Invest Like the Best with Patrick O'Shaughnessy
· Dan Sundheim
[00:00:00] S1: Back then it was like, are these businesses going to ever generate an economic return? I think one analogy was, “AI will be huge. So was air travel.” Airlines were not a good business. There's nothing differentiating about one airline versus the other, and so therefore the returns go down to the cost of capital. Obviously we took a different view, but that was like a 65-35, 70-30 degree of confidence in that. At that point, it was more about the skew. If things played out like we thought and the business models were actually moated, it would be huge. I think at this point we're in a different place in terms of the debate that’s important. The debate that's important now – if you want to look through a positive lens, which we do – you'd say that the businesses have taken slightly different lanes and have excelled at different things within AI. So, OpenAI has been great at consumer and has had good traction in enterprise. Anthropic has been incredibly successful at coding. There was a thesis when we first invested that APIs, or the business of having other software companies plug into you, or other developers plug into your model, would be commoditized. It would just be a race to the bottom. I think that debate is more or less irrelevant, because as you've just seen with Claude Code and even OpenAI's API business, these are durable businesses. And yes, can you switch? You can, the same way you could switch from AWS or Azure, but it's not worth it for a lot of businesses to do it. And there's sufficient differentiation among the models. If you look at the underlying margins of these companies, they are not the margins that you see in a commoditized industry. The gross margins are quite high. The competitive landscape, I think, is not heavily debated. At this point, you probably have four or five LLMs that will be relevant in the long term. I don't see that changing. Not that there's not sufficient talent out there, it's just that the capital required to get into this business is too great and these companies are too big at this point. And then you kind of get the snowball of – the more capital you have, the more compute, you get better researchers. I think it would be very difficult. So the competitive landscape is not really in question. I don't think anyone would say that these business models are commoditized.
· Dan Sundheim
I think the real debate is, these are extremely capital-intensive businesses, capital-intensive to a degree that we've never seen before in the history of business. And the question is, you're spending a ton of capital and the ultimate return on that capital is unknown. So it's not like a normal business who builds a factory and knows what they're going to sell. You are spending tons of capital to train a model. The question is, do the scaling laws work such that the returns on that capital continue to be attractive, which means that you'll be able to attract more capital and build better models? Or, are you going to get to a point where everyone looks back and says, “We raised too much money, we spent too much training models, we didn't get the economic return.” Or – I think equally likely, not more likely – people would say, ultimately you will get the economic return, but it just happened slower than you would've thought. Enterprise adoption just didn't take off as quickly as you thought. And therefore, the problem is when you are this capital-intensive as a business, it introduces financial leverage and operating leverage to a degree you don't see in normal businesses. So you don't have the luxury of two or three years of things going slower than you otherwise would expect. I think the scaling laws, the returns on capital, and the speed at which these tools and AI is adopted throughout the economy are the questions.
· Patrick O'Shaughnessy
[00:00:00] S0: Is there anything that Netflix or something like that could teach us? That's another business that comes to mind where there's a crazy amount of capital that was spent to build an asset, and then it gets amortized over a bigger and bigger user base. And that's turned out to be a great stock and one that I know you've owned a lot. Is there any analogy between those two that's interesting to you?
· Dan Sundheim
[00:00:00] S1: When I was speaking to the executives at the LLMs, the way I framed it is I said, “Look, I think your business is some kind of combination between Netflix and Spotify. Netflix in that, unlike other tech companies, you are spending a ton of money upfront to train these models. Once these models are trained, you go sell them at extremely high incremental margins. You don't know what the revenues are going to be from that fixed asset that you've built. But to the extent that you've built that asset, you want to sell it as much as possible, so that you can get the cash flows to build the next model, and so on and so forth. That's very similar to Netflix in that they invested in content. And when you're an early mover, this kind of fixed asset business – you invest heavily, you get the capital to invest heavily, you get the revenues, you spread it out over an increasing number of people. You invest more in that fixed asset, and that kind of has a flywheel effect of generating more revenue, more content, more revenue, more content. And eventually, you get to the point where it's almost impossible to compete because the first mover advantage is too great. The difference – if you were to say like, what is an important difference of Netflix versus these models – is Netflix's content was differentiated. The models are more similar than they are different. At any given time, OpenAI may have a better model, Anthropic may have a better model, but a lot of the expertise and the innovation gets disseminated pretty quickly. So, these models are not terribly different. And that's where the Spotify analogy comes in, in that I think if you are Google or you are OpenAI, the differentiating factor will not necessarily be that Google gives you a better answer. Like, if we were just like to query Gemini or ChatGPT on something, I don't think it's the case that we would say definitively one will give you a better answer over time. However, the personalization matters, and the first mover advantage is like – the more that these models know about you, how you live your life, your health, all the things that are important to you – you build up this data and history, and it becomes very sticky. The music on Spotify is no different than Apple Music or Amazon Music. Theoretically, it's a pure commodity. What makes Spotify have pricing power, what makes it differentiated?
· Dan Sundheim, Patrick O'Shaughnessy
Why would people be incredibly upset if you said you had to not use Spotify anymore? It's because it's personalized. It's because they've tailored the service to take a product which is a commodity and personalize it to the point where you're willing to pay a premium for that commodity. [00:00:00] S0: If you were giving advice to the executives at these companies and telling them what to lean into and what to look out for over the next five years, I'm curious what you would say. Because the scaling laws are so interesting in the sense that like, the models keep getting unbelievably better and that probably means the revenue available is like – who knows how big it could be, it could be the whole world. But the cost keeps going up by orders of magnitude. The Colossus 2 data centers, this unfathomably big thing, it’s like two gigawatts of power. It's crazy. What advice would you give them based on everything you've learned about these big, massive businesses?
· Dan Sundheim
[00:00:00] S1: The really interesting thing and challenging aspect of these businesses, the LLMs, is that the models they are building now, and especially in the future, can be applied to almost any aspect of the economy. You can take these models and you can make consumers' lives more efficient by having them be personal assistants. You could solve physics problems, you could help with drug discovery. You could make enterprises more efficient. The TAM is certainly not the problem. Focus is going to be a question mark. On the one hand, the more end markets you go after with a fixed asset, the better. You're spreading that cost over more end markets and having more revenue, which then can be reinvested. I think the flip side of that is that I rarely have seen any company succeed trying to go after multiple end markets at the same time. Usually, you have an A team. An A team is focused on one thing. Your culture as a company is oriented towards either consumer or enterprise. Even Amazon, which you'd say is an example of a consumer company that got into enterprise. They got into it like, seven years later, even after they went public. So trying to do everything at once is tempting, because if you're successful, you're effectively just amortizing that fixed asset over more revenue streams. At the same time, you risk not being the best at any one thing. So that is the trade-off. And I think that I'm not sure we have the final answer right now. The market has gone through periods where they thought Anthropic was Lyft and OpenAI was Uber. And up until recently the sentiment on OpenAI was more negative. I think OpenAI is taking the strategy of, “Let's do everything.” We are gonna go after Apple hardware, we're gonna go after robotics, we're gonna go after enterprise, consumer, science. They've been very successful in a lot of ways, but that's hard. I'm sure there are companies I'm not thinking of, but I can't think of many examples where that's been as successful. I understand the temptation to do it, and obviously the difference versus history is that the smartest people in the world are all going to work at these companies. So if anyone's going to pull it off, they will.