Joe Lonsdale: Why AI + the Heartland Will Spark the Next Industrial Revolution

Sure. Well, I guess the background. So I grew up in Silicon Valley. A lot of my friends and I were math and chess competitors. Not very popular in the social scene, but we had a lot of smart friends. And when I was at Stanford, computer science, I was at PayPal because I saw a lot of the people I admired going to PayPal. They rejected me my first year, but they let me in eventually. And the big challenge at PayPal, you guys might have heard of, was that the Russian and Chinese mafia were stealing all of our money. So we had about eight big competitors. And if you want to send a payment online, if we ask you too much about yourself, you don't want to use PayPal. It's annoying. We don't ask you enough about yourself, then you can go to 7-Eleven here and the person behind the counter might be in a bad mood and they could write down your number and they could sell it to the Russians online. And the Russians would aggregate thousands of these things, run them through our system, take out the money, and then you would get something saying, you know, PayPal, $200 charge. You said, I didn't send $200 through PayPal. You charge it back, the credit card company refunds you, takes it from us. So we were losing seven or eight million a month. And so PayPal was very unprofitable at the time, and we had to figure out how to stop the bad guys. So to make a long story short, AI wasn't quite what it was back then. We still did a lot of AI, but we also built these systems. I think we had a team in Omaha actually that we made our investigators, a lot of our best, our smartest customer support people. We give them systems and we built this technology to investigate, cut down the fraud just enough to make it profitable. We were able to sell to eBay as a win. And at the time, I got to know a bunch of these Secret Service and FBI guys because they were the ones who had to arrest people for us. And this was a whole new world to them. This is like 2000. So they'd never actually done cyber crime or arrested people online. So it's kind of fun, like all of us kind of, you know, cops and robbers or James Bond or whatever you want to think about it. It was fun to catch the bad guys. So we started advising them and helping them. And then 9/11 happened and we watched as the US government spent like billions of dollars on very backward stuff that was way behind what Silicon Valley was doing. And we realized, wow, there's like a duty here for our nation. We need to take the best and brightest of Silicon Valley and upgrade the investigative systems of our country. And so that was the initial impetus. We're going to take a functional tech culture, build these systems. And to make a long story short, it turns out the types of problems we had to solve to build these analytical systems and information systems, you actually had to map out all the data in a way to understand, we called it a dynamic ontology, understand how the data could talk to each other and you had to map out workflows. And so we had all these ontologies of workflows. And then it turns out by doing that, not only was it useful for government stuff, it was useful for a lot of companies to take their data and their workflows. And the types of problems we solved made it very easy to then put AI on top of it when AI came along. And so Palantir was probably already going, it was probably already a $20 to $25 billion company. It might be a $30 billion company today except maybe a little better, but the AI came along and it turns out that it's so good at helping people use AI that that's given it this 10x over the last 5 years. And it's very impressive. I'm not running it anymore. It's very impressive to see a lot of people I built it with and hired are still running it today. I hired most of the first 200 people and I'm really proud of the team. We were ranked number one in Silicon Valley for talent. I think that's also a big theme right now. As it turns out, these really strong engineering cultures with the really top talent have been very good at deploying AI very quickly in these contexts and so they've made the best of that. And one more thing I'll say because you asked about the valuation. When we started the company, we would hire this talent and I had a little chart. I think it's a good thing to do with top talent. You say, listen, you know, we want you to have a big piece of the company. We're going to have a trade-off. We prefer people have lower salaries that they could get, but they're taking a piece of the company, and here's three options. You know, more equity or less salary, how much you need, and here's what your options might be worth if we do different amounts. So you say, okay, with this much dilution, if we're worth $2 billion, here's what your shares are worth. So you have a table, and then the bottom of the table, I think the highest one I would do the first couple years was, here's what your shares would be worth if we get to $5 billion. And people are like, Joe, you can't say it's worth $5 billion. That's unrealistic. So it's very satisfying for me today after being shamed for showing them the $5 billion outcome that it turns out it's possible to make these things get pretty big.

Well, now they're all too rich to work anymore, which is annoying. So now you need new friends to work with. Come on, guys. I have them teaching at my new university. They'll help with that, but they won't work anymore.

Yeah, we did Palantir. It was very counterintuitive. You weren't supposed to do defense back then. People thought, why aren't you doing social media? That was the big thing you were supposed to be. They were very annoyed I was stealing talent from social media. The thing about defense that's interesting, I was just with a team here. We're backing Stack Health and healthcare as well. I was just with Alex Fromire who's a great entrepreneur here. The thing about those industries, for those who know him.

We're helping him build it together and backing him and a really strong team already. But I think those industries are both similar actually in the sense that they both have lots of really weird distortions and regulatory barriers so that the actual ultimate product is broken, right? It's like there's possibilities to just be way better than what's there because it's set up in a way that it's kind of kept people out. It's like the animals that evolved on the Galapagos Islands. If you connected them to the mainland, the mainland animals would go and crush them, but they can't get there. So if you can actually get into the space with talent, you could be so much better than what's there because of the artificial barriers. And it's very satisfying for me because you're just fixing things that are so broken. It doesn't mean you shouldn't be great to do it, but it takes a lot of persistence sometimes to break in and to figure out how to do that. So as an entrepreneur, I like solving problems and fixing things that are broken. So it's just a very broken area. And then on top of that, of course, he built Palantir, Elon was building SpaceX at the time after PayPal. He's an old friend. And those were the two companies that kind of broke through and became big companies despite the primes trying to lock everyone down. You guys know what the primes are. People know defense in this crowd.

So basically what happened is America had the most competitive and advanced defense ecosystem in the world. Probably made 70 or 80 big companies and we were just dominant. And then the Cold War ended and the budget was going down. So they kind of merged these 80 companies together to become about seven primes really. So there's seven big companies that are effectively a merger of all these other companies. And then these things became these very bureaucratic arms of the government and everything went through them. And they just spent a lot of their effort to make sure they just locked it down and no one else could play because this is their realm to own those profits. And that was the 90s and then the Silicon Valley tech bubble got going and there's a lot of great things happening. And so a lot of the computer science talent left because it was more interesting to be in Silicon Valley. And we've seen the last 20 years, the best engineering talent has not really gone on to the primes. It's mostly about them locking things down and protecting their area, not about them actually truly innovating as much as they should be. And so Palantir was able to do things much much better and eventually break through. Both Palantir and SpaceX actually had to sue the government, which sounds crazy. I don't like suing anyone. But the government was illegally working with them to block things that were clearly 10 times better and clearly hiding data. So in both cases it was like a scandal. They had to, and you've seen how SpaceX has done now of course, right? This is clearly cheaper, clearly better. So we had a suit to break through. Because of that, you now have Anduril get going. And then what happened, the reason we got Anduril started, three people who started and used to work for me at Palantir and it was my friend Palmer. And I think the reason it got going is a) there's a big gap where it's broken and needed to be done, and b) a lot of us realized, wait a second, Xi Jinping is actually like a crazy communist who actually does want to be a bad guy and confront us. China's not going to open up and be our friend. And he was making a lot of our friends who he knew from the tech world have their best engineers work for the PLA, work for building new weapons. So we said, wow, this is scary. We have a bad guy building really advanced weaponry. We have our space not hiring any of our best engineering people we know. We better get back into this and actually fix this. And so that's what Anduril did. And then that was kind of the mission to then start a few more based on possibilities we saw that were necessary.

I definitely say I'm betting on being able to achieve peace is the most important thing. 20% of what I've invested is in defense and it's true that when there's war in the world you make money on that, which is a very dangerous incentive. But I do think, I spent a lot of time at the Department of War just yesterday with much of my friends are very involved there now. But I think it is true, si vis pacem, para bellum. If you want peace, prepare for war. The way to not have war with China is to be so dominant and so great and have our stuff be so good that they're afraid of us. They know they'll lose face. China does not want to lose face, they do not want to lose battles. If they keep advancing and we don't advance, that would lead to war being much more likely. And so what's necessary is that we do stay ahead of them in order for them to be scared of us. I want to terrify our enemies and I want them not to want to fight with us. And so that's the goal of what we're doing as leaders in defense. That said, obviously these are companies and you want to make them profitable. Well, I think the way we do that is we outcompete the old broken ways of doing things and we convince DC, which means convincing the House Armed Services Committee, the Senate Armed Services Committee, the Generals, the Admirals, the Administration, the National Security Apparatus, that these new things are better. And we've gone a long way towards doing that the last 5 years. And the key thing is we're just trying to get them to give us fair competitions. It used to be you'd have a competition, yours would be way better, and you still wouldn't get in for the next eight years. We're trying to shift it so there's a competition. If you win, you get the funding, not the big giant company that sucks that you're doing things 10 times better than. That's all we're asking. And I do think it's moving that way. So I'm very positive on this current Department of War. It's been going the right way for a while now.

I love it. I mean, this is America's heartland for building these things. A lot of our companies are coming here and are partnering with you and building. So I think supporting these companies, coming up with ways that if you see things that are suboptimal, how can we do better? That's really what entrepreneurship is to me. You see something, you say, wait a second, you've been doing this this way for 30 years, but now based on this new thing I see, we could do this much better. We could have a better result, a less expensive result. I think that's all of our job if you're smart and you have opinions on things and you figure it out and you talk to other friends and you go and build and fix it. And I think a lot of us building these companies who are working here, if you see things we're doing that are not optimal, let's talk about it. Let's fix it. Let's build better things.

There's a big wave.

I think it's different for a new company or a new thing versus an existing legacy company, right? I think sometimes you're doing the new company with this AI thing because you think it's so much better and that becomes a big driving point of the founding of the company. I'm involved in dozens of companies, unfortunately the majority of them are still in California which really annoys me, that are growing very very quickly creating new ways of using artificial intelligence. I think it's a, there's 5 trillion dollars of wages right now in America in the service economy and I think like 2 to 2.5 trillion, 40 to 50%, are in areas where we've already seen you could at least double the productivity if not triple or quadruple. So this is going to affect everything. This is a huge productivity wave, may even end up being as big as the second industrial revolution. If you think about from 1870 to 1900, the average person in the working class or middle class, it wasn't really middle class, the average person in the working class was more than twice as wealthy in real terms by the end of that one generation just because of that industrial revolution. We may be lucky enough to get similar productivity growth right now. I think this is how you get that golden age that we're talking about in DC. But in terms of what you should be doing in legacy companies, which I think is the question for this audience, I think the goal is you focus on pain points and focus on solutions. Don't start with the technology. Don't start with a shiny object. That's a terrible idea. Start with your actual business. And maybe it's quoting, maybe it's invoicing, maybe it's scheduling, customer support, but pick these areas that are pain points that have friction where it looks like AI can do better. I think use off-the-shelf. Do not try to build your own tech team. That's crazy. Like maybe eventually you do that, but if you're a legacy business, it's not your business to create new AI things. There's tons of off-the-shelf things to use. I think your goal should be to augment your workers. I think the framework is that the workers should have to spend less time doing boring repeatable tasks and more time doing higher value work. So I think that's what AI's job is. You pick these areas where there's friction and you automate the boring parts and let the worker have more time for higher value work or for other things you could use them for. A lot of it is the paperwork layer, right? A lot is invoicing and compliance and all this stuff. I think that's where we're starting with this paperwork layer. And then I think what you want to do is you want to pilot it. You want to choose one area, have a tiger team, have a small group, choose one small area, prove the ROI, get that right, and then you scale it. Another thing that we have spent a lot of time on, if you're choosing an area you want an area with good data hygiene. So sometimes before you can use AI, you actually have to digitize your business because in many of these cases, it may not have been worth it to digitize before. It may now be worth it to digitize that area of your business in order to have it be on the computer, have it be there so you can tiger team it and go after it and apply a tool. But those are some of the ideas I'd use. But again, you're starting with the pain points, don't start with the technology is my bias anyway.

Well, Palantir has done a lot obviously to power Anduril and all our defense manufacturers for large manufacturers. I think these old ERP systems didn't really have the intelligence to see and map out and if one thing changes what other things change. Palantir is very good at that. I think it's probably for a larger company though. I'm not sure, it depends. You have to be pretty big for it to be worth it. I think they're trying to get better so it's even easier for smaller companies. But I think for any company where there's the paperwork stuff or where there's the different kind of friction on different very defined workflows, it's not that hard to just ROI, you know, do AI in those areas with off-the-shelf things. There's so many super fast growing off-the-shelf companies right now. You can kind of test them for these areas, see which one makes sense.

Yeah.

It's a very funny thing. To be accused of putting everyone out of a job at the same time that we need to help companies actually get everything done, right? It's a very funny dichotomy.

100%. We need better training programs in our country. I think the thing we're lacking is frankly culture and agency and people who want to work hard. I think if you want to work hard in America, it's not that hard. But at the same time, it's our job to make it even easier. And when you do these training programs, most of what the government does ends up being very sloppy. I think if you make them accountable where you only fund them based on their outcomes, we've seen that works a hell of a lot better. Because in Texas, we changed all of our high-end vocational schools to be funded based on the salaries coming out, and the salaries doubled over the next seven years because they were fully accountable to that. And all of a sudden the board's saying, crap, what are we going to do? Let's better figure out what skills to teach, what businesses to partner with. So I was actually just with my friend VC over here who I hear is very likely to be the next governor here, and we're very aligned on these types of things. I think there's a lot of stuff like that we can push forward to align it. So yeah, I think for AI to work, it should be about the American worker and helping them. But like you said, for me what this does is it raises productivity. We don't have enough people to do things. It lets you do things you couldn't do otherwise, right? So I'll give one example. We have a company, Bedrock Robotics, that I'm really excited about. These guys helped build Waymo. You guys don't know Waymo here yet, do you? Not yet.

But it's working obviously really well. So these guys are now applying that to excavator machines. And it turns out that we've got excavator machines working as if they're driven by a professional with five or ten years of experience with no one in them. And we're using them on a few projects. It's really cool. I'm really excited about using them on quarries by the way, if you think about that, because we can never find enough labor for quarries and you can operate them 24/7 now. But what this does is it brings down the cost of that part of the construction, which actually creates more construction projects because a lot of things don't quite pencil and now they do pencil. And as you get more and more AI, you're going to have more economic activity and you're going to have the few people who are involved each able to create more value by being involved in these things. So overall, it does make each of us and each of the people working for us able to do more and able to earn more, and that's the whole point. Right? I mean, there used to be, you look at people's last names. There used to be people who were Coopers and who were Smiths and who were Bowers and who were all sorts of jobs we don't have anymore because we changed our economy with the industrial revolution. But if we did still have those jobs, those were jobs that paid $5,000 a year. You don't want those jobs in the economy, right? So it's something very similar here.

In five, ten years, you know how to train people to learn how to learn, right? I think it's what our education system needs to do better, by the way. So there's lots of different new attempts at education. I'm very bullish on Joe Lonsdale and Alpha School. I have actually a podcast coming out with him next week. He's one of the most successful tech guys in the nation, and he's made tens of billions of dollars, and he's spending billions of dollars now. He's going to open thousands of schools around the country. And it turns out in two hours a day, you can use these AI systems to learn to figure out what you don't know, teach you, and push you ahead. And that's enough for kids to get in the top percentile. They can use the rest of the time for sports or life skills or whatever else parents want. And a big part of how his school works is you're teaching the kids to learn how to learn. So it's not as much about teachers at a certain point because you actually learn much faster with the systems and with the kids asking questions and helping each other, which is pretty amazing, actually. So I think that framework is going to be a big shift for everyone. You have to learn how to learn. You can't just, you have to learn agency, right? You have to learn to use yourself. I mean with AI, we're each going to be able to build our own worlds, right? This AI is going to be like if you want to start a company, you're going to have your ChatGPT CFO, your ChatGPT CRO. You're going to have all these things around you. We're going to need far fewer people working with the agents. You can do a lot more. So it's all about what is the agency? What are you doing?

Six. I have six young kids. So as you're thinking about raising them, I'm...