The Exchanges

Q Barry Allison yesterday said, How much do you think it is to the audience? How much do you think it is? I don't know if you saw this clip. And then he goes, a hundred billion just to play the game, just to enter. Do you think that's right?

A I've been surprised at how, uh, the scaling has continued, how just throwing more compute, more data has improved these models. I don't know where that ultimately gets to, but I do think a hundred billion, there are not a lot of companies out there that can spend a hundred billion. What I do believe is you don't need 10 foundation models out there. I think you need certainly more than one just for the sake of humanity. And I think China will have one. I think the Western world will have at least one. So maybe there's three to five foundation models in the world. It's not going to be 10. It's not going to be 20. Uh, and so I do think that the pool of spend is going to shrink in terms of the number of companies, but the amount that each individual company is going to spend is also going to go up. So for us, I mean, there is real over investment risk, I think.

AI assessment note: “I do think a hundred billion, there are not a lot of companies”

Q Um, he said that you then showed this specifically nose for longs and had this unique ability to make the transition between the two. How did you make the transition so successfully between the shorts to longs? And how do you assess that today?

A The way we short is a bit different from how other hedge funds do it. So we are not looking for frauds. We're not looking for valuation arbitrage. We're looking to further express a disruptive thematic viewpoint that we hold on the long side, albeit on the short side. So it's really important to see these trends early. And so the view into the private ecosystem is actually quite valuable for that. So for example, If you hold a positive view on SpaceX and Starlink, what does that mean for other satellite businesses? What does that mean for rural telcos? If you are bullish on AI, what does that mean about call centers? So our shoring is really expressing further conviction in the longs, albeit in the other direction.

AI assessment note: “our shoring is really expressing further conviction in the longs, albeit in the other direction.”

Q How does the rest of the business do it then? I'm sorry. I'm just, I don't understand that.

A So I think that's why most hedge funds fail. It is really hard to go build a, a real business. If you are one of the large hedge funds like Citadel, you have strong LP relationships, you have lockups, you've, you've got an incredible long-term track record. That Makes it a lot easier when you have periods that are not up to your expectations. But if you are a One hundred million dollar hedge fund today. You don't know if you're going to be in business in a year. It's hard to go recruit. It's hard to go spend money and say, we're going to go build out a data science team. And you don't know how much time do I spend on recruiting? How much time do I spend on management, business building LP relationships? And that's before you get to the investing, right? So I, I think it's very hard. I have a lot of respect for, for folks who Are trying to do that. It is not easy. We obviously had to do it. And look, there's, there's a bit of, I think for us, one of the things that was most helpful actually in going through that period is that Sequoia does a great job of injecting Sequoia DNA while also allowing these individual businesses to grow up in a way that fits their specific area. So, uh, the Doug and Michael, and I give them a lot of credit for this, the way they set up the various businesses. So SCG, Sequoia, China, now Hong San, Sequoia, India, now peak 15 and Sequoia heritage was to…

AI assessment note: “So I think that's why most hedge funds fail.”

Q And so you're saying the weakness is you can think like a venture investor?

A Well, we get the benefit, right, of seeing things very early on. We get really excited about AI, how it's going to take over the world. Maybe we put on a short that it's just too early, right? It's going to take time for some of these things to actually happen. There are areas within AI where I do think it's very happening very fast. I think there is going to be massive disruption, for example, in call centers, and I don't think that's going to take a lot of time. There are other areas in AI. Now there's this, there's this narrative Uh, that I hear now that software companies are going away, right? AI is going to write all, all software. And I just think that's a naive view. It may happen, you know, over the longterm, but.

AI assessment note: “Maybe we put on a short that it's just too early, right?”

Q If you think about Google on Amazon, you've got Google Cloud, Cash Cow, App AWS, Cash Cow, and then Zuck is like, oh, Instagram is the cash cow, actually. How does that change how he acts? Cause Google and Amazon do also have distribution.

A They have distribution. So on Google, I'd argue GCP is not the cash cow, right? It's, it's search. And with search, I'm not sure. I I'm not sure how AI is going to impact search where it is today. AI is really good at taking a low signal data and translating into high value output. So in the case of meta, taking what you're looking at on Instagram and figuring out, okay, you want to go Buy this type of product. That is incredibly valuable. Google's got an incredibly valuable machine in that you type into the Google search bar exactly what you want, right? You type in skiing in Park City. Well, guess what? I know you, you want to go skiing in Park City. Your intentionality is a hundred percent known at that point for me. Whereas I think for Meta, I think the value of AI is, at least on the ad matching side, I look at a bunch of What you are, uh, browsing on Instagram and I say, okay, Harry wants to go skiing. He hasn't decided yet that he wants to go to park city. Maybe he wants to go to Whistler. How do I then move you in, you know, that path to get you to a different location, right? Or maybe you're even earlier on in the funnel and you say, I just, Harry wants to go on vacation and maybe you haven't even decided on skiing. Maybe you want to go on, uh, you want to go to the beach. So I think AI has a lot of value for that. I'm not sure as much for Google. How do you influence,…

AI assessment note: “on Google, I'd argue GCP is not the cash cow, right? It's, it's search.”

Q Barry Allison yesterday said, How much do you think it is to the audience? How much do you think it is? I don't know if you saw this clip. And then he goes, a hundred billion just to play the game, just to enter. Do you think that's right?

A I've been surprised at how, uh, the scaling has continued, how just throwing more compute, more data has improved these models. I don't know where that ultimately gets to, but I do think a hundred billion, there are not a lot of companies out there that can spend a hundred billion. What I do believe is you don't need 10 foundation models out there. I think you need certainly more than one just for the sake of humanity. And I think China will have one. I think the Western world will have at least one. So maybe there's three to five foundation models in the world. It's not going to be 10. It's not going to be 20. Uh, and so I do think that the pool of spend is going to shrink in terms of the number of companies, but the amount that each individual company is going to spend is also going to go up. So for us, I mean, there is real over investment risk, I think.

AI assessment note: “I do think a hundred billion, there are not a lot of companies out there”

Q If you think about Google on Amazon, you've got Google Cloud, Cash Cow, App AWS, Cash Cow, and then Zuck is like, oh, Instagram is the cash cow, actually. How does that change how he acts? Cause Google and Amazon do also have distribution.

A They have distribution. So on Google, I'd argue GCP is not the cash cow, right? It's, it's search. And with search, I'm not sure. I I'm not sure how AI is going to impact search where it is today. AI is really good at taking a low signal data and translating into high value output. So in the case of meta, taking what you're looking at on Instagram and figuring out, okay, you want to go Buy this type of product. That is incredibly valuable. Google's got an incredibly valuable machine in that you type into the Google search bar exactly what you want, right? You type in skiing in Park City. Well, guess what? I know you, you want to go skiing in Park City. Your intentionality is a hundred percent known at that point for me. Whereas I think for Meta, I think the value of AI is, at least on the ad matching side, I look at a bunch of What you are, uh, browsing on Instagram and I say, okay, Harry wants to go skiing. He hasn't decided yet that he wants to go to park city. Maybe he wants to go to Whistler. How do I then move you in, you know, that path to get you to a different location, right? Or maybe you're even earlier on in the funnel and you say, I just, Harry wants to go on vacation and maybe you haven't even decided on skiing. Maybe you want to go on, uh, you want to go to the beach. So I think AI has a lot of value for that. I'm not sure as much for Google. How do you influence,…

AI assessment note: “GCP is not the cash cow, right? It's, it's search.”

Q There are so many benefits to being tied to such great people and such a great brand. What are the negatives?

A So yes, there are tons of great things. I think the biggest risk for us Is being based here in Silicon Valley, you believe everything's gonna happen overnight? Mobile or cloud and now AI are taking over the world. There's just left, leave all this destruction in the wake. And the reality is these things usually happen very fast, but also much more slowly than peak enthusiasm would suggest, right? So as a hedge fund, we're, we're trading off the wall street echo chamber for the Silicon Valley venture echo chamber. And that certainly got puts and takes. One thing that we try to do is to, uh, to combat that is to host outside voices to speak on occasion. And so I forget if it was Doug or Ruloff who organized this, but we organized, for example, Charlie Munger before he passed to come in and talk to the partnership. Uh, we've had Stan Druckenmiller come in and talk to us. They think in a very different way than we do.

AI assessment note: “we're trading off the wall street echo chamber for the Silicon Valley venture echo chamber”

Q Which year was it? How do you rally a team in a down year?

A It's really hard, right? You know, I think, so we do these surveys, uh, every, I think it's every six months just to get a sense for how, and they're anonymous for how the team is doing. And man, the, you know, The initial, I'd say right post COVID, the surveys were very positive. You know, everyone's super happy. You know, everyone's feeling fulfilled. They're doing good work. And then you can see the surveys just getting worse and worse. And then in 20, 22, they're just downright awful. And I think one of the things that I've learned is that your mentality is so dictated by performance. And that's just, you can't, you can't live that way. Right? You can't manage your business that way. You can't actually go make decisions that way. You have to have, stay even keeled and say like, what is, what about my process is not working? Let me go fix that. But if you're, you know, and this is, this is the discussion about greed when, you know, your others are greedy, fear when others are fearful, right? If you actually let that impact you in that way, you are going to make the wrong decisions. And so I think those surveys are actually quite helpful because it gives you a sense for the team's mindset and how its impact. You could probably almost draw a line on, you know, the NASDAQ and the team survey.

AI assessment note: “And then in 20, 22, they're just downright awful.”

Q Which year was it? How do you rally a team in a down year?

A It's really hard, right? You know, I think, so we do these surveys, uh, every, I think it's every six months just to get a sense for how, and they're anonymous for how the team is doing. And man, the, you know, The initial, I'd say right post COVID, the surveys were very positive. You know, everyone's super happy. You know, everyone's feeling fulfilled. They're doing good work. And then you can see the surveys just getting worse and worse. And then in 20, 22, they're just downright awful. And I think one of the things that I've learned is that your mentality is so dictated by performance. And that's just, you can't, you can't live that way. Right? You can't manage your business that way. You can't actually go make decisions that way. You have to have, stay even keeled and say like, what is, what about my process is not working? Let me go fix that. But if you're, you know, and this is, this is the discussion about greed when, you know, your others are greedy, fear when others are fearful, right? If you actually let that impact you in that way, you are going to make the wrong decisions. And so I think those surveys are actually quite helpful because it gives you a sense for the team's mindset and how its impact. You could probably almost draw a line on, you know, the NASDAQ and the team survey.

AI assessment note: “in 20, 22, they're just downright awful.”

Q Ok, you get it wrong, and it goes the opposite of what you think. It could be up or it could be down depending on your long or short. How do you know when to call the decision wrong versus when to need more data to see if you were wrong?

A I think it's hard. I think it's hard to know, and this is part of the being emotional and attached or not. I think it's really important to be dispassionate and look at the data in a way that, uh, really, uh, synthesizes it for what it is. So if it's good, we have to really call it good. If it's bad, we have to really call it bad. And so one of the things that we've done is we've actually pulled in our data science team as an extension of the investment team. And so we are, uh, they join our weekly pipeline meetings. Uh, we have them very tightly integrated with, with our investment team to make sure that I think we have that data science angle, uh, in, in all of our processes.

AI assessment note: “we've actually pulled in our data science team as an extension of the investment team”