. When TV came, it was broadcasting radio shows. And then, you know, I think that most people are like, why don't we just take a robot that can do sort of human labor? And it really should be just far more sophisticated things that are beyond what our capability is. Even if I was to lift this up, you know, I have to have the physical intelligence quite literally to know how much force do I use, an intuitive sense of physics to lift this up. And then when I unscrew this cap, I have to, I use the torque in my hand, and you see these humanoid robots that are sort of doing that. But the first logic, sort of principles, you would just want an end effector that goes on with some interesting weird contraption that just like quickly spins it off like a screwdriver and lifts it off, and then Swiss Army knife swaps it for something else. And so I just think we're gonna reimagine, really creative people that haven't come onto the scene yet are gonna reimagine what robots look like, and it's not gonna be like Westworld, you know, weird approximations of humans.

I wanna do a quick home court thing. So we're here in Southern California. I know you and Lux have been, investing in aerospace and defense companies for quite some time. Why is so much capital flowing into that space right now?

Well, the first is that I think there's a recognition that the world is not a safe place, and that there's bad actors, and that the best way to thwart bad actors is to have deterrence, and the best way to have deterrence is to have superior firepower and the ability to inflict violence on people. That's just a reality. It would be nice if we all sang kumbaya, and that's effectively what happened going back maybe 10, 12 years ago when Google had people marching out in protest of Project Maven, which some of our companies were involved in, which was taking aerial data from drones phones and various aircrafts and basically orthonormalizing it and putting it in. I actually thought that was a moral thing to do because you were able to distinguish between a bad person with an AK-47 and a good person with a pickaxe coming home from a farming job. But there are bad actors out there, and you had a population, particularly in Northern California, that was aghast at the idea of working for the military industrial complex. We have a peer adversary in China that is not only doing that, but they have an overt military-civil fusion policy so that all the companies are working as part of the government. So that's number one. Then you had a brave soul really in somebody like Palmer Luckey and Brian Schimpf and Matt Grimm and Trey Stephens. And we became early investors in Anduril, and they were unapologetic about this. There was no dancing around, "Hey, we're just gonna make things and it might be used for defense." It was explicit. We are going to build effectively Stark Industries. We're gonna build a war fighting company for America and our allies. People were skeptical. I remember being at the Reagan National Defense Forum when Palmer was first there, in his characteristic garb and his Hawaiian shirt and his cargo pants and his flip-flops, and people were sort of like laughing at him. And it was the classic Gandhi thing. First they laugh at you and then they eventually join you and then you win. Now they're winning the ascendancy of their valuation, the ascendancy of their performance fundamentals and the revenue and the contracts that they're winning are attracting a whole suite of entrepreneurs, all the way from the Gundo where you have this complex, but them, Varda, which just 6 hours ago was the third ever company in the world in history to return something from space. It was just astounding. And the whole team was gathered there at 5:30, 6:00 in the morning as it crashed down safely into Australia. From space. So them, Varda, Hadrian, Saildrone, just air, land, sea, space. You have brilliant engineers that are now saying, "We don't wanna work on advertising. We don't wanna work on, you know, BS e-commerce. We wanna do things that help make the world truly a safer place by making it a more dangerous place for our enemies." You recently added Brett McGurk to your team. Yes.

Why did Luxe add a diplomat to the team?

Well, we started with a 4-star general who ran SOCOM, which was Tony Thomas, some years ago, and had an incredible experience with him. And it opened our eyes. Brett was actually on the board of one of our companies before that faced the IC, the intelligence community, selling into them. We had a great experience. And then he went back into the administration. He was originally in the Bush administration, then went into Obama, then went into Trump. When Trump pulled out of Syria, he had quit because he had promised the Kurds military support. And then came back in for Biden, and then finished with the Israel-Gaza hostage negotiation deal that he played a very key senior role in. We just finished a trip in the Mideast, and it was absolutely eye-opening. And thanks to his relationships, there were people that I never would have met. And I am super bullish about some of these alliances between our firm and them, and our country and them. And yeah, having a diplomat that can open doors and think through a complex geopolitical landscape is important. We have begun to call this sci-tech diplomacy, the idea that some of the companies and the technologies and the scientific progress that they're making is of national import, not only to the U.S., but to our allies. And that is going to be the basis, as you're seeing, whether it's for data centers or for AI, open source versus closed, IP between biotech companies. It's a big deal.

Tell us more about the Middle East trip. How was that? What did you learn out there and why were you there?

I was there originally for a conference for a friend, Steve Drobny, who's actually a local guy here at Clocktower. He's a great guy. It was my first trip last year. I'm hugely bullish on the Emirates. You know, this is a true startup nation, sort of like Israel. 1948 for Israel, 1971 for the Emirates. They're wealthy, they're super savvy, super smart, very commercial, really modernized far beyond some of the other Arab countries. I would say like, you know, some of the Abraham Accord signees in particular. But they don't wanna just like participate in AI. Like they want to dominate and win, but in a way that is pro-social with their allies. They made an explicit decision to cast their lot with the United States over China. And that was, I think, a hard decision. But, um, I, I just think, uh, it's pretty incredible. So Sheikh Tanoun, who's really a key player, um, in, in, in the AI and technology and data center and energy and pretty much every facet, um, Abdullah bin Zayed, who's the brother of the leader as well and, uh, foreign minister, uh, was the assignee for the Abraham Accords and normalizing relations with Israel. These are extraordinary, intelligent, very sophisticated people that see the world as, you know, this sort of chessboard that it is. Saturday was very interesting, because we went from the Emirates to Saudi as guests of MBS, and he had his third annual National Day. The country was founded in 1727, but I think he recognizes the importance of not only having their Founders' Day, or rather the National Day, which is in September, but this Founders' Day, the third ever, because you need a national spirit. I think they all all don't like the Islamists. They all don't want religious dispersion. They want a national sensibility, because that's what creates stability in these countries. They are not democracies. Israel is the only democracy in the region. I don't think that they should be democracies. I think they are run by benevolent rulers that really understand what is best for the region, the people.

You talked about Israel, or you brought up Israel. Also, these countries are in what can be a tough neighborhood sometimes. Talk about what's going on out there, and I know we got a whole talk about it just right before this, but what's your perspective on what's happening out there? What's happening in technology out there? Where do you see things headed?

Hugely bullish on Israel. Obviously, I'm very vocal about it since the war, particularly on October 7th. I was raised Jewish. I'm atheist. And I always say that it wasn't October 7th in Israel, but it was October 8th in America that really radicalized me in a very pro-Zionist kind of way. We made our first two ever investments in Israel. Both of them are defense companies. One is sort of leaked, one is a little bit more stealth. But I truly believe that this is going to be the greatest generation in Israel, much like you had in World War II, the greatest generation in the United States. You have people that already were sophisticated engineers and computer scientists and material scientists and truly like Renaissance people that are dedicated to national service, a sense of community, and a sense of serious education. But now they are driven, I mean, absolutely driven by this existential threat. And at least for the next year or two, they're all cooperating with each other. It will surely devolve into competitive spirit as we do in Silicon Valley, but for now, it is just extraordinary seeing the companies that are being built. And frankly, us and probably 4 or 5 other venture funds that didn't have a presence there are spending a lot of time there. My partnership has made, you know, will be 3 trips in just the past few months. And, you know, we're going to have a very serious presence there. So very bullish on the region. Very bullish on our prior speaker. I've met with him before, had dinner with him. Absolutely brilliant. It would be the greatest peace dividend in the world to see a free and open Iran. So that is like super bullish. And, you know, I think you have a reasonable probability that in addition to Bahrain and UAE and already, you know, Jordan, Egypt, but you could see Syria, you know, coming in and joining the Abraham Accords. You could see Saudi normalization, you know, nearly imminent. And it would just absolutely transform the region economically, technologically, militarily. Under a security umbrella that would really thwart CRINC, which is China, Russia, Iran, North Korea, and would be a great thing.

Yeah, let's talk about China for a second since you brought it up. Venture capital sort of returning to its roots as we've talked about before. I think you and I agree that's a wonderful thing. There is a bit of a neoconservative bend to the technology companies that are being built in defense, and sort of the foil for that has been China most of the way. What are your views on the currency of global power as it relates to China and the technology that they're building?

Well, technology is increasingly becoming the currency of global power. It's, you know, this idea of, again, of sci-tech diplomacy. As it relates to China, I would love to see a détente. I would love to see leadership that was not doing sort of cunning, insidious things. I would love that TikTok was just a social media company like every other. I've been very involved behind the scenes. I think TikTok should exist. I think it should be divested from the control of the Chinese Communist Party. And there are very different aims, and they've been explicit about those aims, and they're published in documents. And you can see, again, this military-civil fusion and some of the undermining tactics that are being done.

How real is that? Like, the— all the sort of psyops that we hear about on TikTok and all that stuff, is that— is that like pretty—

I think it's very real in that they have found an effective vector to find seams of divisiveness in our society and rip them apart further. And I think, you know, in the Cold War, we never gave Russia a dedicated channel to give us propaganda and influence youth who might oppose military or certain policies we have. We certainly in World War II never gave Germany, you know, its own dedicated channel. 30% of Americans 18 to 34 effectively getting their news from TikTok. And we know vis-à-vis empirical comparisons to all of the other social media networks that they tip the scales in the favor of the things that they want, and forgetting about the things like not being able to talk about Uyghurs or Tiananmen Square or Xinjiang Province and those kinds of things. So, I think that that is a complicated portfolio, so to speak. And then militarily, we have great allies between AUSUK, Australia, US, UK, Philippines, and South Korea. And, if you talk to Sam Paparo, who runs INDOPACOM, and is responsible for that region, he would describe it that his mission is to create a hellscape, which I guess is a technical term, But he will describe, you know, the military surface area there. You've got 40 amphibious ships that you're up against. You've got 50 subs, you've got 200 surface combatants, 2,000 fighter jets, 3,000 unmanned systems, 750 satellites, and then 200,000 mechanized infantry. This is like the game board. And he wants to have a ton of attritable systems, drones of all types, you know, large, small, aerial. And it really is, despite the war on AI, and chips and all of this stuff that matter. It still is mass on target. It still is about projectiles and missiles and thwarting physical space. Now, all that said, my prediction is that there is no war for Taiwan. There is no physical kinetic, not a single bullet is fired. If I'm China and I play Go, not chess, not checkers, the long game that I would play is let TSMC diversify into the US so that you have a substantial hedge away from TSMC. Then the administration, now or future, might look and say, "Well, why is Taiwan in our national interest?" I mean, sure, democracy, freedom, et cetera, but I think Taiwan just gets Hong Kong sometime in the next few years, and the key for us is just ensuring that the semiconductor supply chain is redistributed. That would be the release in the pressure valve, which is effectively just let time take its course, and these forces will avoid a kinetic war.

Um, I think we have time for maybe one more question. I, there, there's a lot of optimism in tech right now. Advancements in AI, healthcare, energy production, a lot of the stuff we've talked about here. Let's end on a hopeful note. Um, I've seen some receipts on X of you being right about GPUs and some other mega trends from 8 years ago. You say most of what we see as crucial today won't matter in 10 years. What will matter in 10 years? In 2035, what do we have to look forward to?

Yeah, and I always say this because, um, when I was young, my My mother said, you know, you need to learn Japanese and golf, 'cause that was the rising power. You know, the past generation it's been, you need to learn Mandarin and coding. And I would argue that AI has gotten so good, like you probably in the next half generation don't need to know either of those things, 'cause you can have effectively Babelfish and code generation through Cognition and Cursor and others. I think that the most interesting thing that is not heavily funded yet, there's no bubble in it, but it feels like a directional inevitability, is automation of science itself. I believe that in the same way if you were in a band in the past 15, 20 years, you would go to a studio, you would rent time, you would rent instruments, you'd, you know, coordinate physical presence. All of a sudden you have Logic Pro Tools, GarageBand, you can be anywhere and do that. I think the same thing will happen with science, that you will effectively have robots running automated scientific labs. I will be able to be on the beach in the Bahamas, pull up an iPad, dial up an experiment with the same sort of time sequence that you would of a musical notation in a composition, The robots will run the experiment with perfect fidelity. You will solve the reproducibility crisis. You will probably have to engineer in some stochastic randomness because so many great scientific experiments happen and it's like, huh, that's funny, how did that happen? So you need some, you know, happy accidents. But then at the end of it, it will come up with the experiment and the result, and the result will then reverse prompt you and say, hey, there's a paper from 1972 and a paper from 1996. This might be a spurious correlation, but would you like to rerun the experiment changing the titration of this from 5 milliliters to 10 milliliters, introducing this chemical? And you're just sitting there on the beach and you say, "Yes," and it runs it again. You can't go faster than the speed of chemistry and reactions, and you can add catalysts. But I think that the countries and the companies that are able to basically automate science, which results in the production of knowledge, drugs, materials, new breakthroughs, are gonna absolutely rule the world. So that would be a thing that I think in 10 years is probably, for my kids, I would say, figure out how to get involved in the automation of science.

Awesome. This is great. I wish we had more time. We'll just have to have you back at some point. Thank you so much for being here. I know you have a million places you could be right now, but we really appreciate that you're here.

Thanks, Nick.

Yeah, thanks, Josh Wolfe.