The Exchanges

Q been eight years. I'm hoping that my question asking ability has gone up in terms of quality in those eight years. Now, listen, I want to start. Fifteen billion dollars you raised today, and I was just looking at that, and I was wondering, in an age of venture today, do you have to go really big or go crafts and very small and boutique to win in venture today?

A Yeah, I mean, I think this is, um, this sounds like a bad word when I say death, but there is this kind of death of the middle that happens to a lot of asset classes in general. In venture capital, it was a tiny, tiny asset class at the beginning. Um, right now it's gotten bigger, but it's really more of the end state of a lot of these companies is huge. I mean, Sequoia used to brag about, I think it was like, 20% of the market cap of the Nasdaq was Sequoia companies. Millions of, like, you know, Apple and Oracle and all of these, these amazing names. Um, they're very, very big, right? And companies go public much, much later today, so the ability to deploy more capital, more money into, kind of, venture capital, which is no longer, um, you know, kind of sidetrack here. Series D didn't exist in, like, 1992, right? It's like, that was an IPO. Like, companies would go public. I think Amazon went public at, like, a six hundred million dollar market cap or something. Like, that was the norm. There was no Series I, Series K, Series, you know, W. You would just go, you'd raise, you know, Series A, raise Series B, raise Series C, then go public, um, and consequently, venture firms back then were very, very small, but also, the exits tended to be quite small as well. I mean, like, if, if a very, very good scenario is you have a company that goes public at a sub-billion dollar market ca…

AI assessment note: “there is this kind of death of the middle that happens”

Q being the labor displacement in the kind of one of the three kind of pinnings that you have. I completely agree. My friend Jason Lemkin said this year will be the year where we see the demonization of technology leaders, and that we see labor displacement materially showing up in labor markets. Do you think that's true, and will we see labor displacement in labor markets materially show this year?

A I'm not sure about that. I think in certain areas, for sure. Um, I mean, in general, like, I could even click up a notch, which is, if you think about SaaS, right, like, broadly speaking, I think there are kind of three types of SaaS companies right now. There are the ones that are almost impervious to everything that's happening with AI. And if anything, it's a huge tailwind because they're going to start being, they have the distribution, they're going to start adding features. And that's things like Workday and NetSuite and these things where it's like, they have the hostages, um, never going anywhere. On the other side, you have things like Zendesk, right? Where it's like, how many licenses per seat do you need of Zendesk? If now every customer support ticket can be answered automatically, you need zero license. Like their revenue could go down a hundred percent. These are very, very different. And then you have things in the middle, like Adobe, Where it's like, ooh, maybe I, now whenever I want a logo, I just go to ChatGPT. I don't go to, like, the graphics team. So maybe you'll need fewer graphics designers. Maybe you'll need, you know, Zendesk, you'll need fewer customer support people. That probably is true, right? Like, there are gonna be certain areas that will get hit harder than others, but what technology has always done is, you know, people shift into other jobs, …

AI assessment note: “I'm not sure about that. I think in certain areas, for sure.”

Q What product does Andreessen not have today that you would most like Andreessen to have? You mentioned GC having, like, the fund there that does that roll-up play. They've got the, like, consumer performance marketing fund. I can't remember what that's called, but what product do you not have that you'd most like to have?

A I probably think, um, I don't know if we would ever do this, but, uh, something around credit for a lot of our companies. So, you know, we have equity products, but we don't have debt products. Um, and they have very different return profiles, obviously, but every one of our companies, they need, you know, general capitalists actually has one of these. They have a credit fund. Um, so either for customer acquisition or if you're FinTech and doing lending, so that, that would be interesting. But in general, we just kind of, we listen, we don't want to be at odds with our entrepreneurs. Like there's a very solid reason why we don't have that, which is like, oh, you didn't pay back the bill. I need to go foreclose. Like that's a, as a venture capital firm, like you can earn a thousand X on a, on a winner. You don't really want to like kind of, Beat up the, ah, the companies that are struggling, and that's kind of what the credit instrument needs to do, but I, I think it's a good product.

AI assessment note: “something around credit for a lot of our companies.”

Q No, I love it. Let's finish today on your most recent investment and why you said yes, publicly announced, that is. Do you think it's a good thing to publicly announce?

A Well, I think it can, it can be, and it cannot be. So going back to the inflection point and like, how do you find almost non-defensible, but very, very unique points of distribution? If you're doing a really good job, the last thing you want to do is tell the world that you're doing a really good job and just create competition. I think you want to do it when you get to a certain inflection point where you realize, all right, we're past the point of no return. Like we're not going to get 20 competitors that just emerge out of the woodwork. And start copying what we're doing, or even if we do, it doesn't matter because we have our catch-up is in front of every single chip or fry stand in the UK. Like, great. We're in good shape now because we can go talk about like our phenomenal margins on catch-up. Nobody's going to be able to catch up with us. So probably the one that I made most recently that's been announced, I've done a couple after that, is a company called Propel. And what Propel does is they basically digitize access to the, it's called the SNAP program, the Supplemental Nutrition Assistance Program, also known Colloquially is food stamps in the U S so it turns out that tens of millions of people in the United States, they're on food stamps. Um, they're the most economically needy in the country. It's a really valuable program, but it involves interfacing with governme…

AI assessment note: “Well, I think it can, it can be, and it cannot be.”

Q No, I love it. Let's finish today on your most recent investment and why you said yes, publicly announced, that is. Do you think it's a good thing to publicly announce?

A Well, I think it can, it can be, and it cannot be. So going back to the inflection point and like, how do you find almost non-defensible, but very, very unique points of distribution? If you're doing a really good job, the last thing you want to do is tell the world that you're doing a really good job and just create competition. I think you want to do it when you get to a certain inflection point where you realize, all right, we're past the point of no return. Like we're not going to get 20 competitors that just emerge out of the woodwork. And start copying what we're doing, or even if we do, it doesn't matter because we have our catch-up is in front of every single chip or fry stand in the UK. Like, great. We're in good shape now because we can go talk about like our phenomenal margins on catch-up. Nobody's going to be able to catch up with us. So probably the one that I made most recently that's been announced, I've done a couple after that, is a company called Propel. And what Propel does is they basically digitize access to the, it's called the SNAP program, the Supplemental Nutrition Assistance Program, also known Colloquially is food stamps in the U S so it turns out that tens of millions of people in the United States, they're on food stamps. Um, they're the most economically needy in the country. It's a really valuable program, but it involves interfacing with governme…

AI assessment note: “So probably the one that I made most recently that's been announced, is a company called Propel.”

Q David George before the show. And he said, one thing he's never talked about publicly. I don't think that he's a phenomenal master on his advice on selling companies. Ask him about that. I know it's a bit broad and random, but I do want to touch on it because David said I had to, what's your biggest advice on selling companies having seen so many and living it yourself?

A Yeah, so I'd say a couple things. Um, you know, this is a very highly choreographed dance, so you can't just say, oh, I should, like, so if you're raising money, um, you're like, oh, I should raise money. I have the best metrics ever. I'm going to talk to five firms, and they're going to compete to the death over winning my deal. Like, that was my experience with my Series B at trial pace. So it's like, ah, so it's like, and, and kind of corp dev is like, I'm either raising money or selling my company. It's the same thing, right? No, it's completely different. Um, If you're selling your company, you have to spend, you know, in many cases years, like, just getting to know people at the potential acquirer. It's never the CEO, unless you're like, you know, what, you know, Yankum at WhatsApp. Like, let's just say that you have a company, you, you do something amazing, um, somebody at Salesforce should buy it, um, you would rather go public, but you're like, ooh, you, you kind of see the writing on the wall, like, I'm going to hit a wall in a year and a half. What you should start doing then Is, um, I kind of call it a background process. Like if you know what cron is in, in Unix terms, right? It's like you should have a little cron job where it's like five percent of your time as CEO should just be like getting to know people at the three or four companies that might buy you. You n…

AI assessment note: “five percent of your time as CEO should just be like getting to know people”

Q What product does Andreessen not have today that you would most like Andreessen to have? You mentioned GC having, like, the fund there that does that roll-up play. They've got the, like, consumer performance marketing fund. I can't remember what that's called, but what product do you not have that you'd most like to have?

A I probably think, um, I don't know if we would ever do this, but, uh, something around credit for a lot of our companies. So, you know, we have equity products, but we don't have debt products. Um, and they have very different return profiles, obviously, but every one of our companies, they need, you know, general capitalists actually has one of these. They have a credit fund. Um, so either for customer acquisition or if you're FinTech and doing lending, so that, that would be interesting. But in general, we just kind of, we listen, we don't want to be at odds with our entrepreneurs. Like there's a very solid reason why we don't have that, which is like, oh, you didn't pay back the bill. I need to go foreclose. Like that's a, as a venture capital firm, like you can earn a thousand X on a, on a winner. You don't really want to like kind of, Beat up the, ah, the companies that are struggling, and that's kind of what the credit instrument needs to do, but I, I think it's a good product.

AI assessment note: “something around credit for a lot of our companies... we don't have debt products.”

Q to you, but I know quite a bit about lending, where it's quite easy for me today to see a lending business and go, blech, fucking horrible. It's a hard market. I don't want to be there. Look at Lending Club. Look at the market cap there. Very dismissive. As many were with Stripe when they knew payments. How do you prevent yourself knowing too much that it's a negative?

A I think this is a great question. And this is the number one thing that that, so I do a couple of things. Number one, if it's like an ad tech company, I know a lot about ad tech. I know a lot about payments. I will force somebody else to join me for the pitch. That is like a beginner's mindset mind. Um, so I think that's one is just like have a sparring partner internally. That has that, you know, what if it works? You always have to be like, what if it works? So that's number one. Number two is, I, I like to ask the entrepreneurs, like, what is different? And the thing that's different, like, the, the way that Patrick, the reason why Patrick, um, and John made Stripe work partially is it's like, they just believed that a great number of new companies will be created, and they're gonna pick the best product, and they're gonna have the best product. And actually, this informs a big part of my investment thesis now. I mean, I call it Greenfield. But, um, there's a saying that I use a lot, which is the best companies have hostages, not customers, right? It's like, you'll appreciate this if you're an enterprise SaaS guy, right? It's like the best companies have hostages, not customers. So if there's a company that has something marginally better than Workday, right? They're not going to go, like, Workday has hostages. They don't have customers. They're not going to go be able to se…

AI assessment note: “I will force somebody else to join me for the pitch. That is like a beginner's mindset”

Q I actually am in trouble with my team because I just tweeted today. Series A is the worst place to be investing. Company progression is minimal. Price is four to five X, the seed price. And we're paying a 150 to 200 X ARR with little signs of product market fit. Do you agree with me? It's the worst place to be investing.

A Well, I think the problem is that there's the nomenclature, which kind of varies company to company. So, like, when I started TrialPay, we raised, our Series A was 3.1 million on 9.5 million pre. Um, and that was expensive. I remember, like, arguing with the partner at Battery. It's like, this is the most expensive deal we've done. This was 2006. At, at Site Advisor, I think we raised 2.7 on 2.7 pre. So even lower, so that, you know, hey, hence, hence he was right. So now you have a pre-seed, a seed, a seed extension, a seed extension two, like, what is a series A, right? There's not like this, like normally a series A would be like the first institutional round of money. Now there's so much variance because like, oh, there's the series A where it's like five superstars out of open AI and they need tons of money for compute. No moral hazard on that. You're not going to go spend money on people. You're going to spend money on GPUs. That's one form of Series A. Another form of Series A is, like, I just did a Series A where the company had, like, ten million dollars of ARR when I invested in it. So it's just all over the place. So I, I think it's just hard to kind of cast a generality. Um, there are certainly ones where, like, I used to call this the Series B trap, but again, I think the nomenclature has shifted. Like, I would have agreed with your team if you called it the Series…

AI assessment note: “So I, I think it's just hard to kind of cast a generality.”

Q I actually am in trouble with my team because I just tweeted today. Series A is the worst place to be investing. Company progression is minimal. Price is four to five X, the seed price. And we're paying a 150 to 200 X ARR with little signs of product market fit. Do you agree with me? It's the worst place to be investing.

A Well, I think the problem is that there's the nomenclature, which kind of varies company to company. So, like, when I started TrialPay, we raised, our Series A was 3.1 million on 9.5 million pre. Um, and that was expensive. I remember, like, arguing with the partner at Battery. It's like, this is the most expensive deal we've done. This was 2006. At, at Site Advisor, I think we raised 2.7 on 2.7 pre. So even lower, so that, you know, hey, hence, hence he was right. So now you have a pre-seed, a seed, a seed extension, a seed extension two, like, what is a series A, right? There's not like this, like normally a series A would be like the first institutional round of money. Now there's so much variance because like, oh, there's the series A where it's like five superstars out of open AI and they need tons of money for compute. No moral hazard on that. You're not going to go spend money on people. You're going to spend money on GPUs. That's one form of Series A. Another form of Series A is, like, I just did a Series A where the company had, like, ten million dollars of ARR when I invested in it. So it's just all over the place. So I, I think it's just hard to kind of cast a generality. Um, there are certainly ones where, like, I used to call this the Series B trap, but again, I think the nomenclature has shifted. Like, I would have agreed with your team if you called it the Series…

AI assessment note: “So I, I think it's just hard to kind of cast a generality.”

Q and a half million. Dude, raising their Series A, like, you know, eight or 10 on fifty-ish, you know, memory serves me correct, was a bear. Was fucking horrible. And I was just like, oh my god. The triple, triple, double, double is so dead. Like we're in lovable as well. That obviously is a completely different fundraise journey. Is the triple, triple, double, double dead? I don't think so.

A Yeah, no, I, I think, um, I mean, it might, it might be harder for a certain set of people that are maniacally focused on like growth over everything else, but like what really matters is growth and stickiness and the triple, if you're a triple, triple, double, double with like, you know, terrible retention data. Like, okay, that, that's gonna be very hard. But if you actually have, you know, again, system of record, or in that case, you know, it sounds like vertical operating system, that should not be hard. If, if you find the right, I mean, like I, I would do, I, I, I love those things, right? Like I would much rather have a slower growing, you know, permanent system of record that will never get ripped out than the fastest growing thing on the planet that has 9000 competitors that are all built and lovable by 17 year olds. It's like, I, I think there's no comparison. I mean, there, there are plenty of people that would be attracted to both. Would be my answer. I'm, I'm surprised that it, that was, it was as challenging as you, as you portray it.

AI assessment note: “I'm surprised that it was as challenging as you portray it.”

Q and have a higher win rate, specifically with your profile of fund, because I get it in other funds where you don't have the ability to follow on and lead the B, the C, the D. You may even not be able to do the pro radars, in which case I get that thinking. But when you can, why are we not having a higher win rate and doing 10%?

A Well, I mean, this is actually one of the things that we looked at, um, because I, I kind of feel like my job here is kind of quasi portfolio manager. So I, I run our apps fund, right? So we have seven different funds and my job is to make sure that like that fund is as successful as possible. Um, and you know, we're winning the right deals that we, and it's like, if we just say, Hey, everybody win every single deal, just win every deal. It doesn't matter. That's all I'm going to optimize for. And we end up with five percent checks in every series A, like, you know, that's not going to work, right? We can win every deal that way. What is the front? How far on this curve can you go? Um, and it's the, again, it's the exact inverse conversation that an entrepreneur is having, where it's like, I want a tier one investor. I want, you know, an amazing specialist. I want whatever I want on, you know, this person that I want on my board. What is the least amount that I can give up to get an amazing person? And they would love to get five percent A round deals done. But they're like, oh, wait a minute. Like, that's not going to work. And like that, that's the, the tension between the two. So I, I agree with you, but I think, you know, where do you, it's like Zeno's paradox. You know, that is right. It's like, you will never get to the destination if you go halfway each time. Like, is it…

AI assessment note: “if we just say, Hey, everybody win every single deal... that's not going to work”

Q hope it's not too forward and you can say, dude, don't want this in there. Um, by all means, I'm not a journalist, but like you do the successive B. Because you lose the A. When you sit down and we're sitting down as a team, how do we reflect on that? When you reflect on like a Relit review, what was the takeaway from that when you sat down?

A Well, I mean, there are a lot of deals that we, we lose because we're not willing to kind of go the distance on price. I mean, that, that is a common thing where it's like, did we really lose it? Like this has happened to us a number of times. It's like, all right, we want to do the deal. And this is again, like consensus and non-consensus, a lot of times the difference is just on price. Um, or ownership, right? Like, if we had shown up and said, hey, we'll do 10% of this company for an A round, like, we could win every deal. It's actually, I think one of the, one of the competing elements that has shown up, I'm, I'm interested to watch how, uh, Standard Capital does. This is kind of the YC offshoot. Yeah, it's like, um, I'm going to take 10%. Like, that's very, very bad for big funds, because in order to make the math work for a big fund, you have to have high ownership, and you know that your ownership will get Depleted or will, will get diluted over time as option pool expansions happen, even if you take your pro rata in every single successive round. So, I mean, we can win all these deals, but a lot of times, um, you know, I, I, I am much more preoccupied with ownership, um, at the A, because we're buying it out of the money call option. And, and the reason why I, I kind of tell this story is because, uh, you know, there, there's, there's something that I've used as a bench…

AI assessment note: “a lot of deals that we, we lose because we're not willing to kind of go the distance on price.”

Q and a half million. Dude, raising their Series A, like, you know, eight or 10 on fifty-ish, you know, memory serves me correct, was a bear. Was fucking horrible. And I was just like, oh my god. The triple, triple, double, double is so dead. Like we're in lovable as well. That obviously is a completely different fundraise journey. Is the triple, triple, double, double dead? I don't think so.

A Yeah, no, I, I think, um, I mean, it might, it might be harder for a certain set of people that are maniacally focused on like growth over everything else, but like what really matters is growth and stickiness and the triple, if you're a triple, triple, double, double with like, you know, terrible retention data. Like, okay, that, that's gonna be very hard. But if you actually have, you know, again, system of record, or in that case, you know, it sounds like vertical operating system, that should not be hard. If, if you find the right, I mean, like I, I would do, I, I, I love those things, right? Like I would much rather have a slower growing, you know, permanent system of record that will never get ripped out than the fastest growing thing on the planet that has 9000 competitors that are all built and lovable by 17 year olds. It's like, I, I think there's no comparison. I mean, there, there are plenty of people that would be attracted to both. Would be my answer. I'm, I'm surprised that it, that was, it was as challenging as you, as you portray it.

AI assessment note: “what really matters is growth and stickiness and the triple”

Q But I do want to touch on a hot topic being ICOs. So we did do a little Snapchat survey. We got 184 questions in two hours. 176 were on ICOs for you. So let's start on ICOs and more broadly blockchain tech. Starting with underlying technology being the blockchain itself, how do you look to assess the question of public versus private blockchain and the kind of ensuing debate?

A So I'll take a very, not controversial approach, but one that I think is somewhat direct, which is, I was watching the US Open tennis tournament, and there were all these advertisements from IBM, and IBM says, we have developed something that keeps your food, like your tomatoes, safe by putting it on the blockchain, which is the most absurd sentence that I've ever heard in my life. When I first started at this firm, I met with a large CEO. It was one of the biggest banks in North America, and he said, you know what, I don't like Bitcoin, but I like Blockchain. But I want the blockchain to be in my office and not available to anybody else entirely within our own data center for our own data. And I said, oh, actually, you know, there's a company that does that. He's like, oh, really? Which one? And I said, well, they're actually, they're in Redwood Shores. They have these big, five-tall office buildings. And he's like, oh, what's it called? I said, it's called Oracle. And I was joking, of course. Actually, I wasn't joking, because a lot of people that say, I want a blockchain, the same people that a few years ago were saying, we want big data, which in a couple years were saying, like, oh, AI is very important to us. I think it's become a buzzword that almost means nothing. They mean something, but for most cases, there's no reason not to use a centralized database. There are a f…

AI assessment note: “for most cases, there's no reason not to use a centralized database.”

Q And then how did that early entrepreneurial start then translating to VC later on?

A Well, there are many, many stops in between. So I started a couple of venture-backed companies in between. So from the shareware business that I had, one of the things that I learned was that most people don't want to pay for digital goods. So you're willing to pay five dollars for a bottle of water at a sporting event. I mean, you might not be willing to, but you have to. That's your only option. People pay five dollars for that. To pay five dollars for a song on iTunes, that would be, that would be ridiculous. Nobody would want to pay that. And I found this to be true with my shareware, which is I had one product that was used by millions of people and maybe one percent of people would pay. So I had this idea of, all right, I'll let the other 99% pay me effectively by trying something else. So maybe they'll sign up for Netflix. Maybe they'll go buy a pair of jeans at the Gap. Maybe they'll go switch their car insurance to Geico. All of those companies will pay money to acquire new customers. It's the whole basis of online advertising. And these are called affiliate programs. So back in my shareware days, I basically came up with this idea for trial pay, which was a venture backed company that I started. And actually Chris Dixon and I started a company called fraud eliminator, which became site advisor, which was also venture backed, but all of these different shareware and so…

AI assessment note: “there are many, many stops in between. So I started a couple of venture-backed”

Q It's so great when you have a show like this. My question to you is, you said hostages, not customers. How should I think about that then in a world of cursor or Any of the foundation models, your anthropics or your open AIs, where they are customers, not hostages. They can switch very easily. The promiscuity of customers has never been higher. How should we think about that?

A It's, it's a really good question. I mean, I think this is where behind every technology revolution and kind of go back to like Silicon, then the personal computer, then the internet, then kind of internet two point O where you could write to the internet, things like things like Facebook and Google and YouTube, then mobile, then cloud. There's always been an infrastructure layer and an application layer. So, you know, you go back, like the infrastructure layer for PCs was, I don't know, like Microsoft and Apple, like the operating system players, the infrastructure player for the internet was like Cisco and Akamai. Um, the infrastructure player for everything AI are all of these backend model, right, you know, providers, and then there's the application layer on top. So if I do something, you know, we were talking about Ask Leo, right? Like that's an application layer company. They look, I mean, if I were Vlad, I would love to be promiscuous with all the backend models because I should be. And then the infrastructure players are like, oh shit, you know, all of our customers are being promiscuous. Let's figure out how we specialize in a particular I imagine that, like, that's why Anthropic, I imagine, has gotten very good at coding. Um, but it's kind of, the application layer tends to be a little bit stickier, but the problem is you might have 9000 competing companies at the ap…

AI assessment note: “I would love to be promiscuous with all the backend models because I should be”

Q have paid 300 and doubled them. Because when I map out 18 months time, I looked at their revenue projections, and in 18 months time, when they need to go raise, their revenues would have been so much that I could still see a three X on that 300. Idiot. That's how I get comfortable with paying up for something. How do you get comfortable preemptively paying up so much?

A I mean, I think it's the same, it's the same math, but it's, it's dangerous on both sides, right? It's like, um, I always have this speech that works, you know, maybe one time out of a hundred that I give it, which is kind of like the Spider-Man speech of like, with great capital comes great responsibility, and if you raise it too high of a price, you're fucked, because I lived this, let me tell you my story, I raised it this price for my Series C, then I had like Google that wanted to buy me, but it was like at the same price, so therefore it tanked the thing, and then my next round, everybody asked me what was the price of my last round, and nobody wants to invest, like, I go tell this story, um, I can introduce the founder to 10 other founders that have lived the exact same thing. It's like, I wish I hadn't raised my round at such a high price, but who starts a company, right? Like, let's just think about this for a second. The people that start a company are irrationally exuberant. Like, if they thought that the company was going to fail, if they thought they had a zero percent chance of raising a Series B, they wouldn't start the fucking company, right? So that's why the speech doesn't work. Um, because I always tell people, like, hey, you know, the reason why you shouldn't raise your Series A, like, there, there was a deal that I guess we should have done, candidly, becau…

AI assessment note: “I think it's the same, it's the same math, but it's, it's dangerous”

Q And then how did that early entrepreneurial start then translating to VC later on?

A Well, there are many, many stops in between. So I started a couple of venture-backed companies in between. So from the shareware business that I had, one of the things that I learned was that most people don't want to pay for digital goods. So you're willing to pay five dollars for a bottle of water at a sporting event. I mean, you might not be willing to, but you have to. That's your only option. People pay five dollars for that. To pay five dollars for a song on iTunes, that would be, that would be ridiculous. Nobody would want to pay that. And I found this to be true with my shareware, which is I had one product that was used by millions of people and maybe one percent of people would pay. So I had this idea of, all right, I'll let the other 99% pay me effectively by trying something else. So maybe they'll sign up for Netflix. Maybe they'll go buy a pair of jeans at the Gap. Maybe they'll go switch their car insurance to Geico. All of those companies will pay money to acquire new customers. It's the whole basis of online advertising. And these are called affiliate programs. So back in my shareware days, I basically came up with this idea for trial pay, which was a venture backed company that I started. And actually Chris Dixon and I started a company called fraud eliminator, which became site advisor, which was also venture backed, but all of these different shareware and so…

AI assessment note: “there are many, many stops in between. So I started a couple of venture-backed companies”

Q I love that unburdened by what has been memo that is there. Final one for you, dude. What does venture look like in five years time? When we look at the fifteen billion dollars that you raised today, I mean, it is obscene to even think that would happen five years ago when we go back. What does it look like five years out?

A I think it ends up eating even more of the world. So, um, this is kind of going back to To Mark's essay around software eats the world, that, that largely has happened. As I mentioned, like the five biggest companies on earth, they're technology companies, which was un, like unthinkable in 2005. Like technology companies were little service providers to big companies like banks and oil companies, right? So I, I think this momentum of kind of everything becomes a software company. It, it kind of goes into this like thesis too, that I mentioned around software does the job of labor. You're going to have all of these areas where it's like, there's going to be like, you know, Toast, Vertical SAS proved this, or kind of V-one. It's like, oh, how is Toast worth twenty billion dollars? You're gonna have a lot of things like this, where it's like brand new markets that have grown like crazy. Um, AI is now allowing software and technology to do so many things that it didn't do before, and this is before even things like robotics. Like, if robots actually work, wow, like, now you've expanded the market like another hundred x. So I, I just, I'm just so bullish on the ability of technology, um, To create enduring value. So, uh, you know, my, my, my guess and my hope is that it's going to go up into the right.

AI assessment note: “I think it ends up eating even more of the world.”

Q and have a higher win rate, specifically with your profile of fund, because I get it in other funds where you don't have the ability to follow on and lead the B, the C, the D. You may even not be able to do the pro radars, in which case I get that thinking. But when you can, why are we not having a higher win rate and doing 10%?

A Well, I mean, this is actually one of the things that we looked at, um, because I, I kind of feel like my job here is kind of quasi portfolio manager. So I, I run our apps fund, right? So we have seven different funds and my job is to make sure that like that fund is as successful as possible. Um, and you know, we're winning the right deals that we, and it's like, if we just say, Hey, everybody win every single deal, just win every deal. It doesn't matter. That's all I'm going to optimize for. And we end up with five percent checks in every series A, like, you know, that's not going to work, right? We can win every deal that way. What is the front? How far on this curve can you go? Um, and it's the, again, it's the exact inverse conversation that an entrepreneur is having, where it's like, I want a tier one investor. I want, you know, an amazing specialist. I want whatever I want on, you know, this person that I want on my board. What is the least amount that I can give up to get an amazing person? And they would love to get five percent A round deals done. But they're like, oh, wait a minute. Like, that's not going to work. And like that, that's the, the tension between the two. So I, I agree with you, but I think, you know, where do you, it's like Zeno's paradox. You know, that is right. It's like, you will never get to the destination if you go halfway each time. Like, is it…

AI assessment note: “if we just say, Hey, everybody win every single deal... that's not going to work”

Q to you, but I know quite a bit about lending, where it's quite easy for me today to see a lending business and go, blech, fucking horrible. It's a hard market. I don't want to be there. Look at Lending Club. Look at the market cap there. Very dismissive. As many were with Stripe when they knew payments. How do you prevent yourself knowing too much that it's a negative?

A I think this is a great question. And this is the number one thing that that, so I do a couple of things. Number one, if it's like an ad tech company, I know a lot about ad tech. I know a lot about payments. I will force somebody else to join me for the pitch. That is like a beginner's mindset mind. Um, so I think that's one is just like have a sparring partner internally. That has that, you know, what if it works? You always have to be like, what if it works? So that's number one. Number two is, I, I like to ask the entrepreneurs, like, what is different? And the thing that's different, like, the, the way that Patrick, the reason why Patrick, um, and John made Stripe work partially is it's like, they just believed that a great number of new companies will be created, and they're gonna pick the best product, and they're gonna have the best product. And actually, this informs a big part of my investment thesis now. I mean, I call it Greenfield. But, um, there's a saying that I use a lot, which is the best companies have hostages, not customers, right? It's like, you'll appreciate this if you're an enterprise SaaS guy, right? It's like the best companies have hostages, not customers. So if there's a company that has something marginally better than Workday, right? They're not going to go, like, Workday has hostages. They don't have customers. They're not going to go be able to se…

AI assessment note: “I will force somebody else to join me for the pitch”

Q bad answer to your question, because the biggest mistakes in venture have been when you underestimate market size and you don't see what it can be. But I'd sit down with you and I'd go, ok, 10% entry, five percent on exit, assuming a 50% dilution. Do we think this can reasonably be a fifteen billion dollar company? If so, that is a number that returns the fund with comfort.

A I know, but the problem is it's kind of garbage in, garbage out. It's like, you can always say that for something, because otherwise you're like, oh wow, I underestimated the size of the black car market, you know, hence I'm not going to do, it's hard. I mean, the way that I do it, just kind of to be pithy about it, is like, we either want to buy any percent, any percent, right, of something that could work, that is absolutely working, or high ownership of something that could work. If you really kind of draw a line of like, you have to bifurcate the market. It's like, Facebook, Um, if you look at that round, I think Greylock put twenty-five million into Facebook at a, um, actually, I think the round was maybe twenty-five million at 500. I think that was the B round for Facebook split between Meritech and Greylock, but that was absolutely working, right? So it's like, are they getting 10%? No. Are they getting five percent? No. Like, but it's like the market winner, and things can go wrong, but like, holy shit, it's absolutely working. And like, I don't see that many things that look like that, but when they, when you do, you throw away all the rules, or it's like this Is not working, but this person looks like a super genius. Um, they have high agency, they can get, they can materialize labor capital and customers, but it's not working yet, right? So like, I have to have high …

AI assessment note: “I know, but the problem is it's kind of garbage in, garbage out.”

Q hope it's not too forward and you can say, dude, don't want this in there. Um, by all means, I'm not a journalist, but like you do the successive B. Because you lose the A. When you sit down and we're sitting down as a team, how do we reflect on that? When you reflect on like a Relit review, what was the takeaway from that when you sat down?

A Well, I mean, there are a lot of deals that we, we lose because we're not willing to kind of go the distance on price. I mean, that, that is a common thing where it's like, did we really lose it? Like this has happened to us a number of times. It's like, all right, we want to do the deal. And this is again, like consensus and non-consensus, a lot of times the difference is just on price. Um, or ownership, right? Like, if we had shown up and said, hey, we'll do 10% of this company for an A round, like, we could win every deal. It's actually, I think one of the, one of the competing elements that has shown up, I'm, I'm interested to watch how, uh, Standard Capital does. This is kind of the YC offshoot. Yeah, it's like, um, I'm going to take 10%. Like, that's very, very bad for big funds, because in order to make the math work for a big fund, you have to have high ownership, and you know that your ownership will get Depleted or will, will get diluted over time as option pool expansions happen, even if you take your pro rata in every single successive round. So, I mean, we can win all these deals, but a lot of times, um, you know, I, I, I am much more preoccupied with ownership, um, at the A, because we're buying it out of the money call option. And, and the reason why I, I kind of tell this story is because, uh, you know, there, there's, there's something that I've used as a bench…

AI assessment note: “there are a lot of deals that we, we lose because we're not willing to”

Q But I do want to touch on a hot topic being ICOs. So we did do a little Snapchat survey. We got 184 questions in two hours. 176 were on ICOs for you. So let's start on ICOs and more broadly blockchain tech. Starting with underlying technology being the blockchain itself, how do you look to assess the question of public versus private blockchain and the kind of ensuing debate?

A So I'll take a very, not controversial approach, but one that I think is somewhat direct, which is, I was watching the US Open tennis tournament, and there were all these advertisements from IBM, and IBM says, we have developed something that keeps your food, like your tomatoes, safe by putting it on the blockchain, which is the most absurd sentence that I've ever heard in my life. When I first started at this firm, I met with a large CEO. It was one of the biggest banks in North America, and he said, you know what, I don't like Bitcoin, but I like Blockchain. But I want the blockchain to be in my office and not available to anybody else entirely within our own data center for our own data. And I said, oh, actually, you know, there's a company that does that. He's like, oh, really? Which one? And I said, well, they're actually, they're in Redwood Shores. They have these big, five-tall office buildings. And he's like, oh, what's it called? I said, it's called Oracle. And I was joking, of course. Actually, I wasn't joking, because a lot of people that say, I want a blockchain, the same people that a few years ago were saying, we want big data, which in a couple years were saying, like, oh, AI is very important to us. I think it's become a buzzword that almost means nothing. They mean something, but for most cases, there's no reason not to use a centralized database. There are a f…

AI assessment note: “for most cases, there's no reason not to use a centralized database.”

Q two, five x on two 50, than, I don't know, 15 x on 10, or whatever it is? Um, Yes, but there's an opportunity cost of dollars, and for an endowment fund, they are able to put it into the smaller fund, and so do you accept with that then, that you just scale out of certain LPs, and it's no longer the best risk-adjusted place to put money then?

A Well, I bet I, I think it's a, obviously you can't, you can't disprove a, an unknown future. Um, but I would pause it to say that if you were trying to find pick and win the best deals, and maybe you disagree with me on like the kind of the small specialist, um, or a large generalist, but who wins the best consensus deals? Every now and then there will pop up a non-consensus deal that everybody thinks is terrible. Nobody wants to, Sequoia doesn't want to do it. We don't want to do it. You don't want to do it. Nobody wants to do it. Um, and then it ends up being a thousand X and then somebody who is not the best known venture firm, you know, ended up winning that deal or being sold that deal, I should say. And then it ends up with a, with a great return. But a lot of the best deals will go to the best firms. Like that's, what's very different about venture capital than like private equity. Like if, if you and I are trying to take a public company private, You know, you're KKR and I'm Blackstone. We're both trying to, you know, take over RJR and Abisko or something like that. They're just going to sell to whoever offers them the highest price per share. I mean, they have to. Whereas in venture capital, as you know, you have to win the hearts and minds of the entrepreneur and win that deal. And a lot of the best deals are somewhat obvious. Like, it's not surprising. Like, everybod…

AI assessment note: “But a lot of the best deals will go to the best firms.”

Q When we look at companies going from one to 20 to 30 to 40, there's actually quite a few that do that today. Before that was completely unheard of. How much weight should we place on revenue growth today versus not? And is there a world where these companies that are going from one To three or four, three or four, right? This used to be good. I'd left behind.

A So if, if you want to know the, the three investment theses that I have for our fund, um, I'll, I mean, this is exactly what I told LPs, and it will answer your question in a second. I think we have, we have three. We have one, which is we invest in system, like I call it Greenfield Bingo, and most of the green, like these are existing software companies, but selling to new companies as opposed to, you know, selling to the hostages that will never leave. They tend to be systems of record. Right. Like the re or vertical operating system. So like the reason why real it, I love that company so much. That's never going to grow like zero to a hundred in like a month, but it is very, very sticky revenue. Like once you're on, like NetSuite has hostages, not customers. They're not going to leave. You know, if this can, if, if real it can sell into every new company, like they're going to do great. The revenue growth will be slower, but it will be so sticky and they have infinite option value and adding like, Hey, do you want to have a, Collections AI agent that runs on top of, you know, overdue invoices, blah, blah, blah. And that's, like, optionality on top of your sticky system of records. So number one is Greenfield kind of systems of record. Number two, and this goes to the fastest growing companies in the world that you're talking about, is, like, software that does the job of lab…

AI assessment note: “The revenue growth will be slower, but it will be so sticky”

Q It's so great when you have a show like this. My question to you is, you said hostages, not customers. How should I think about that then in a world of cursor or Any of the foundation models, your anthropics or your open AIs, where they are customers, not hostages. They can switch very easily. The promiscuity of customers has never been higher. How should we think about that?

A It's, it's a really good question. I mean, I think this is where behind every technology revolution and kind of go back to like Silicon, then the personal computer, then the internet, then kind of internet two point O where you could write to the internet, things like things like Facebook and Google and YouTube, then mobile, then cloud. There's always been an infrastructure layer and an application layer. So, you know, you go back, like the infrastructure layer for PCs was, I don't know, like Microsoft and Apple, like the operating system players, the infrastructure player for the internet was like Cisco and Akamai. Um, the infrastructure player for everything AI are all of these backend model, right, you know, providers, and then there's the application layer on top. So if I do something, you know, we were talking about Ask Leo, right? Like that's an application layer company. They look, I mean, if I were Vlad, I would love to be promiscuous with all the backend models because I should be. And then the infrastructure players are like, oh shit, you know, all of our customers are being promiscuous. Let's figure out how we specialize in a particular I imagine that, like, that's why Anthropic, I imagine, has gotten very good at coding. Um, but it's kind of, the application layer tends to be a little bit stickier, but the problem is you might have 9000 competing companies at the ap…

AI assessment note: “the application layer tends to be a little bit stickier”

Q have paid 300 and doubled them. Because when I map out 18 months time, I looked at their revenue projections, and in 18 months time, when they need to go raise, their revenues would have been so much that I could still see a three X on that 300. Idiot. That's how I get comfortable with paying up for something. How do you get comfortable preemptively paying up so much?

A I mean, I think it's the same, it's the same math, but it's, it's dangerous on both sides, right? It's like, um, I always have this speech that works, you know, maybe one time out of a hundred that I give it, which is kind of like the Spider-Man speech of like, with great capital comes great responsibility, and if you raise it too high of a price, you're fucked, because I lived this, let me tell you my story, I raised it this price for my Series C, then I had like Google that wanted to buy me, but it was like at the same price, so therefore it tanked the thing, and then my next round, everybody asked me what was the price of my last round, and nobody wants to invest, like, I go tell this story, um, I can introduce the founder to 10 other founders that have lived the exact same thing. It's like, I wish I hadn't raised my round at such a high price, but who starts a company, right? Like, let's just think about this for a second. The people that start a company are irrationally exuberant. Like, if they thought that the company was going to fail, if they thought they had a zero percent chance of raising a Series B, they wouldn't start the fucking company, right? So that's why the speech doesn't work. Um, because I always tell people, like, hey, you know, the reason why you shouldn't raise your Series A, like, there, there was a deal that I guess we should have done, candidly, becau…

AI assessment note: “it's the same math, but it's, it's dangerous on both sides”

Q Every LP says the canonical wisdom and the theory of venture As you scale, performance goes down. Do you legitimately think then that with the expansion of these markets, you can maintain five X plus net funds at scale?

A Well, I think the difference though is that imagine that you're an LP and you have a billion dollars to invest. Would you rather get, would you rather invest fifty million dollars and get a five X on that? Or would you rather invest all billion and get a three X on that? And the answer is you'd rather get a three X on a billion than a five X on, you know, five million. Or, um, one of my good friends is this guy, Mickey Malka at Ribbit. I was lucky to be an investor in his fund one personally. And it's like, that was like a 55 X fund, um, on, I think it was like an eighty five million dollar fund, but 55 X, like that's insane. But, you know, at some point, um, you, you could ask Mickey this too. It's like, you're better off with like a five X on like a very, very large fund. Like the harder thing to do is to just return gross dollars, uh, period. Like that, that's what LPs actually want. It's amazing to get a hundred. Like I've, I've had two funds that I've invested in. One is Mickey. This other one is this fund called AngelPad. Which was, uh, kind of like a third rate competitor. I don't want to call it third rate, but it was like, it was not, you know, there was Y Combinator and then it's like, there was AngelPad. It was just like this, this small little experiment. That was a 120 X. I got a 120 times the capital that I get DPI.

AI assessment note: “the harder thing to do is to just return gross dollars”