David Cahn: Why Servers, Steel and Power Are the Pillars Powering the Future of AI | E1186

1. 0:00 – 4:41

   Intro

   1. DCDavid Cahn0:00

      No one's ever gonna train a frontier model on the same data center twice, 'cause by the time you've trained it, the GPUs will be outdated and the data center will be too small. The bigger these models get, the more that scaling laws become the dominant thing. And so I think there's a really good argument to be made that basically the data center is the most important asset. I'll propose my own three things that I think are the three things that matter. I would summarize it as servers, steel, and power. The industrial revolution is just getting started.

   2. HSHarry Stebbings0:25

      Ready to go? David, I am so excited for this. I'm so excited for this because I've heard for a while from Pat Grady, no one works as hard as Mr. David Karn, and how early you get into the office, which has made me feel terrible because Pat like up at 5:00 AM. So thank you so much for joining me.

   3. DCDavid Cahn0:55

      Thanks for having me.

   4. HSHarry Stebbings0:56

      I wanna start though with a little bit of context. I think we're really very shaped by some elements of our childhood. When you think about your childhood and your early years, you're a twin. How did being a twin influence who you are today, David?

   5. DCDavid Cahn1:10

      It's a great first question, and I, I, I would have to say that actually being a twin is the single most important fact about me, and so it's a good place to start. Um, I would s-

   6. HSHarry Stebbings1:19

      Why is it the most important fact about you?

   7. DCDavid Cahn1:22

      It's the single biggest thing that's influenced my life. Um, and I would say, first of all, I love being a twin. My brother and I are extremely close and we spend a lot of time talking to each other. We talk to each other every day. There's two ways that it's really influenced my life. I would say the first way is, I think being a twin gives you a bit of a license to be non-conformist. And especially in technology, especially in the Valley, there can be a lot of herd mentality, and we're gonna get into that a little bit on the AI stuff. But I think that being a twin growing up, you just feel this sense of this person is with me, this person is at my side, and I can kind of be who I wanna be and say what I wanna say. And I think that was really empowering growing up. I think the second way that being a twin really influenced me, and I think Andrew Reid would probably agree with this, who's also a twin, being a twin just makes you hypercompetitive. You're growing up, you're, you're, you know, we, my brother and I used to swim competitively when we were seven, eight, nine, 10 years old. And you're just one stroke behind, you know, you're so close and, or you're one stroke ahead and you're afraid of being one stroke behind, right? And so I think there's a bit of you're so close and you don't have this, I think with other people you can say, "Oh, well, this person's smarter than me," or, "This person is stronger than me." Or... You have the same deck of cards. So it becomes about how do you play that deck of cards really well. And I think being a twin kind of pulls you into this direction of, how do I make the most of myself? How do I make the most of what I have in order to be competitive?

   8. HSHarry Stebbings2:48

      I think that's the best advert I've ever heard for being a twin.

   9. DCDavid Cahn2:51

      (laughs)

   10. HSHarry Stebbings2:51

       I always thought it sounded quite hard, um, uh, constantly needing to prove yourself, but when you actually put it in the same way that you did there, I, I love that. And I also love the close relationship you have there. I, I also think that you're often shaped by challenging elements of your childhood. I remember when my mother got MS. People don't know this, dude. Well, I, my mother got MS and I, we didn't make any money from the show, and I said, "I'll pay for the show." And we used the podcast to pay for all of her treatment, hundreds of thousands of dollars.

   11. DCDavid Cahn3:18

       Wow, that's amazing.

   12. HSHarry Stebbings3:19

       But I, I didn't know that you could make money from podcasts, but that shaped me, her getting MS. What challenging moment from your childhood do you think really shaped you?

   13. DCDavid Cahn3:27

       I don't know if this is a challenging moment, but I think my family history and, and heritage is something that's really shaped me. My dad's family fled Nazi Germany in the 30s, uh, to the United States. My mom's family immigrated from Syria to the United States. And both my parents were the first in their families to go to college. And so I think that learning from them and learning from their families and seeing that struggle, I feel the, the pressure of, you know, this is the, this is the chance, this is our chance, you know? 100 years of, of struggle and, and sacrifice has sort of gone into putting me in this position where I have this incredible opportunity to be in Silicon Valley investing in technology, working with founders, and so I feel that, I think about that every day, and I think that's something that really drives me.

   14. HSHarry Stebbings4:13

       You were an incredibly high achiever growing up, everyone told me. Did your parents tell you that they were proud of you with all your achievements?

   15. DCDavid Cahn4:20

       I think yes. Um, I mean, look, m- my dad was a, was my coach growing up. He's someone who success sort of doesn't matter. You don't celebrate wins, you sort of, it's always, "What's next? How can I be better? If you get a 99, how do I get 100?" And so I don't know that they were proud of u- I'm sure they were proud of us, but it's n- it's not, not the topic of conversation. The topic of conversation

2. 4:41 – 8:09

   Exploring AI's Impact: Two Critical Questions

   1. DCDavid Cahn4:41

      is, what's next?

   2. HSHarry Stebbings4:42

      I mean, the topic of conversation being what's next would only lead us to one thing if we're actually gonna talk about technology or venture today-

   3. DCDavid Cahn4:48

      (laughs)

   4. HSHarry Stebbings4:48

      ... which would be artificial intelligence. And you've written, I think, probably two of the most seminal pieces on AI this year. And I wanna start with one actually, which is, you said at the top of, you know, the game theory of AI CapEx, will AI change the world, and are CapEx levels too high? You said they're two different questions. Can you explain why to me?

   5. DCDavid Cahn5:14

      Yeah. And let me set a bit of context here too. I've been investing in AI for about six years. I was on the board of Weights & Biases starting in 2019. Led the pre Stable Diffusion round in Runway ML, so this is before any of the AI video generation had been created. Um, and led the Series C round in Hugging Face. And so I've been around AI for a little bit. I'm very passionate about AI, and I think one of my core beliefs, in fact, I would say my strongest core belief that drives m- my investing behavior is when I'm 80 years old, my life is gonna be completely different because artificial intelligence is gonna change everything about how our, how our society operates. And I think as a VC, I'm living, you know, many years in the future in terms of thinking about what are the companies to invest in to get there? So I think that sets the stage on AI as this incredible thing, right? And I think a lot of us kind of believe that, a lot of people in Silicon Valley are talking about this. And yet at the same time, we're investing hundreds of billions of dollars-Those are big numbers, right? I think people sort of, we struggle to grok numbers. The entire SaaS economy is a $250 billion market, right? So we're talking really big CapEx numbers. And I think that for the last year, there's been a lot of kind of proponents or, or, or promoters, if you will, in AI, who are basically like, "No matter how high that number gets, it's fine. AI is gonna change the world. Don't worry about it. Like, that number is fine." And I don't even really disagree with them, but I think it's important, and this is why I wrote The AI $600 Billion Question. I think it's important just to say, okay, at some point we gotta start looking at the numbers and asking ourselves, like, "How is this gonna play out?" And that affects everybody in the ecosystem. And so to, to, to specifically answer your question, believing in AI and believing that infinite CapEx in the next two years, those are very different concepts. You can believe in AI but also believe that the amount of CapEx in the next 24 months may be difficult to pay back.

   6. HSHarry Stebbings7:04

      I completely agree with you. I actually was just speaking to someone the other day within one of the largest companies with, you know, bluntly access to all of the budgets and he said, "You don't get it, Harry. This is like the Manhattan Project. When you're in, you're in and-"

   7. DCDavid Cahn7:17

      Yeah.

   8. HSHarry Stebbings7:17

      ... you've got, you can't keep, you can't pull out."

   9. DCDavid Cahn7:20

      The, I think, you know, we're talking 48 hours after Mark Zuckerberg just acknowledged the overbuilding that's happening. Sundar talked about this. And so I actually think that something meaningful just changed this week in the sense of, I actually think Mark Zuckerberg, Sundar, and I all basically believe the same thing, and I think we had all these kinda promoters coming into AI trying to say, "Hey, look, the CapEx doesn't matter. The budgets don't matter." And I think what you heard this week from Sundar and from Mark Zuckerberg is they understand that this is risky. Now, it's a risk worth taking, and they have to take it, and we can talk about the game theory behind that and why it makes sense for them to take that risk. Even Mark Zuckerberg is not sitting here saying, "Hey, it's risk-free. AGI is coming. Don't worry." He's saying, "If AGI comes, great, fantastic. This is gonna be a great investment. If it doesn't come, this may be a difficult investment. It's a risk we have to take."

3. 8:09 – 12:02

   Impact of Conscious CapEx Overspending in AI

   1. DCDavid Cahn8:09

   2. HSHarry Stebbings8:09

      Can I ask you, you mentioned the realization that across the board from, as you said, Sasha and, and Mark more specifically, um, what happens then? A realization that leads to a reduction in that CapEx spend? A realization that means it's now conscious overspending of CapEx? What happens then?

   3. DCDavid Cahn8:27

      I don't think it changes much. I mean, eh, my opinion, and this is my opinion, but my opinion is that Mark, Sundar, and I have always agreed, right? My opinion is actually that, like, it is the, these sort of, uh, let's call them crypto proponents who've sorta came into AI in the last 12 months who are, you know, shouting from the rooftops that AI is going to the moon, that's the group of people who've been saying that CapEx is all rational. I don't think that any of the big tech companies are sitting in their boardrooms saying, "We know exactly how we're gonna pay back this CapEx." Now, that's fine, right? They're making a strategic calculation, "If we don't make this investment, then my, our competitors are gonna get ahead." And I think one thing that people weren't paying attention to and maybe are now is that this is one of the most powerful oligopolies in the history of business that we're dealing with, right? Microsoft, Amazon, and A- and, and, and, uh, and, and Microsoft Azure and Google now represent $7 trillion of market cap. That's 10% of the global market cap. So we're talking about one of the most powerful oligopolies in the history of business. Of course they're gonna be willing to spend aggressively to protect their oligopoly. The wrong takeaway is, "Oh, Google knows something we don't know and therefore is making this investment, and therefore hundreds of billions of dollars of AI revenue are guaranteed in the next two years." That's the extrapolation I think people were making that I was questioning, and I think increasingly people are gonna realize these investments are happening but they are speculative and there's a lot on the come.

   4. HSHarry Stebbings9:49

      Could you argue this is great, the speculative investments are being made by incumbents with a huge amount of money, and if they work out, we will have amazing products and consumers will get great value from them? And if they don't, great. We weren't the ones who paid for it, and incumbents with huge cash reserves were. Could you take that view? And is there an opposing side to that?

   5. DCDavid Cahn10:13

      I've taken that view. I mean, that's the view I, I, when I first published The $200 Billion Question nine months ago, I said, "This is great news for startups." And when I published The $600 Billion Question, I said, "This is great news for startups." So one, one refrain that I've sort of been consistent on is, this is fantastic for startups. These big tech companies are producers of compute. Startups are consumers of compute. So if you believe there is an overproduction of compute and that compute prices go down, startups win because they are buying those compute. Lower cost of compute directly translates to higher gross margin for startups. Higher gross margin for startups translates directly to more valuable companies. So I'm very happy that this is happening, and I think it's fantastic for the economy. I think it's fantastic for the technology ecosystem, but I think we do need to be clear-eyed about what it is and we do need to be clear-eyed about w- the next few years and what's on the horizon.

   6. HSHarry Stebbings11:04

      Is there an alternative argument that it allows for the continued concentrating of power, the continued oligopolies becoming even stronger, that we must consider as well?

   7. DCDavid Cahn11:14

      I think there is, right? And I th- I think that's a super valid perspective. I think the perspective y- what you would, what you would argue is these companies are too powerful. They're basically erecting barriers to entry, right? Now, in order to be an AI cloud, you need to be willing to light a bunch of money on fire, right? And so I think that that is a pretty significant barrier to entry for new entrants. And by the way, I think what, what you heard from the commentary this week is that's not an accident, right? That's pretty explicitly what they're trying to do is say, "Hey, we cannot afford to let anybody else, you know, attack our golden goose." This cloud business today is a $250 billion business. So the cloud business today is the same size as the SaaS sector. That's the business that Azure, Google, and, uh, and, uh, and AWS control, is the same size as the entire SaaS sector. So of course they're gonna do everything they can to

4. 12:02 – 17:35

   Reconciling Compute Cost and Its Future Importance

   1. DCDavid Cahn12:02

      protect it.

   2. HSHarry Stebbings12:02

      Can I ask you a hard one? But I'm struggling... Again, the, the joys of the 20VZ, uh, show is I just learn from amazingly smart people like you and Sam Altman, and I struggle to reconcile two thoughts, which is, like, the overproduction of compute, meaning the cost comes down for startups, the consumers of it, and then Sam Altman very bullishly stating that compute is the currency of the future, and that is the single most important thing. Are those two beliefs at odds?... or can they coincide?

   3. DCDavid Cahn12:30

      I think the thing that bridges, so I think they are a little bit at odds. And I'll tell you why. The compute is a euphemism, right? Compute is the thing that you consume, but how do you make compute really matters. So what is compute? Compute is a data center somewhere in the middle of Illinois with a bunch of GPUs and a bunch of liquid cooling systems and a bunch of physical things in the world. So I think the part of the reason that the dialogue around AI has become so extreme is that people use these euphemisms, compute, cloud, all this stuff, and it doesn't actually, it doesn't actually capture the physical (laughs) reality of what this is. And so the big question is, okay, you're a big tech company. You, in the next two years, data centers take about two years to build, you are going to build a, and it costs about $2 billion per data center. So you're gonna build a $2 billion data center in the middle of Illinois and you're gonna set it up. You don't actually know how to set it up, right? Because we actually don't know how to build GPU data centers yet. It's very new. So you're gonna set it up in the best way that you think possible, you're gonna buy H100 chips 'cause that's the best chip on the market today, and you're gonna put them in this data center. Now imagine two years from now, NVIDIA's B100 chip is now the dominant chip. So now you need to take all those chips in the data center and you need to move them, or you need to put in th- these new chips 'cause these new chips are better. Now, liquid cooling systems have changed so now you need to change the liquid cooling system. So I fully agree, compute is the future. However, compute is generated from a physical asset that is physically built in the world, and if we build those incorrectly, we are going to have to deal with that and we're going to have to build new ones. And so there is, it's not just this pure thing where it's like, "Hey, we're gonna build it all now. We're building 15 years of compute today." And like, "Great, we have more compute." There's complexity to it. There's trade-offs.

   4. HSHarry Stebbings14:09

      So, (laughs) my hand's getting more and more filled with notes. Uh, when you write, it takes up too much. It's too noisy. Uh, so I, I have to ask that, um, you know, we, we mentioned there's a couple of different elements, but you mentioned there about kind of needing to replenish your, you know, H100s with the newer model, whatever that is, in two years or three years, whatever period that is. If models just become a lot more efficient, would you still need to replenish them? Could you see model efficiency to the extent where you wouldn't need the replenishment of newer models?

   5. DCDavid Cahn14:39

      I think model efficiency is gonna improve, and I actually think this is one of the points I made in the game theory piece, which is, if you're optimistic on AI, you are actually more concerned about data centers becoming outdated. And there's this great line that I, that I heard somebody say and, and I've just been thinking about it a lot ever since, which is, you know, no one's ever gonna train a frontier model on the same data center twice because by the time you've trained it, the GPUs will be outdated and the data center will be too small. I just thought it was amazing and it's like, it's reminiscent of the, you know, no man ever steps in the same river twice, right? Um-

   6. HSHarry Stebbings15:11

      I, I actually s- I actually said that. That was me.

   7. DCDavid Cahn15:13

      (laughs) Yeah, yeah.

   8. HSHarry Stebbings15:15

      (laughs) . Direct attribution to Harry. Thank you so much.

   9. DCDavid Cahn15:17

      Um, (laughs) and I, I just thought it was just such a fantastic way of just succinctly summarizing the issue here which is, let's say models get way better, let's say models change, let's say we're, scaling laws continue to hold, all this good stuff happens. Actually that means that we probably need to change the architecture of the data center. We probably need new chips. You, you know, right now everyone's chasing 100,000 GPU cluster, right? That's the sort of bleeding edge. Elon says he's gonna build a 300,000 GPU cluster. If all that actually happens, it's gonna dramatically change the physical architecture, the real estate of what actually needs to happen for these data centers, and I think increasingly you could think, you could argue a model's just a data center, right? The bigger these models get, the more that scaling laws become the dominant thing. All these researchers, they jump from one lab to the other, it's not like there's that much differentiation anymore between the, the actual models that these companies are, are launching. And so I think there's a really good argument to be made that basically the data center is the most important asset and that we're gonna have to learn how to build these really, really big data centers.

   10. HSHarry Stebbings16:18

       How does the data center change? You mentioned it changes in real estate moving from, you know, 100,000 to 300,000. What would one expect to see in that changing landscape?

   11. DCDavid Cahn16:27

       Here's what's happening, and I think basically no, a lot of people are not paying attention to this right now, Amazon in the last six months has announced $50 billion of new data centers, right? This is a, when I talk about AI $600 billion question, like this is the cost line on the $600 billion question, like these data centers are getting built. They have to go hire people to go build these things, right? Like, you're gonna go to some town, there's a company called CyrusOne, there's another company called QTS, these companies are the real estate developer for data center. So Microsoft or Amazon goes to this real estate developer and he says, "Hey, I want you to build me this data center." The real estate developer goes to this company called DPR, DPR is the biggest general contractor in building data centers, and the guy, DPR is like, "Hey, I need you to build me this data center." DPR goes and finds a subcontractor. The subcontractor needs to go find you a thousand electricians and all these people. I mean, labor is the single biggest cost in data centers. And now you're putting ads on Facebook and you're like, "Electricians needed. Please come to this random town in the middle of Illinois. We need you to come build a data center," right? And then you have all these people who like, they're getting on planes, they're gonna be put up in hotels. Like, just the sheer physicality of what's gonna happen in the next 12 months is pretty awesome and I think it's gonna be really fun to see.

5. 17:35 – 21:38

   Is Model Progression Outpacing Data Center Development?

   1. DCDavid Cahn17:35

   2. HSHarry Stebbings17:35

      Are we overestimating the speed with which we are able to build these data centers? Given the complexity that you mentioned there, will we not see model progression way exceed data center development in a way that is very challenging?

   3. DCDavid Cahn17:49

      Well, this is the scaling law question, right? This comes back to like the bitter lesson that people talk about, and I, I don't actually h- have a very strong opinion on this, but I would say the question comes down to are the researchers gonna come up with research breakthroughs faster than we can build bigger data centers? There's a lot of people in Silicon Valley who say all that matters now is scale. There's other people, and I've met a lot of these people and I think they're quite smart, who will say, "No, actually we're gonna make a breakthrough on reasoning, we're gonna figure out how to use data better, we're gonna figure out how to make the models more efficient." So I think there's two schools of thought on this. The evidence today I think is more strongly in favor of scaling laws and I hope to see more evidence of, of the, of these other breakthroughs being able to drive progress because I want to see as much progress as possible and if we can have both, that's better than having just one.

   4. HSHarry Stebbings18:33

      We mentioned startups being the beneficiaries in terms of, you know, cost of compute coming down and them as the consumers benefit from that.Maybe. My question is, when you look at, say, a Canva, is a great example. They've integrated a lot of AI features into their core products and whenever you use those features, they are paying to OpenAI, to Anthropic, to s- you, you name whatever vendor it is, and they are not charging additional revenue per seat. And so actually, it's just denigrating their margin already. And so my question to you is, does it actually make worse companies because they're not able to charge more, but they have a higher cost 'cause it costs to integrate AI?

   5. DCDavid Cahn19:14

      Look, uh, E- Elon came to Sequoia's base camp event this year and he said something that stuck with me, which was, "The only thing that matters is building useful stuff." Like, everything else in business kind of figures itself out. You do something useful, people generally will pay for it. If you don't do something useful, people won't pay for it. And so I think (laughs) the, the question for Canva and the question for all of these software companies is, can you make your product more useful with AI? I think if the answer is yes, then you probably have pricing power, you can probably earn a margin if you're driving value to the end consumer. And this is your point, I think, Harry. However, if you are not making your product more valuable and if you are simply injecting AI so that you can s- tell investors that you have AI and so that you can, you know, talk on your marketing page about how you have AI, then you're just injecting a new cost item with actually no benefit to the consumer. And if there's no benefit to the consumer, they're not gonna pay more. And something I like to talk about on this is, like, Netflix. I mean, think about how much value you get from Netflix. Netflix is amazing. Think about how much val- value you get from Spotify. It's amazing. Like, products that consumers buy are really, really good and you don't pay that much for them, and capitalism's amazing that way. And so the burden to deliver value is very, very high in business and we're not seeing that so strongly in AI yet.

   6. HSHarry Stebbings20:24

      I'm not sure if I agree that if you add value, you can charge more-

   7. DCDavid Cahn20:28

      Mm.

   8. HSHarry Stebbings20:28

      ... because if you have parity of features and the commoditization of features, and you have Canva offering it and Adobe offering it and Sketch offering it, and, and 10 others, even though it adds more value, if you have a mass of other providers offering the same, all pricing power goes and you are forced in a race to the bottom, which I think we're already seeing across a lot of spaces. How do you think about that one, am I wrong?

   9. DCDavid Cahn20:54

      Uh, no, I think you're right and it's nuanced, right? I think the nuance here would be in businesses with barriers to entry, you have pricing power. In businesses with low barriers to entry, you don't have pricing power. So I actually think that will be sector by sector. There will be some sectors where you're right, things get commoditized and they just basically get bid down to 0% gross margin. In industries with structural low gross margins, it's very hard to raise the gross margin, so if cost goes up, you'll raise price a little bit to compensate for it, but you won't raise price that much. In industries with higher barrier to entry, for example, industries that have data moats, industries where if, if I put all my data on your platform then I can't really move, I think those people have pricing power and if you had pricing power before AI, you'll have pricing power after AI. If you didn't have pricing power before AI, you probably won't have pricing power

6. 21:38 – 24:26

   The Future of Vertical Integration in the Compute Stack

   1. DCDavid Cahn21:38

      after AI.

   2. HSHarry Stebbings21:38

      W- going back to compute, I do just have to ask, we've, we've seen more and more players talk about kind of the owning the vertical stack, talk about, you know, actually starting their own chips. Apple were very open in terms of talking about how they plan to, you know, how they do now, but use and implement their own chips and move away from kind of the NVIDIA monopoly. Do you think that will be an ongoing trend, where we see the continuing, uh, verticalization of the stack from different players, and, and how should we think about that?

   3. DCDavid Cahn22:05

      Yes and no, and I'll, I'll break it down into two pieces. There's one element in which I say yes and one element in which I say no. The element in which I say no is, you know, I, I have ha- had this mantra for six years and it's served me well, it's like, "Don't bet against Jensen." Like, I think it's really, really difficult to bet against Jensen and NVIDIA. Jensen, I, I, I, I had friends joining NVIDIA years ago and I said to them, "You're making a super smart decision. Jensen is the Steve Jobs of our era." And that seems increasingly true, right? Jensen's just a phenomenal CEO, I think it's really hard to bet against phenomenal CEOs. And, and a, and a visionary CEO, right? Like, this was not obvious when Jensen started doing this 30 years ago. So I, I wouldn't bet against NVIDIA. Now, is there gonna be more verticalization? Yes, and I think that actually comes back to this, like, data center construction piece and running the data center. One opinion that I've changed, that my mind has changed on over the last few months and, and year, you hear Elon talking so much about data center construction and he's gonna build his own data centers and vertical integration. Increasingly, I think that's a pretty smart approach, and I think that vertical integration between the model layer and the data center matters. By the way, the other company that has vertical integration is Meta.

   4. HSHarry Stebbings23:08

      Mm.

   5. DCDavid Cahn23:08

      Zuck controls his own data centers, right? He also is building the models. And so I think the vertical integration piece is gonna matter in the sense of you can't have a separate team running the data center and a separate team building the model. That is just not gonna work as these models get bigger and bigger. You will need to deeply couple those things, and I think you look at Elon, you look at Zuck, they are doing that. And I think actually one of the big questions that Microsoft and OpenAI are gonna face, uh, and Amazon and Anthropic are gonna, gonna face is, how do they, how do they unify those efforts, right? Anthropic and OpenAI are separate from the parent company. The parent company controls the data centers.

   6. HSHarry Stebbings23:40

      That was gonna be my question-

   7. DCDavid Cahn23:41

      (laughs)

   8. HSHarry Stebbings23:41

      ... which is like, do you think this is something that a startup can feasibly do, given the CapEx heavy nature of this and the verticalization that you spoke of there? Even though OpenAI or Anthropic can raise a billion, two billion, three billion, Microsoft throw off 330 million in three- free cash flow per day. Is this a game that anyone but Mag7 can play?

   9. DCDavid Cahn24:06

      I think the evidence of the last couple years is, in order to play in the big model game, you need a cash machine and that cash machine cannot be the AI (laughs) business, right? So Facebook has a cash machine called Instagram, uh, Amazon has a cash machine called AWS, Microsoft has a cash machine called Azure. I th- um, I think you need a cash machine in order to

7. 24:26 – 28:51

   Facebook's Unique Position Without Cloud Revenue

   1. DCDavid Cahn24:26

      compete.

   2. HSHarry Stebbings24:26

      How does Facebook, being the only one not having cloud as their cash machine, change how they behave, do you think?

   3. DCDavid Cahn24:35

      I think that, you know, the cloud guys are playing defense, Meta's playing offense, right? And I th- so I think that's, like, the easiest way to think about it. The cloud guys are protecting their existing business. I think Meta can afford to be pretty creative (laughs) and also I think Meta...... has to play less defense because if they decide it's not worthwhile, they don't have to keep investing. The cloud guys are stuck in more of a prisoner's dilemma. They have to keep investing. If they do not invest, they risk losing market share in one of the greatest businesses of all time. And Meta gets to sort of play for the future. And Zuck is still young, and he's doing a fantastic job. And I could imagine Zuck playing a pretty important role in the future of AI.

   4. HSHarry Stebbings25:09

      If you were Zuck today, what would you be doing?

   5. DCDavid Cahn25:12

      I think exactly what he's doing. I mean, he's super, super sharp. And I think what he's doing makes sense. He is creating an open source, or open source alternative. I think that is good for the world. I think that is good for, for Meta. And so I think LLaMA is fantastic. I think it's great for startups. Startups are gonna build on top of LLaMA, they're building on top of open source. I think the consumer killer use case hasn't emerged yet. Obviously Meta has fantastic distribution, and so if they can unlock that killer use case, they'll figure out how to, how to do that and they'll make a lot of money off of that. So I think the Meta approach really makes a lot of sense.

   6. HSHarry Stebbings25:44

      Given the need to be so tied to one of these cash cow businesses over the long term for Anthropic, for, um, OpenAI, for any of the smaller players, we have Cohere on the show next week actually, will they be acquired and they will be a part of Amazon or of Google? Is that the only pathway forward?

   7. DCDavid Cahn26:05

      I don't know. I mean, it, it does seem like a lot of what's going on in AI has to do with antitrust, right? Like Microsoft probably doesn't want to acquire OpenAI. Amazon probably doesn't want to acquire Anthropic. And so I think you have this setup, and this is unique, and as a student of business I kind of love this stuff, I think it's cool. It's like this is kind of unique where you have these kind of like they're, they're not wholly owned, they're not controlled, they're kind of like 49% owned, right? So you have some amount, some, a strong amount of ownership, strong amount of incentive alignment, but separate businesses. I think that will probably sustain because it's not really possible to acquire these businesses. Now you can acquire the smaller ones, and we've seen that with Inflection and Adept, you can acquire some of the teams. But it's gonna be much more difficult to acquire something like Anthropic.

   8. HSHarry Stebbings26:46

      To what extent do you think about the ones that aren't investing so heavily in it, in your, you, we mentioned obviously, you know, Google, we mentioned Meta, Microsoft. What about Salesforce? What about Netflix? What about the $100 billion plus, but who aren't in that core leading division? Are they at an advantage or a disadvantage?

   9. DCDavid Cahn27:05

      I don't know. I mean, I do think the like tier below those three big companies are scared right now. I think there's probably like, those boardroom conversations are probably tense, right? I think on the one hand, huge opportunity, huge opportunity, right? Make the product better, you already have existing distribution, distribution is a huge advantage and a huge moat. So on the one hand you're like, this could be a huge opportunity. On the other hand, you need a lot of money to compete with those big companies, and they're gonna have a lot of power. And so I, I think the smaller companies are very concerned, and I think that they are doing everything they can in order to compete. But again, there's a lot of game theory here in terms of the big three companies can afford to go to war with each other and do this arms race, and the other guys can't, or maybe they can, and maybe they'll try and, right? It's, it's a sort of a less, it's a less obvious to what is the strictly rational calculation. Zuck used this phrase this week in his, uh, in his quote, right? "It's, uh, strictly rational for the big companies to invest." It's less obvious what's strictly rational if you're Salesforce. It's less obvious what's strictly rational if you're Workday.

   10. HSHarry Stebbings28:04

       Do you think it's strictly rational for Wall Street? And what I mean by that is how CEOs and leadership understand the importance of investing in AI is not always the same as the perception on Wall Street. Do you think the messaging around AI investing will or needs to change around how these big companies make these massive CapEx investments?

   11. DCDavid Cahn28:26

       Look, I think Wall Street is quite sophisticated, and so I think Wall Street's doing their homework, and I have a lot of admiration for what those folks do. And so my suspicion is they know what's going on. That said, there's always a, there's always a gap between Wall Street and Silicon Valley. Like we have our sort of, especially like VCs, we have our nose to the ground, we're like sniffing around all the time, like, "What's going on?" You know, "What's happening?" And so there's some gap I think in their, in the understanding, but I think the, the, those guys are super sophisticated and I wouldn't bet against them either.

8. 28:51 – 30:41

   Core Bottlenecks in AI: Compute, Algorithms, or Data?

   1. DCDavid Cahn28:51

   2. HSHarry Stebbings28:51

      I had Alex Wang on the show not too long ago from Scale, and he said, "Really there's like three things. There's compute," which we've discussed, "there's, uh, algorithms, and then there's data."

   3. DCDavid Cahn29:01

      Mm-hmm.

   4. HSHarry Stebbings29:01

      He said, "I take the alternate view to a lot of people. I think data is the core bottleneck on model and AI progression today." To what extent do you agree with him? And when you look at compute algorithms and data, which one do you think is the core constraint?

   5. DCDavid Cahn29:18

      It's funny, I, I used to kind of agree with Alex, and I think that my mental model has actually shifted. I kind of think compute models and data have kind of converged. Like, all the big guys know what they're doing. They're all doing the same thing. Everybody's using Scale. Now, I love Alex. Alex is a fantastic CEO. He's doing a great job, right? Like he is kind, kind of this arms dealer, everybody has to buy from him, he's providing all these big, big model companies with data. But I think it's really hard to argue today that any of the big model companies has a data advantage. Compute, it's just a commodity that you pay for. So it's hard to argue that any of them has a compute advantage. And then models, they would all argue that they have some secret sauce, but again, if you believe the bitter lesson and you believe that scaling laws are the things that matter, then the secret sauce is like not that material. And so I'll propose my own three things that I think are the three things that matter, and I would summarize it as servers, steel, and power. So I'm just much more interested in the industrial nature of AI and like what is happening is, is this industrial revolution. And so servers, that's NVIDIA, AMD, Broadcom, like the chip innovation, NVIDIA has an amazing gross margin. There's gonna be a ton of competition. The chip wars are just getting started. So there's a lot of interesting stuff happening there. Steel, we talked a little bit about this before, there's tons of construction that's gonna happen. The, a lot of the big beneficiary are literally construction firms and real estate firms basically. Um, and then power we haven't talked about as much, but I think the power element here is super interesting. We've ha- we've, we've been talking about this energy revolution for a few years. Maybe it's finally gonna happen

9. 30:41 – 32:26

   The Future of Chip Pools & Nvidia's Product Roadmap

   1. DCDavid Cahn30:41

      because of AI.

   2. HSHarry Stebbings30:42

      I love alliteration. One, I love threes. You know, I, I never worked at McKinsey, but I think I'm like a secret McKinsey consultant because I love things in threes and I love alliteration. So, uh, I, I want to start on the chip pools that you mentioned. How do you expect the chip pools to play out? And when you say that we're at the start of NVIDIA's unbelievable product roadmap, can you just take me to that and how you see that playing out?

   3. DCDavid Cahn31:05

      One of the history of Silicon Valley is...... Moore's Law. Chips get cheaper, chips get better, and I think that this is one of the things Silicon Valley is amazing at, right? Is prices come down, you make things better. You look at the B100, the B100 is- is for the price, the price that you're gonna get f- based on what they've announced publicly, right? The price that you're gonna pay for more performance is amazing, right? So the price to performance ratio continues to improve. I don't see that stopping. NVIDIA's a fantastic company. They've, they've tr- have a track record of innovating, they have a track record of new products. You better believe Jensen's driving his team super hard on what the next chip is. So I think NVIDIA is gonna- gonna continue to surprise us to the positive. At the same time, AMD, Broadcom, these other companies, they see this huge business, right? And, and you're- we've, you know, you see a lot of startups now also wanting to compete. This is gonna be a huge business, NVIDIA earns a big gross margin, and when, when you, when anyone sees a fat gross margin in a big market, people come chasing.

   4. HSHarry Stebbings31:56

      To what extent do we think that import/export policies imported by a potential Trump Administration could impact the chip wars?

   5. DCDavid Cahn32:03

      Yeah, I mean, the politics of this stuff is crazy. But I think the political ramifications of all of these chips are coming from Taiwan, obviously w- everyone knows there's risk there. Does that mean the US invests in the US supply chain? I think the answer is probably yes, we're seeing that. How much of that gets on-shored? How do you do that? There's obviously a very heavy reliance on TSMC, and so, uh, I don't know how that plays out but I think that will be an important political issue in the future.

10. 32:26 – 37:48

    Key Considerations for Steel Supply & Demand Dynamics

    1. DCDavid Cahn32:26

    2. HSHarry Stebbings32:26

       In terms of, like, steel, um, I'm just intrigued, what are the core considerations that we must address with steel? Like, if we have a shortage of steel, what happens then? If we have an over-supply of steel, what happens then? What are the core considerations that people aren't thinking about with steel that we should consider?

    3. DCDavid Cahn32:44

       Yeah, and I think it's the, I use steel as, like, this catchall for kind of all the industrial stuff, so include generators there, batteries, right? There's all this industrial stuff, like, you have to manufacture. When I talk to the big cloud companies, here's what I hear. "We are calling our fact- our- our manufacturing partners, our factory guys," like the steel guy, the generator guy, et cetera. And we're saying to him, "We have a humongous order coming. Here's our next five years. We're gonna be ordering tons and tons of your product. Please increase your manufacturing capacity to support us." And what they're hearing back is, "We don't believe you. We do not wanna double our..." Because the, 'cause your manufacturing partner, let's say you're the guy making the diesel generator, you have to build a new factory in order to produce more diesel generators and if Microsoft stops buying your diesel generators, that factory's gonna sit dormant and you've just put a lot of capital under the ground. And so there is this, these supply chain dynamics here are just fascinating where you need to convince your supply chain (laughs) to increase their capacity and you- you're telling them all these orders are coming. The supply chain guy is not like an AI person, right? The supply chain guy is like a manufacturing person who runs a factory, who's really good at running factories. And so I think the, how the supply chain gets managed is, is also really interesting and the big com- tech companies actually do a very good job of this, and I think it's another advantage they have that's underestimated, is kind of convincing the supply chain to change their behavior is something that those guys have a lot of practice doing.

    4. HSHarry Stebbings34:03

       How do they do it? Is it with upfront payments? Is it with, eh, I'm just trying to understand how they do that.

    5. DCDavid Cahn34:09

       Well, Microsoft can call you and say, "I'm just gonna buy all of your capacity (laughs) for the next five years. Please build a new factory," right? And you're gonna do that if you're the factory guy. So I think this is another advantage that the big tech companies have.

    6. HSHarry Stebbings34:18

       Wait, should we not be investing in the factory guys? (laughs) I'm just-

    7. DCDavid Cahn34:22

       A lot of people are trying. A lot of people are trying.

    8. HSHarry Stebbings34:24

       I'm being, I'm being serious. You talked to me about the real estate developers in the beginning. I was like-

    9. DCDavid Cahn34:27

       (laughs)

    10. HSHarry Stebbings34:27

        ... "Why are we investing in AI CRMs?"

    11. DCDavid Cahn34:30

        (laughs)

    12. HSHarry Stebbings34:30

        That sounds like a better business. Like... (laughs)

    13. DCDavid Cahn34:33

        I'll- I'll give away my secret here which is, that's where I'm spending time, right? I'm spending a lot of time in, in this supply chain area where I think there's a lot of interesting stuff in, in the supply chain happening and I think-

    14. HSHarry Stebbings34:42

        Yeah, I-

    15. DCDavid Cahn34:42

        ... again, it's, it's less picked over.

    16. HSHarry Stebbings34:43

        Okay. Unpack that for me. Which parts of the supply chain do you find most interesting and least picked over?

    17. DCDavid Cahn34:50

        Well, I think the real estate developers, you know, KKR owns CyrusOne, Blackstone owns QTS, fantastic investments. There's gonna be huge money-making investments for those firms. So those are great investments. There's other companies in this space that are really interesting so I think that space will be interesting. Real estate developers in real estate, if you just look at real estate, it's a good business, right? And so being in that business is a good business. I think the power area is something that everyone is talking about and, you know, I- I went, I was in West Virginia, uh, two weeks ago, or three weeks ago maybe, and I visited this company that's building long duration batteries that I had invested in at CO2. They literally built a billion dollar factory in the last 12 months. So there's a just tremendous industrial movement happening and I think in 12 months, you'll see all these charts coming out of, like, more factories are getting built than ever before and all this stuff. The industrial revolution is just getting started.

    18. HSHarry Stebbings35:38

        Can you unpack that for me in terms of the industrial revolution just getting started? And also the power requirements. Do we have sufficient power supply to facilitate the industrial revolution, the compute requirements that we hear and we just discussed?

    19. DCDavid Cahn35:52

        We don't and I, you know, there, there's sort of an interesting phenomenon here which is, you know, the Biden Administration passed the IRA, the Inflation Reduction Act, and this was this huge stimulus bill that basically incentivized everyone to build solar and build batteries and all this stuff. And I think (laughs) one of the great ironies, and, and this is me as a capitalist speaking, right, is that the forces of capitalism, AI, will drive more energy revolution than any amount of political regulation could have, right? And so I think (laughs) that's just amazing.

    20. HSHarry Stebbings36:17

        Why is that? Because of the requirement for us to reduce energy costs to facilitate the AI developments that we need?

    21. DCDavid Cahn36:23

        Exactly. And because the demand for power outstrips the amount of power we have. So f- for example, everybody pretty much agrees we need more generation capacity. So if you can build a new solar farm, you're gonna do that because the economics make sense, you have a buyer, there's a lot of demand on the other side. Now, that takes time and you, you, and there's, but there's a lot of mature companies doing this. NextEra is the biggest, one of the biggest utility companies in the US.... fantastic company, $100 billion plus market cap company. Most people in Silicon Valley have not heard of this company. And th- this company is investing, i- has been investing, by the way, to their credit, for the last 10 years they've been doing this, i- i- through their unregulated business. They've been investing in batteries, solar, new technologies. When I look at clean tech investments, I see them all the time around the table. You know, this is, uh, really positive on America, right? Like, America has these great companies, like NextEra, that are doing a fantastic job and they're gonna keep innovating and they- they're gonna see the- th- the demand on the other side now, and people will build against that.

    22. HSHarry Stebbings37:13

        So do you not think we're entering an energy crisis?

    23. DCDavid Cahn37:15

        Look, I think we need more energy. Um, I think we've always needed more energy. And I'm- I may be a little bit more optimistic here, where the last... 20 years ago, everyone was saying solar is gonna be a big thing. Solar is a big thing now. It's kind of happened, right? The thesis has played out. No one's ta- no one talks about it, it happened quietly, and it's sort of funny how sometimes the biggest changes happen silently. Like, w- and solar's a huge portion of the US generation capacity today. And so, I don't know that it's gonna be this, like, loud crisis that people anticipate. I think it's gonna be a slow trek for more renewable energy, more power on the grid, and we're just gonna- we're gonna plot our way

11. 37:48 – 40:00

    The Risks & Benefits of Open vs. Closed AI Models

    1. DCDavid Cahn37:48

       there.

    2. HSHarry Stebbings37:48

       There are two core elements that I- I really want your thoughts on. I'm so enjoying this. Um, but the first is, we mentioned earlier Zuck and his obviously open sourced approach, um, versus others which have obviously got more closed approaches. How do you think about actually societal ramifications, implications on whether we should have open or closed models? Vinod and Marc Andreessen have taken two very opposing sides, saying that once it's out of the box, you can't put it back. And, you know, Alex Wang on the show said this is more dangerous than nuclear weapons. Quite a showstopper there, Alex. Thank you for that title. Um, but how do you think about that and that core consideration of, is this too dangerous to be open?

    3. DCDavid Cahn38:33

       Yeah, I think this comes, like, this is sort of the AGI debate a little bit, right? Like if you think that AGI is gonna happen tomorrow then you're, like, very afraid, and if you don't then you're not afraid, and that informs whether you believe in closed source or open source AI. I sort of challenge the question a little bit, of like, I don't really believe this, that- that- that... I don't really worry too much about any of this stuff. It's like, we're probably gonna be fine. AI is probably gonna change our lives and make it better. It's gonna improve productivity. Maybe AGI is gonna happen when you and I are, like, 90. But I don't think AGI is happening, like, tomorrow. I feel pretty good. We have closed sourced options, we have open source options. Good to have both. Would definitely not wanna live in a world where we only have one option. So I'm happy that we have both. But I'm not too worried about this stuff, and Marc Andreessen t- has r- has talked about this. I really like, actually, a lot of the stuff Marc Andreessen says on this where he says, like, "People don't have religion in Silicon Valley so they, like, worry about AGI." And I think that is sort of just, it's sort of this unknown that people wanna talk about.

    4. HSHarry Stebbings39:25

       I had Ethan Mollick, who's a professor of AI, like societal implications on yesterday actually, and it was fascinating. He said that no VCs actually make sense today who believe in AGI and are still investing in SaaS.

    5. DCDavid Cahn39:36

       (laughs) I love that, yeah.

    6. HSHarry Stebbings39:37

       So it's complete bullshit. Because if you believe in AGI, your tax accounting product for SMBs will not fucking exist if AGI is there. And so you can believe in AGI but then only invest in AGI, or you could not believe in it and keep investing in SaaS. But to hold two is not mutually compatible.

    7. DCDavid Cahn39:57

       I love that. Humans love consistency, right? So yes, that's inconsistent.

12. 40:00 – 45:03

    China's AI Progress: Behind or Underestimated?

    1. DCDavid Cahn40:00

    2. HSHarry Stebbings40:00

       I do have to ask, uh, we mentioned, um, you know, the- the potential dangers there. Another one that is just concerning a little bit for me is just how we think about China's development. Everyone says China is two years behind the US, two years behind the US. Do you believe that, or do you think actually we're underestimating their capabilities and where they actually are?

    3. DCDavid Cahn40:19

       I mean, I- I- I (laughs) I think that China's gonna catch up. I think that, you know, America's a fantastic country. We have a lot of advantages. We have great immigration. We have a great capitalist economy. We have a lot of freedom. So there's a lot of reasons to believe that America will do well. China's gonna do well also. Um, so I- I guess where do I stand on that? On net, I'm probably like, America is gonna do great. I actually think A- like, the AI th- revolution in America has a lot to do with the fact that America is a really good place to do business. It's a really good place to move if- if you want to do something great. We have a great culture in Silicon Valley. So I think that America has a good shot to stay ahead, but don't underestimate your competitors. I- I think you always gotta assume your competitor is gonna perform super well, and hopefully that motivates us to do even better. I don't know that I see a scenario in which America just completely falls behind. If anything, the last two years of evidence have demonstrated that America's doing pretty darn well.

    4. HSHarry Stebbings41:07

       Will we see a new piece from you in a couple of months called The $900 Billion Question? Like-

    5. DCDavid Cahn41:14

       (laughs)

    6. HSHarry Stebbings41:14

       ... uh, or does it shrink? Like, I- I mean that genuinely. Like, is that what we should expect, do you think, or do you think it will s- revert now with the realization that we mentioned earlier?

    7. DCDavid Cahn41:23

       I think it's f- I, first of all, I probably won't publish the piece because I only try to publish pieces when I have something to say. I'm not someone who li- I don't really like to talk about, talk for the sake of talking. So it's like, I had something to say on the $600 billion question.

    8. HSHarry Stebbings41:35

       You won't be a very good VC if you only talk when you have-

    9. DCDavid Cahn41:37

       (laughs)

    10. HSHarry Stebbings41:37

        ... some knowledge to say, David, I promise you (laughs) .

    11. DCDavid Cahn41:40

        I'm trying, trying my best. But I would say, it's very possible we get to a trillion dollars. I mean, it's just m- the math behind it, right? It, just to d- and to remind folks of what the math behind it is, you basically just take NVIDIA's run rate revenue, you- you multiply that by two to get the total cost of the data center, 'cause about half the cost of the data center is GPUs, half's- half is everything else. And then you multiply by two again to get the- the implied revenue from the companies consuming AI. And so, can it get to a trillion? Absolutely. And if anything, I think what we heard this week from Zuck and Sundar is, we have to keep building. We have no choice. We're stuck in this prisoner's dilemma. So I don't see an end in sight. I don't know what the catalyst is for people to stop building. One thing they are doing that I think is really interesting and that's changed in the last few months is, they've shifted... Basically before everyone was doing these, this CapEx off their own balance sheets, so Microsoft builds their own data center, Amazon builds their own data center, they hire the construction firm, they hire, you know, all the people doing it and it comes straight off their balance sheet. It's CapEx, it's cash flow out the door. The numbers are getting pretty big now and so I think one thing that we're gonna see more of is off-balance sheet financing. And what that means is, somebody else builds the data center and then Microsoft agrees to lease it for 20 years. Somebody else builds the data center, Google agrees to lease it for 20 years.What that does is it changes the financial profile of what it looks like to investors. Before it looked like Microsoft just spent two billion dollars. Now it's gonna look like Microsoft committed to spend $200 million for the next 20 years. What is that? That's basically debt, right? That's a, uh, a capitalized lease is a form of debt, but I think people are... I, I wonder if some of this stuff is to minimize the amount that people perceive is being spent because you can spend in this other way that is actually kind of a, a overhang on your future business, rather than cash out the door today.

    12. HSHarry Stebbings43:23

        So do you think we will see the introduction of a new financial in- instrument to facilitate the movement of on-balance sheet to off-balance sheet?

    13. DCDavid Cahn43:30

        It's very possible, and I've spoken to a bunch of real estate investors. So something that people don't know is a lot of the capital going to fund data centers comes from real estate developers, uh, real estate investors. And these real estate investors, I, I, I talk to them, and I say, "Hey, do you believe that the data center you're financing is gonna get used? Like, is there gonna be enough demand for this data center?" And they say, "I don't really care. For me, I'm giving Microsoft money. This is a loan to a big tech company. My, the deal that I'm doing is backed up by the credit of Microsoft or the credit of Amazon or the credit of Google, and I'm earning, you know, I could buy Microsoft's bonds at X percent yield, and I'm earning X plus 2% yield. And so this is a, you know, good risk-adjusted investment." And I, to me what that says is, you know, these big tech companies are basically issuing debt against their own balance sheets, but it's happening through these intermediaries so it's perceived as kind of off-balance-sheet financing.

    14. HSHarry Stebbings44:21

        What are the rates on those debt prices, average? I know it's different, but like...

    15. DCDavid Cahn44:26

        I'm not a debt guy. The, the debt guys will basically tell you, "Hey, my, my, my unlevered return is 10%. My levered return is 15%" 'Cause the debt guys are taking their own debt. Debt is complicated, so I won't go too much into the debt stuff, but what I will say is these guys are basically earning a slightly higher yield, and they think it's really good risk adjusted. And by the way, they might be right. You're basically getting Microsoft debt. Through this kinda financial engineering, you're getting a slightly better return on that Microsoft debt, and you can deploy a lot of capital against that because, hey, Harry, if I offered you, instead of buying Treasury bonds, you could buy, you know, really safe debt in probably the biggest and safest company in the world today at a premium to what you can earn buy- buying Treasuries, you might

13. 45:03 – 46:55

    Lessons in Deal Selection from Leading AI Companies

    1. DCDavid Cahn45:03

       wanna buy that.

    2. HSHarry Stebbings45:03

       How do you think we'll see that supply of capital for the financing of these facilities to change over the next few years? Because I'm a believer in like, you know, Adam Smith's invisible hand, um, and capital will flow to kind of the best opportunities. That is a great structured deal that no one would not take if you have available cash. How do you think that ecosystem changes with the realization that this is too good a deal?

    3. DCDavid Cahn45:27

       It's a pretty good deal. (laughs) So I think prices will get a bit... Like, like the spread between the risk-free, you know, return and the sort of risk-adjusted return will compress, which just means that it's gonna be less of a good deal as more people do it. And I think there's effectively tons and tons of money that wants to do this deal. And so, um, you know, Blackstone did, did this CoreWeave deal, they were one of the first, and, and Blackstone are generally very good at structuring these type of creative deals, good on them, um, and I think we'll see more of that type of deals going forward.

    4. HSHarry Stebbings45:55

       Bit of a weird one. Are you a strong believer in Adam Smith's invisible hand?

    5. DCDavid Cahn45:59

       I'm a student of s- of social psychology. I, I find psychology to be very interesting. I actually spend a lot of time reading about psychology. I think psychology drives the economy more than people realize. And so I think Adam Smith just sort of deeply understood human psychology, and that's what the invisible hand is about, and, you know, I'm, I'm a big fan of Isaac Asimov's Foundation, right? He w- he basically describes this, it's one of the gr- all-time great sci-fi books, and he basically says, "In the future there's gonna be so many people that..." And he invents this character, Harry Seldon, and Harry Sel- Seldon is a psychohistorian, and Harry Seldon sort of says, "Humans are so predictable, and we have such clear behavior that when there's 100 billion humans, you can predict the future of history." And that's sort of the premise for this sci-fi book. And so I tie that back to the, to, to Adam Smith in the sense of I think Adam Smith is doing that (laughs) on a less grand scale of saying, "Hey, human behavior is understandable, capitalism sort of takes advantages of these components of human nature, and it's one of the greatest systems that we've been able to invent so far."

14. 46:55 – 54:07

    Leading at 27 & Working with Top Companies

    1. DCDavid Cahn46:55

    2. HSHarry Stebbings46:55

       I would love to take a little bit of a change of tack though, David. This has really been one of the best shows I've, I mean, the best show I've done on AI, period. Uh-

    3. DCDavid Cahn47:04

       Wow. Thank you.

    4. HSHarry Stebbings47:04

       ... I was thinking, wh- I was thinking when I was doing this with you, "God, I can't wait to tweet like, 'I've done Sam Altman, I've done all the biggest, and David's was the best.'" Um, but I, I wanna start actually in this new shift of the show, just in terms of your investing. So I was chatting to Pat before about, bluntly, you, and the venture career that you've had so far, and he mentioned you being 27 and being the co-head of, you know, the venture division at Co2. And you worked with Databricks, UiPath, Snowflake, some of the best businesses of our time. And the first thing I really thought was how did that and those companies impact and teach you about deal selection?

    5. DCDavid Cahn47:41

       I mean, first, just to calibrate, I was the associate on those deals, so I built the models, I did the customer calls, but I was not the lead investor on those deals. What I will say, and just because you picked those companies, I'll give you a kind of a through line on those four, which is, I think the lesson from those companies is listen to what people do, not what people say, and I'll tell you why. When I called the customers on Marqeta, you call DoorDash, Instacart, Square, those were kinda the big customers, and you asked them, "Do you guys like Marqeta?" They would say, "It's fantastic, but it's too expensive. We're gonna rip it out. We're gonna build this ourselves." (laughs) That's what everyone said when you called them. Now fast forward five years later, Marqeta's a public company, none of those companies have churned, it continues to do super well. UiPath, same thing. You call all the customers, they say, "Oh, it's a Band-Aid solution, we're just using it for now, eventually we're gonna fully automate." Again, fast forward five, six years since that investment, company went public, fantastic company, people continue to use it. Even Snowflake and Databricks, same thing. You call customers in 2018 and you ask them, "What do you think about Databricks?" They'd say, "Oh, well, Amazon has their own competitor, Google has their own competitor." Snowflake, "Oh, there's BigQuery, there's Redshift, they're all kinda the same, you know, it's not that special." And then, of course, (laughs) fast forward six years, these are two of the most valuable companies that came out of this cohort. So I think the lesson is you have to listen to what customers do. They continue to pay for the, if, especially expensive products. If people are paying for expensive products and telling you they're gonna churn-... you probably shouldn't listen to that. They're paying for it for a reason. And so, the thing I focus on is how much value is this company delivering as opposed to our customers telling me they're gonna stick with it forever, 'cause you just don't know if that's right.

    6. HSHarry Stebbings49:17

       It's so funny. So many of the great investments have been missed because people say, "Ah, but they could build it themselves."

    7. DCDavid Cahn49:24

       (laughs) Yeah.

    8. HSHarry Stebbings49:25

       And they so rarely do. Can I ask you, on those early days of Coatue, what did you learn most significantly that you did not expect to learn? I think often when you're going from zero to one in a new environment, it can be some of the hardest learnings. Just like the gym. The first week or month is the toughest. (laughs) What were some of those learnings that you didn't expect to but were most impactful?

    9. DCDavid Cahn49:46

       I think the biggest thing, I mean, I was so naive when I started. I, I really didn't know what venture capital was. I didn't understand what the industry was. I didn't really know what the job was. I sort of knew, okay, I wanna be an investor. I like investing. I like businesses. But I didn't really know what to expect. And I think the biggest thing that served me well over those years was every time I meet someone, I just try to figure out, like, what are they really good at and how can I learn from this? So if there's one person who's really good at sourcing, why are they really good at sourcing? What are the things they're doing to be really good at sourcing? And I would, I would learn a bunch of that. Another person, this person's a really good leader. Why are they a good leader? Oh, it turns out people really like to follow them. People voluntarily choose to follow them. That probably means this person's a really good leader. What are the characteristics? What are the things that they do? Oh, they're really loyal. Loyalty is really important when you're a leader. Anyways, I think there's a bunch of these dimensions where a big part of the learning in the early years was just kind of drinking from the fire hose and it was such a fantastic experience. See people who are really great at their craft, understand what their superpower is, understand what they're doing. What are the actual behaviors that lead to that superpower? And then my goal that I set for myself is I'm, I'm probably not gonna be as 100% as good as you if this is your superpower but I'd like to be 80% as good as you at your superpower. And if I can be 80% as good as 10 people on each of their 10 superpowers, then I'll, I'll probably be pretty, pretty good.

    10. HSHarry Stebbings51:02

        How do you think about that in the face of there's no one way to do venture? How do you think about the two opposing, everyone has their own way and trying to learn from other people?

    11. DCDavid Cahn51:12

        Well, I totally agree on these kind of two opposing forces. I would almost describe one as, like, there is a definition of success in this business.

    12. HSHarry Stebbings51:18

        Mm-hmm.

    13. DCDavid Cahn51:18

        You're either a slugger or you are not. If you are generating billion-dollar gains, you are a slugger. Doug Leone is a slugger. Pat Grady is a slugger. Andrew Reed is a slugger. Right? So you either are slugging or you're not, and I think one of the delusions of this business is, oh, you can be good at this business without being a slugger. No. There is one definition of success in this business and that is generating billion-dollar-plus gains. On the other hand... And you can get a lot of advice and you can learn from those people and you can see what they're doing. And so, and I think, frankly, the reason I'm at Sequoia is, like, sitting next to the sluggers is pretty cool. You get to learn-

    14. HSHarry Stebbings51:47

        (laughs)

    15. DCDavid Cahn51:47

        ... a lot from them. You see Roelof, he invested in YouTube and Instagram, right? What enabled him to do that? What enabled him to see that? Especially given my whole modus operandi is this, like, let me study people and figure out what they're good at and get 80% as good. (laughs) Man, Sequoia's this, like, amazing place, right? Like drinking from the fire hose. You have all these amazing people and you get to kind of extract, okay, Pat is really disciplined. He's really good at waiting for a fat pitch. That's something to learn from Pat. Andrew is really good at connecting with founders and he deeply understands product. That's something to learn from Andrew. So there's an element where you take these sluggers, you kinda learn from them. But I think to your point, the other side of that is venture's a game where you're on the field every day. No one's on the field for you. Nobody can do this job for you. You can get all the advice, you can get all the help, you can get all the learning, but at the end of the day, you gotta be on the field. You gotta deliver a product to founders that they want. You have to be someone that they wanna work with. And you have to develop your own style to do that. And the hardest part of this job, and that's the thing that no- (laughs) you can't outsource to anybody is push comes to shove, Monday morning, you're in that partner meeting. Y- you have to put your neck on the line. Do I wanna do this investment? And something I like to say to people is, "I don't think the hardest part of venture is figuring out what are the good companies. I think the hardest part of venture is figuring out when are you gonna put your neck on the line for this company? And all good partnerships force you to put your neck on the line." That's the whole system. That's why these partnerships work is because they're testing your conviction in that way. And so conviction is the hard thing and conviction is something no one else can, can give you.

    16. HSHarry Stebbings53:12

        How does Sequoia force you to put your neck on the line?

    17. DCDavid Cahn53:15

        Look, I think the whole structure of the partnership is we are here to support you. We are here to teach you. You're here, you're gonna learn from us. But at the end of the day, you gotta make these decisions. And at the end of the day, there's a clear definition of success. And that's being a slugger. And I think the, the constraint... By the way, we talk a lot about constraints at Sequoia. Companies need constraints, right? When you have too much capital, there's no constraints. Roelof talks all the time, I- I love this story. Roelof is always telling this story about how at PayPal the best moments of PayPal were when the company was running out of money and the company just had to figure it out and we talk about the PayPal mafia. Everyone was so great. Roelof says, "Hey, we weren't that great. We were great when we had to be great." And I think Sequoia creates the same constraint. Hey, David, you get one to two investments this year. Is this one of the one to two? Are you sure about that? (laughs) and that constraint... So I think the constraint is the thing that forces you to say, "This is, this is a company I'm gonna stand for for the rest of my life," right? "And this is something that

15. 54:07 – 1:02:04

    How Sequoia Welcomes & Empowers New Partners

    1. DCDavid Cahn54:07

       I really believe in."

    2. HSHarry Stebbings54:08

       To what extent is one very nervous entering the Sequoia partnership in terms of the discussion? I ask this, I think of both Pat and Doug, which is, like, it is a very, uh, challenging environment for someone new to come into but you also want to enable someone to feel comfortable to share their thoughts, their wisdom, their experiences. How do you think about how Sequoia creates the freedom for new entrants to say what they feel?

    3. DCDavid Cahn54:32

       I think Sequoia does a good job of saying the only thing that matters is the next investment and it is equally likely to come from you as anybody else. And I think Sequoia, uh, Pat does a good job of this especially, of, like, "This is the business model. The whole business model is we need to attract good people and then you need to make good investments and we need to set you up to do that." So I think Sequoia does a good job of that. Now, at the same time, Sequoia is not a low-pressure place to go work, right? You're sitting next to people who did these incredible investments, DoorDash, Insta- Instagram, uh, Instacart as well, Zoom, Snowflake, right? So there's pressure, right? And I think it changes how you invest. It changes, like I th- I think it, like, kind of changes your brain. It rewires you. There's this great line that people at Sequoia say which is, "Sequoia hires investors, breaks them down and then rebuilds them in the Sequoia mold." I certainly feel like I've gone through that, right? Which is, are you gonna build this company for the long term? Sequoia cares so much about company building. And so I th- and I think the thing that really reshapes you is not just the company building but the, the bar for success is so, so, so high. Unless this company is one of the-... 50 most important companies in the world, you, you haven't done your job well. And so I think that, that is good pressure, right? I think that is good ... a source of, hey, this is just demanding. You, y- you need to pursue excellence at a very, very high level. And I think Sequoia does a good job of giving you the framework and tools and learnings to go do that.

    4. HSHarry Stebbings55:52

       I did this with Pat, and he gave me answers. And so I expect the same from you. (laughs)

    5. DCDavid Cahn55:56

       Okay. (laughs)

    6. HSHarry Stebbings55:57

       Who is the best sourcer in Sequoia?

    7. DCDavid Cahn56:00

       I think Sonya is very, very good. I would say Sonya.

    8. HSHarry Stebbings56:03

       Hmm.

    9. DCDavid Cahn56:03

       Sonia's really good in AI, she's super aggressive, she's out there, she's talking to founders. Sonya's phenomenal.

    10. HSHarry Stebbings56:08

        Who is the best picker? Who's the best at selection?

    11. DCDavid Cahn56:12

        Pat is very, very good. Uh, Pat is just disciplined, right? I think he waits for the fat pitch. He's waiting for the right companies. And I think that means his portfolio, as a result, is very strong.

    12. HSHarry Stebbings56:22

        I agree. What I find astonishing is his movement earlier. He was a B, C investor, and now he's moved earlier, it seems, with more Harveys of the world and how he thinks about intercepting with companies. I'm just fascinated with how he's able to sustain that quality with the movement earlier. It's easy to ... Not easy, but to pick your pitch when you're at B or C is very different to pick it when you're at C than A.

    13. DCDavid Cahn56:45

        I think that comes down to adaptability. And I think Pat is someone who's really adaptive. You look at the market landscape today, it's just different. And so I think the right strategy is this barbell approach, right, where you're doing early companies and then you're also doing late companies. Pat invested in Grow Therapy this year, right? This is a c- this is a later stage company. This is a, this is a mature company, generating cash. So I think Pat is actually good at kind of picking spots and he's picking spots 'cause he's paying attention to the market dynamics. And to have done both Harvey, which is early, and Growth Therapy, which is late, um, that's pretty amazing.

    14. HSHarry Stebbings57:14

        We, we mentioned kind of the lessons from the amazing businesses that you worked with at Coatue. When I did Pat on the show, uh, or when I spoke to Pat on the show, um, uh, (laughs) uh, he gave a g- incredible framework for how he thinks about founder assessment.

    15. DCDavid Cahn57:29

        Mm-hmm.

    16. HSHarry Stebbings57:29

        I hate the question of, "What do you look for in founders?" But my view's actually changed a lot. How do you think about founder assessment frameworks in how yours has changed?

    17. DCDavid Cahn57:39

        Man, I think about this so much. One thing that occurred to me earlier this year, a- as I've been thinking about this a lot, was I was reading ... I went down the rabbit hole. I read the Elon, uh, Walter Isaacson bio, which was just fantastic.

    18. HSHarry Stebbings57:50

        Mm-hmm.

    19. DCDavid Cahn57:50

        And so I thought, "W- hey, this Walter Isaacson guy, he's a good writer. Let me just read everything else that he's read, h- he's written." So I read his Steve Jobs bio, I read his Einstein bio, I read his Benjamin Franklin bio. So he's written all these, all these really great biographies and I was just thinking about, wow, these, these sort of historic visionaries, these amazing figures. And at the same time, obviously my day-to-day job (laughs) is I meet with a lot of founders and I ... At the time, I was organizing this dinner. It was a, I call it the hardo dinner. It was for like the most hardcore young founders, you know, who are like, "We're gonna go conquer the world." And it was a great dinner. And the day before the dinner, it's just like these two ideas were just colliding in my mind and I was j- I was like, "What do these amazing historic figures have in common with these people that I'm gonna sit down with at dinner tomorrow who have the potential to become historic figures," right? What occurred to me, I, I sort of came up with this framework. It is two dimensions, a- and I'll just describe it. There's one axis is basically science and one axis is, uh, intuition. And so that's s- sort of two by two matrix. You have science and intuition, and then you have human and technology. And so science applied to technology is kind of the easiest one, which is that's just an engineer, right? You apply science to technology, you become a good engineer. You apply science to yourself, to the human, and I think what that becomes, that's what I call this kinda hardcore mindset. Elon obviously has sort of coined that term. And sort of a hardcore person says, "How do I maximize my own potential?" They're thinking c- they're thinking consciously about that. They're thinking tangibly about that. One thing they might do is work super hard, because it turns out people who work hard generally (laughs) do produce more than people who don't work super hard. But there's other things that you might figure out. "Here's how I can maximize my efficiency, et cetera, et cetera." So I think that is sort of the science element. But the intuition element is almost more interesting and it's a thing that I've been trying to get better at and, and sort of trying to improve on and think about. So you have intuition applied to technology. That's like Ivan at Notion, right? It's like how is this supposed to look? How is this supposed to work? And, and you think about Einstein, who is this intuitive person. Everything was a thought experiment for him. It wasn't science, it was intuition. You think of Ivan and you think of Brian Chesky at Airbnb, right? "This crazy idea! We're gonna have people sleeping on everyone else's couches and renting out their apartments." I mean, eh, yes, they built great technology, but like, that intuition was just so brilliant at that time. And then thinking of the final piece, and I think this is one of the most underestimated pieces, is intuition applied to humans. And I think this is leadership. Leadership is, "Hey, I understand you. I understand what you need. I understand what motivates you." And I think that good founders and I, uh, need all four of these dimensions. And so when I look at founders, I sort of ask myself, "Hey, I think if you have one of these, you can be very good. I think if you have two of these, you can be great. I think if you have three of these, you can build a billion dollar company. You need all four of these to build a $10 billion company, $100 billion company." And so I think that's a visionary founder. And I think one founder who I've worked with that really embodies that is Paul at Supabase. I think he has all four of these qualities and it's so rare. It's something I look for and think about. And maybe one final point on this is, I don't think you start with all four (laughs) of them, right? Like, I, I do th- I believe ... I'm a big believer in like growth mindset. Like, you can decide, "I'm gonna become good at these things." And I think you can work on these things over time.

    20. HSHarry Stebbings1:00:47

        Funny, you mentioned Paul at Supabase there. I spoke to him before the show, and he mentioned continuous, uh, mornings where it was like 4:00 AM your time, I think you might have been in Singapore, uh-

    21. DCDavid Cahn1:00:57

        (laughs)

    22. HSHarry Stebbings1:00:57

        ... and he w- on with Carrie from Coatue and him just continuously. And what I actually thought, before we get to like actually how you organize time, is just like a young person's game. It's, it's easier when you're younger to do 4:00 AM every morning and have less sleep. Do you think venture is a young person's game?

    23. DCDavid Cahn1:01:17

        Look, I think there's advantages to youth and there's advantages to experience. And I think you have to, you make the most with the cards you've been dealt. And so I'm young and so I need to play the best young person's game there is. I can't play the Doug Leonie game. Doug and I are not the same. Doug is way more experienced than I am, right? And so I think the value proposition that I offer to founders is quite different.... the value proposition that I offer to founders is, I'm gonna grow with you. You're probably my age, right? Like, we're the same age. So like, over the next 20 years, we're gonna become great together. That's the value proposition that I have to offer to founders. Doug is gonna offer a different value proposition. And so, for me, it's about delivering on what you have to offer and doing the best you can with what you have.

    24. HSHarry Stebbings1:01:56

        How old are you, David?

    25. DCDavid Cahn1:01:58

        I'm 28.

    26. HSHarry Stebbings1:01:59

        Oh, fuck.

    27. DCDavid Cahn1:02:00

        (laughs) I'm the same age as you, Harry. Come on, man.

    28. HSHarry Stebbings1:02:02

        Oh, I know.

    29. DCDavid Cahn1:02:03

        Come on, Harry.

    30. HSHarry Stebbings1:02:03

        That's

16. 1:02:04 – 1:08:06

    Ranking Core Pillars of Venture: Sourcing, Selecting, Servicing

    1. HSHarry Stebbings1:02:04

       what makes me feel awful. Uh-ha. But you're so much wiser. (laughs)

    2. DCDavid Cahn1:02:09

       I don't know about that. You built a pretty successful franchise.

    3. HSHarry Stebbings1:02:12

       Dude-

    4. DCDavid Cahn1:02:12

       I could turn this interview on you. (laughs)

    5. HSHarry Stebbings1:02:13

       No, no, no, no, no. No, definitely not.

    6. DCDavid Cahn1:02:16

       I won't. I won't. (laughs)

    7. HSHarry Stebbings1:02:17

       No, okay. Alfred, every time is like, "Harry, when, when are we gonna do your show?" I'm like, "Fucking never, dude. Never." Uh. (laughs)

    8. DCDavid Cahn1:02:23

       I was looking at your birthday, Harry. I think I'm four months older than you, so I'm, I'm slightly older.

    9. HSHarry Stebbings1:02:27

       Oh, thank God. Well, then now I feel great. Um, a da- it's a really unfair question to ask. You got founders, you got market, you got product. Can you rank them one through three?

    10. DCDavid Cahn1:02:36

        Founder, market, product, easy. I don't think it's a hard question for me.

    11. HSHarry Stebbings1:02:40

        Okay, great. Well done. Sourcing, selecting, servicing. The core pillars of venture. Rank them one through three on why you're great.

    12. DCDavid Cahn1:02:48

        Well, I was gonna say an important selection is the most important. Uh, and I think, and then I think winning comes behind that. And, and, and sourcing probably is actually, eh, maybe it's tied, but I think it's probably third. I would say in terms of my strengths, I think sourcing is something that I, I, I've sort of just worked at. I, I done a lot of founder meetings. I've hustled super hard. I think in the beginning of your career, you have to get phenomenal at that. I think actually winning the deal was something that was really important early in my career because I wa- you know, at Sequo- at Sequoia, I think winning the deal is actually s- a slightly less important probably because you, you can leverage the Sequoia brand.

    13. HSHarry Stebbings1:03:18

        Is it much easier when you have the Sequoia brand?

    14. DCDavid Cahn1:03:21

        I just think the value proposition to founders is different, right? Before, the value proposition is, "Hey, you get to work with me. I'm gonna be great. I'm gonna work with you. We're gonna, we're gonna try to build this company together." Now, the value proposition is, "Hey, I could be an idiot, okay?" But like, you're gonna go to tell engineers that you're working with Sequoia, and p- engineers are probably gonna wanna work with you. Now, hopefully, I'm fantastic also, but I think even if I'm not, then the Sequoia brand itself is pretty valuable. So I think it does change the game a lot in terms of winning the deal.

    15. HSHarry Stebbings1:03:45

        Did your deal flow change immensely when moving to Sequoia? Uh, there's the-

    16. DCDavid Cahn1:03:50

        No.

    17. HSHarry Stebbings1:03:50

        ... idea that everyone just wants to, uh, everyone will send their best companies to Sequoia, and you see everything. Is that true?

    18. DCDavid Cahn1:03:57

        I think you would think that's true, and it's not true basically. I think my deal flow basically didn't change at all. And I think what that tells you is I'm just an outbound hunter. That's who I am. I'm always... The best founders don't wanna meet with VCs. That's the reality of it. You have to convince them. I, I have this phrase I like to use. Like, you have to earn the right to win. It's like a mantra that I had for many years, especially when I was coming up. And I just think you have to do the work to earn the right to even spend 30 minutes with a founder. A founder's time is super valuable. And that doesn't change at Sequoia. Founders don't wanna spend time with VCs for the sake of spending time with VCs. And all of my sort of, sort of deal flow, if you will, is just me being out there hunting in the market.

    19. HSHarry Stebbings1:04:32

        I agree. Founders don't wanna spend time with VCs, which is why you should start a podcast. 'Cause they're like-

    20. DCDavid Cahn1:04:36

        (laughs)

    21. HSHarry Stebbings1:04:36

        ... "Hi, we're not meeting investors." I'm like, "I agree, they're all twats." Uh, I'm a, I'm a media person. I'm just a podcaster. And then whenever it comes something else, I'm like, "No, I'm an investor." It's amazing how you can switch hats. Um, you mentioned kind of they don't wanna spend time with investors. What's the craziest thing you've done to win a deal, David?

    22. DCDavid Cahn1:04:54

        Look, I think you probably actually maybe can relate to this. Like, in the beginning, it's super hard, and nobody wants to give you the time of day. They ignore you. They don't respond to you. And so I think in the beginning, it was about, "How do I just get people's attention?" Once I get in front of people, then I think you can make your case, etc., but I think getting in front of people is really hard. And so one thing I used to do is I went through this phase where every time I met a founder that I really liked, I would go through their Twitter, I would figure out something they really liked, let's say some TV show, and then I'd fi- go in Cameo, and I would find the actress or actor that they, like, really liked, and I would get that person to record a video for them, and I would send them that video. In hindsight, it's kind of ridiculous. It was kind of a ridiculous thing to do. And yet, it was shockingly effective. Like, I will tell you-

Episode duration: 1:13:01