Invest Like the Best with Patrick O'Shaughnessy
· David George
[00:00:00] S1: I think it's hard to describe what it'll be, I think it'll be some form of an affiliate thing that happens, it's like a new native thing. The thing I always say to people is, again, we got to be humble in how we think about this, we never would've predicted what a feed-based advertisement is, no one would've known what that is because we didn't even know what the feed-based product was. It turns out it's probably the best advertisement format in history, it's really, really compelling, so it's not surprising that it monetizes really high, and people actually really like it, and really like Instagram ads. So, a year ago, this light bulb went off for me, maybe it's six months ago, I did deep research on, you'll probably relate to this, a new baseball bat for my son. He's nine years old, and it's pretty complicated. It needs to be a certain length, and drop, and all these certain specifications, and there's this year's version and last year's version, and if I had to do that on Google, it would be a total mess. I would struggle with it. Amazon, no chance because of the ads, Deep Research was really, really, really good at it, and it solved my problem for me. So, light bulb went off for me at that moment, one, the models are going to get so much better, and two, to me, it's an execution problem of building the capabilities to go execute that stuff on your behalf on the web. So, I think that's a really exciting future, there's going to need to be tons of guardrails built into it, you got to build a ton of product and piping to do so, it's really hard. Instagram famously tried to do shopping natively, it's just too hard. But I think that's a pretty exciting future, and shopping is just one category. So, if I take a step back and I think about AI today, really active users spend almost 30 minutes a day in the products. For context, users spend 50 minutes a day on Instagram, 70 minutes a day on TikTok, they're monetizing only a slight few of them today. Consumers get a ton of value, there's going to be a ton of consumer surplus available, and I think that could lend itself to the creation of a huge company, a massive company. And again, I think ChatGPT's in the lead today, but it's early.
· Patrick O'Shaughnessy
[00:00:00] S0: In that specific area of the world, this pure AI part of the world, where do you feel the most different than your peers in what you think matters, what you think is exciting or not exciting, worries you have, where do you feel most divergent from your friends?
· David George
[00:00:00] S1: I feel like I'm probably reasonably consensus on the excitement on the consumer side, I can put it into context around this upside around price that you get on the P x Q, especially if time spent continues to go up, which I think it will as the models get better and they have memory and things like that. I think on the enterprise side, one of the lessons I learned from SaaS and cloud, which by the way, the advancements of SaaS and cloud are tiny compared to the advancements of what AI is going to do, is I think maybe a little bit more expansively on what the companies can become on the enterprise side, but maybe I'm slightly more skeptical about what their ultimate business models will be. So, one of the really fun topics that people debate with high degrees of confidence that I have very low confidence in is what is the ultimate business models of these companies, and people put up these super compelling slides that are like, hey, the whole software industry is only $400 billion, but look at how big white collar labor is, and we're going to go get a ton of that. And to me that's a little bit hand-wavy, so there's a couple of areas where the business model has progressed in a compelling way to go tackle that directly. So, customer support is one, but because there's a very discreet task with very simple completion analysis that you can do, it's simple to price it on that. You can shift the business model from a seed-based thing for Zendesk or something to a new business model where if you successfully complete the task, you can charge on that. Maybe the next furthest developed area is coding, but it's not completion of a task, it's consumption-driven, and especially in the developer world, that whole world is used to paying things on consumption, it's how it has all shifted over the last 10 years, everything else I think is pretty TBD. It's going to be very hard. And I think when you see major technological shifts, it's very tempting to say, oh my gosh, there is so much economic value that all these companies are going to capture top down. The reality of doing it is much harder. And I always say to people, 90% of the technological surplus is going to go to the end users.
· David George, Patrick O'Shaughnessy
Just start with that as the assumption, whether it's consumer, whether it's enterprise, a funny analogy that I heard from somebody else is how is the steam engine ultimately priced? It wasn't priced based on replacing 50 laborers, competitive forces drove it to a certain price where there was an appropriate return on capital, but the vast majority of those productivity gains went to the end users of those machines, not the maker of the machines. So, I think something similar will probably happen in the enterprise, even with that, you can create the biggest businesses in the world. So, an analogy would be Apple, what would you pay for your iPhone? [00:00:00] S0: A lot more than I do. [00:00:00] S1: The sky's the limit. 90% consumer surplus is probably low, if the iPhone costs $1000 or something like that. So, I'd say the same for Google, I'd say the same for Facebook, it's going to happen in consumer, consumers are going to be the ones who realize the surplus. The same is going to happen in business, but I think the next generation of business companies can still be much bigger than the previous generation of companies given the capability gains.
· Patrick O'Shaughnessy, David George
[00:00:00] S0: When I last ran into you a couple years ago in person in San Francisco, we were talking about Waymo, and you were in the mode of intensely studying that company and thinking about it, which makes me very interested in this class of companies where you heard about Waymo, and self-driving as a service for a really, really long time with nothing happening, and then all of a sudden, the last time I was in San Francisco a couple weeks ago, it's just every other car, and the explosive nature of Waymo, as an example, is really cool to watch. There's all these other technologies, you might call them American dynamism and Andreessen Horowitz, whether that's robotics or small modular reactors, really exciting big technology ideas, which you understand the potential. Like if we had an in-home robot, that'd be awesome, but it's really hard to figure out how long it will take, maybe similar to how long Waymo took or something. How do you think about investing in those kinds of companies, where it's incredibly exciting, clearly if we had it and it worked, it would be really valuable, but it's really hard to know how long it's going to take to work? [00:00:00] S1: Often these are the ones that are the biggest market opportunity. Robotics is the biggest market opportunity, we are all obsessed with LLMs and diffusion models. [00:00:00] S0: Yeah, if you know it was going to work in five years, you'd put all your money in.
· Patrick O'Shaughnessy, David George
[00:00:00] S0: Yeah, if you know it was going to work in five years, you'd put all your money in. [00:00:00] S1: You'd put all your money into it. I happen to think it will take a little bit longer, part of that is informed by my experience with Waymo. I'd contrast maybe what Waymo does and increasingly Tesla and some others with what a robot needs to do, and it's very different. A car needs to basically stay in a lane, avoid anomalies, collisions, go a certain speed limit, find places to park... [00:00:00] S0: It's not simple- [00:00:00] S1: When I describe it that way, it's much more complicated than that, but simply put, that's what it has to do. I contrast that with what a robot has to do. What does a robot have to do in your home? [00:00:00] S0: A lot more degrees of freedom. [00:00:00] S1: Endless degrees of freedom. Make a cup of coffee, go do my laundry, but it took Waymo 10 years, and if you go back to the DARPA challenge, the whole industry, decades to get this point, two decades to get to this point roughly. So, my expectation is technology is advanced, obviously the generative AI techniques can be applied to robotics to help it go much faster, but I think it's going to take a long time. [00:00:00] S0: So, how do you invest in that?