The Exchanges

Q What do you, what do you mean by that? If you're in?

A I mean, if you are a venture fund that is staring down the barrel of having to deploy three billion dollars, I think that is hard because again, at the early stage, it is hard to capture disproportionate ownership in the few companies that actually generate all of that liquidity. If you're a small, if a small venture fund, I think it's super possible in today's world. You don't actually have to be in, you don't have to catch the seed of SpaceX. You'd really like to because those are the only, the platform companies generate liquidity. But at the end of the day, like, you can get by with not capturing all of the great outcomes. If you have a three billion dollar venture fund, the math is really hard, right? You have to capture a lot of those. The growth funds are, growth funds are a little bit different math. But I think, just go back to your question about, you know, can these, can a five billion dollar growth fund scale and work? The answer is yes. The reason why is the markets change in two different ways. Change number one, these companies are staying private longer. They're staying, they're getting bigger while they're private. There are more opportunities to invest over time. So now we're 10 years ago, you couldn't put a billion dollars in a company. Now you can invest a billion dollars in any given round. If I invest a billion dollars and I 10 X that billion dollars, that…

AI assessment note: “can a five billion dollar growth fund scale and work? The answer is yes.”

Q What was your biggest lesson from working with Mamoon?

A Again, so many. I think the thing, the gift that Mamoon hasn't, I think from the SaaS era, Mamoon, my view is he was the best Series A investor in the SaaS era, period. Right? I mean, if you look at his track record, it's incredible. Figma, Glean, Rippling, Slack. I mean, it's just this unbelievable, like, hit after hit after hit. What Mamoon is special at, and what he pays attention to, and what I learned from him is, there are distinct inflection points in companies. There are moments where they really kink, right? They kink up. And he is the master at seeing that around the Series A, right? With very little data, being able to see it. I remember going back to, um, I worked on Figma with him when I was an associate at Kleiner. And I cut all the data for Mamoon. This is a, it was, it was a very fun time. And I remember he took one look at it, and within 30 seconds he was like, we're doing it. And there was this big company in Vision at the time, and it was a great company, and everybody thought it was the winner. And he looked at that data, and he was like, this is, this is going to happen. And what he saw is the net retention curves, the customer behavior of really big companies. I think, I can't remember exactly, but I think the companies were Google and Square and Amazon, right? Like, really insane customers. And this is when Figma had . And he saw the usage curves inside t…

AI assessment note: “what I learned from him is, there are distinct inflection points in companies”

Q When founders come to you and they're like, um, how dare you? Like, it started off as a pillow company. And now it's doing enterprise payments. How am I to know?

A And listen, that's part of the game. And, um, as a founder, I complete, I completely empathize and understand, understand that, right? I can understand how that would be, you know, a really tricky situation. From our perspective, it's when you're investing in large markets, oftentimes you are going to end up in assets that compete because they naturally expand terms. A great example of this is I think we were the only private investor that was Invested in Snowflake and Databricks when they were both private. They started off in completely different areas, right? Databricks didn't have a data warehousing product, and Snowflake didn't really do a lot of ELT. It was mostly built around the ecosystem. They grew together, and they, we weren't invested when they were, you know, starting to compete because Snowflake went public a lot earlier. But at the same time, that happens in big markets.

AI assessment note: “when you're investing in large markets, oftentimes you are going to end up in assets that compete”

Q Is that three X enough to be exciting?

A No. A three X is not enough to be exciting. And the math is really simple, right? Say, say I'm a fund and I'm KOTU. And I want to make a three X net return for my investors, which I think is, which I think is sort of the baseline for what people would say is like a top quartile return and people get really excited about three X net return for a fund, you know, 25% net IRR, something around those bands. I'm going to have some things where I swing and I miss. And say I have a one X, I need a five X on the other side of that. Heaven forbid, I have a loss rate. I have a loss and I have a zero. I need a six. We obviously really try to avoid those, right? If I have a two, I need a four. So for me, I need to see a, a steady case where you can get that three X, but I really need to believe that if the company three X's, I want to put more money in because it can three X again. And I think this is a really critical thing that a lot of folks end up missing over time is ultimately I need to imagine a case where after I've made my three X, somebody else thinks they can make their three X because otherwise one, I'm not going to get those six X pluses that I'm going to need in my fund. But two, the company's not going to exit. I have to imagine this is why the big idea have being in big ideas really matters. Somebody's got to sit on the other side of that stock. I have to be able to walk dow…

AI assessment note: “No. A three X is not enough to be exciting.”

Q What do you, what do you mean by that? If you're in?

A I mean, if you are a venture fund that is staring down the barrel of having to deploy three billion dollars, I think that is hard because again, at the early stage, it is hard to capture disproportionate ownership in the few companies that actually generate all of that liquidity. If you're a small, if a small venture fund, I think it's super possible in today's world. You don't actually have to be in, you don't have to catch the seed of SpaceX. You'd really like to because those are the only, the platform companies generate liquidity. But at the end of the day, like, you can get by with not capturing all of the great outcomes. If you have a three billion dollar venture fund, the math is really hard, right? You have to capture a lot of those. The growth funds are, growth funds are a little bit different math. But I think, just go back to your question about, you know, can these, can a five billion dollar growth fund scale and work? The answer is yes. The reason why is the markets change in two different ways. Change number one, these companies are staying private longer. They're staying, they're getting bigger while they're private. There are more opportunities to invest over time. So now we're 10 years ago, you couldn't put a billion dollars in a company. Now you can invest a billion dollars in any given round. If I invest a billion dollars and I 10 X that billion dollars, that…

AI assessment note: “I mean, if you are a venture fund that is staring down the barrel”

Q Pretty much. I put my feet on the ground. I say margin matters. Um, but I start to question whether margin actually does matter in the early days. If your company is rocking, you're spending on inference, and that is a sign of good usage and love. Does margin matter?

A Yeah. I think the same business principles that have applied to businesses for the last three decades in technology are the same business principles that matter today. Margin matters, but that is new, but it's nuanced. Margin, I would, I would add an addendum to that. Margin matters at scale. The best businesses in particular infrastructure, whenever there's a technology wave happening and an architecture shift, Some of the best businesses, not all of them, but some of the best businesses have had horrific margins early. The hyperscalers. The hyperscalers were low margin early. Those are the best software platform businesses in the world, right? Snowflake and Databricks. Very low margin early. A lot of people passed on those early rounds because they're, ah, in SaaS, you have to have 80% gross margin. Look at Snowflake. It's got 20, you know, like margin matters, but early it can be a misleading indicator, right? Especially when an architecture shift is happening. The reason why in AI, and I'll give you the the bull case on this, right, is the reason why margin might not matter early on in a company's life in AI is the cost curve is coming down so fast. Say my inference margin is 10% today. It may have been negative a quarter ago and super negative two quarters ago, but the token costs are coming down so fast. Maybe I'll be, if I'm an application AI company, I'll probably be ab…

AI assessment note: “Margin matters, but early it can be a misleading indicator”

Q previous software revenues were. And now we have, you know, this transience of technology superiority, which sounds really kind of wanky, but like technology cycles have changed so far. So Gemini is better, and then Claude's better, and then OpenAI is better. That your durability of revenue seems to be more questionable and transient than ever. Should we ascribe value to revenue in the same way that we used to?

A Yeah, I think, I think you're absolutely right. I think it's changing, and I think it changes during every architecture shift, right? This is the really critical part of technology, right? As you moved from on-premise technology to SaaS technology, as you moved from the internet to mobile internet, right? You had a potential for all of the companies in the prior generation to completely evaporate. And the big question is, can you find the companies that have the talent density, that are the most forward-thinking, That are willing to reinvent themselves over and over and over again, which is so hard. And those are the companies that you want to find out. I think of a great example every time I think of this point, which is Databricks. Right. If you talk to Ali from Databricks, um, we've been an investor since 2019. I think one of the things that we've seen from there and even before, I mean, I looked at the company when I was at Kleiner Perkins, um, I've seen this company for a very long time is his ability to reinvent that company over and over and over again and ride multiple S-curves from being, you know, basically an ELT data transformation layer to, you know, running and training models to being the center of all data in The enterprise. Those are like multiple S curves that he's hopped and multiple times he's reinvented the company. And I think it's not revenue growth that …

AI assessment note: “I think it's not revenue growth that you want to chase. It's that.”

Q is like when we did Lovables A, it was like at three million in revenue. By the time the legals were done, it was at 20. And so the multiple had gone from 70 x to 10 x. Correct. I mean, Anton should have been asking for a trade Like, I want to renegotiate this. How do we value assets that are growing in such disproportionate or previously unseen ways?

A Yeah. Again, I think it's, this is one of the, this is one of the hardest things. It's why we actually think the, the framework that we use internally is we think about valuation. Everybody has to think about valuation, but when a company is growing exponentially, 10 X year and year, 50 X year and year, right? The things that we're seeing now, We think about valuation last. It's the last question we try to answer is, is the valuation great? Because, like you mentioned, you may invest in a Series C at twenty million of ARR at three billion post, and that seems insane. But if the 20 goes to a 200 in one year, and then 600 the next, and three billion the next, all of a sudden that looks extremely cheap. And so our job is, how do you find the things that are on that curve?

AI assessment note: “when a company is growing exponentially... We think about valuation last.”

Q to three X the next year to seven hundred and fifty million. Wow. Well, you paid four and a half. So even if it double or triples and then doubles again, you're still not at the six or seven X that it will be valued at in a public market. How do you just get your head around the hard dynamics of what it will be in a public market?

A Yeah. No, I think this is, I think it's a great question and it, It filters down into every decision that we think about all the time, right? And I think the key is, first, you want to be in gigantic TAMs. Big ideas only, right? Because if you ever compromise on that very first principle, and you're paying high valuations, you're in trouble. Medium TAM, small TAM, you better believe that this thing can be absolutely gigantic. We have this test internally, right? Where it used to be, five years ago, we called it the ten billion dollar public company test. Can this be a ten billion dollar plus public company? That bar has changed, right, in this, in this new world because we are tackling much larger markets than we used to. And so now that test is, can you be just an enduring public company? And that may mean fifty billion of market cap. It may mean a hundred billion of market cap. It really depends based on the stage, but really it's big idea first. And then is the market absolutely yanking you into that giant market? Right. Do you feel that market pull such that that revenue curve and subsequently down the line that earnings path is really achievable? And so what you need to really believe is like, take this fifty million ARR five billion post type company, right? You need to believe that someday you can get to five billion of revenue with 30% margin minimum growing really fast…

AI assessment note: “You need to believe that someday you can get to five billion of revenue”

Q When founders come to you and they're like, um, how dare you? Like, it started off as a pillow company. And now it's doing enterprise payments. How am I to know?

A And listen, that's part of the game. And, um, as a founder, I complete, I completely empathize and understand, understand that, right? I can understand how that would be, you know, a really tricky situation. From our perspective, it's when you're investing in large markets, oftentimes you are going to end up in assets that compete because they naturally expand terms. A great example of this is I think we were the only private investor that was Invested in Snowflake and Databricks when they were both private. They started off in completely different areas, right? Databricks didn't have a data warehousing product, and Snowflake didn't really do a lot of ELT. It was mostly built around the ecosystem. They grew together, and they, we weren't invested when they were, you know, starting to compete because Snowflake went public a lot earlier. But at the same time, that happens in big markets.

AI assessment note: “when you're investing in large markets, oftentimes you are going to end up in assets that compete”

Q Is that three X enough to be exciting?

A No. A three X is not enough to be exciting. And the math is really simple, right? Say, say I'm a fund and I'm KOTU. And I want to make a three X net return for my investors, which I think is, which I think is sort of the baseline for what people would say is like a top quartile return and people get really excited about three X net return for a fund, you know, 25% net IRR, something around those bands. I'm going to have some things where I swing and I miss. And say I have a one X, I need a five X on the other side of that. Heaven forbid, I have a loss rate. I have a loss and I have a zero. I need a six. We obviously really try to avoid those, right? If I have a two, I need a four. So for me, I need to see a, a steady case where you can get that three X, but I really need to believe that if the company three X's, I want to put more money in because it can three X again. And I think this is a really critical thing that a lot of folks end up missing over time is ultimately I need to imagine a case where after I've made my three X, somebody else thinks they can make their three X because otherwise one, I'm not going to get those six X pluses that I'm going to need in my fund. But two, the company's not going to exit. I have to imagine this is why the big idea have being in big ideas really matters. Somebody's got to sit on the other side of that stock. I have to be able to walk dow…

AI assessment note: “No. A three X is not enough to be exciting. And the math is really simple”

Q I adore Cliff and Mel, and I think they're amazing, but you said about, like, the platform companies, and you included Canva. If you were to be a harsh critic, and you'd say, well, Figma's worth eleven billion dollars today, and image generation, graphic generation is right in the pathway of a lot of large AI companies, Is Canva really a platform company?

A What I love about Canva is they've shown that same ability that Databricks has, where they're able to hop multiple S-curves and develop multiple products, right? They started as, I'm sure you know the story, but it's incredible. Mel started, Mel and Cliff started this business as a yearbook business, making yearbooks. They successfully transitioned that online. They successfully transitioned that to SaaS. And now they've transitioned to many, many, many products, right? Canva is a suite of like a dozen products that are all growing extraordinarily quickly. So you have that dynamic. And then the other thing that I love is they were one of the first companies that really leaned into AI. I remember Cliff called me about this very early on because we were early investors in, uh, in stable diffusion. If you remember the image, the image generation company in open AI and a few of these other businesses. And he called us Really early in this wave, like pre ChatGPT in this wave and was like, hey, we're going to start integrating AI into our business now. And so that type of mentality, both the ability to develop multiple products and hop TAMs, and to stay ahead of the curve in AI, I think is going to serve them very well.

AI assessment note: “they've shown that same ability that Databricks has, where they're able to hop multiple S-curves”

Q What was your biggest lesson from working with Mamoon?

A Again, so many. I think the thing, the gift that Mamoon hasn't, I think from the SaaS era, Mamoon, my view is he was the best Series A investor in the SaaS era, period. Right? I mean, if you look at his track record, it's incredible. Figma, Glean, Rippling, Slack. I mean, it's just this unbelievable, like, hit after hit after hit. What Mamoon is special at, and what he pays attention to, and what I learned from him is, there are distinct inflection points in companies. There are moments where they really kink, right? They kink up. And he is the master at seeing that around the Series A, right? With very little data, being able to see it. I remember going back to, um, I worked on Figma with him when I was an associate at Kleiner. And I cut all the data for Mamoon. This is a, it was, it was a very fun time. And I remember he took one look at it, and within 30 seconds he was like, we're doing it. And there was this big company in Vision at the time, and it was a great company, and everybody thought it was the winner. And he looked at that data, and he was like, this is, this is going to happen. And what he saw is the net retention curves, the customer behavior of really big companies. I think, I can't remember exactly, but I think the companies were Google and Square and Amazon, right? Like, really insane customers. And this is when Figma had . And he saw the usage curves inside t…

AI assessment note: “what I learned from him is, there are distinct inflection points in companies.”

Q But they can destroy the economics. Is seed still a business when you have mega fund entry with different economics in the way that we do?

A I mean, I think it's gotten harder for two reasons. One is you do have this mega fund dynamic, but the other thing is we're in a different world than we were five years ago, right? Which is, In general, people are coming out of the gate with bigger check sizes and bigger valuations, right? And that just raises the risk like dramatically over time, right? So I think that's, those are the two dynamics that are, that are really at play is it's harder for a seed fund to buy 20% today or 10% today or five percent today than it was a few years ago because of this dynamic. And that has to do with a lot of different things. One of them is in a SaaS world, You didn't need that much capital. You started up, you kind of get going, whatever. In this world, businesses tend to be more capital intensive, right? They may be actually more durable at scale because of this. Makes it harder for the next entrant to come in. But the reality is they're harder to start. They take more capital. And that has led to some of these, like, very big ballooning seed rounds. I think that makes it harder to be a seed investor in today's world. And again, why? Having a flexible mandate where you can row up and down that river and not have to be there is really a nice place to be.

AI assessment note: “I think it's gotten harder for two reasons. One is you do have this mega fund dynamic”

Q How do we determine value in this pool of reductive market caps?

A I think it's really, really hard right now is the short answer, right? This is the debate that we have all the time inside of our building, right? Which is, you know, you take a design tool, for example, you can make an argument that that design tool is super well positioned in a world of AI because they're going to integrate AI into all the design process and generate so much more value than before. But then you could say, well, I just create all my designs in ChatGPT now, right? So why would I even need this design tool? And I think that's the argument that you're going to have on both sides of this at all times. I think the things that you're going to want to look for, the leading indicators that you're going to want to look for are, is the revenue still continuing to grow sequentially? Is net new AR still continuing to climb? What's happening with the retention dynamics of these businesses? And I think, like, the more you can see that, the more, um, the better you're going to feel. But the reality is, for the next three months, six months, nine months, we're not really going to know what's really happening in the world, right? Because things are happening so fast, and all of the earnings that happen are retroactive. Right? So you can only see into the past that way. So I think that's why you're seeing people basically walk away from the sector.

AI assessment note: “the leading indicators that you're going to want to look for are, is the revenue”

Q is like when we did Lovables A, it was like at three million in revenue. By the time the legals were done, it was at 20. And so the multiple had gone from 70 x to 10 x. Correct. I mean, Anton should have been asking for a trade Like, I want to renegotiate this. How do we value assets that are growing in such disproportionate or previously unseen ways?

A Yeah. Again, I think it's, this is one of the, this is one of the hardest things. It's why we actually think the, the framework that we use internally is we think about valuation. Everybody has to think about valuation, but when a company is growing exponentially, 10 X year and year, 50 X year and year, right? The things that we're seeing now, We think about valuation last. It's the last question we try to answer is, is the valuation great? Because, like you mentioned, you may invest in a Series C at twenty million of ARR at three billion post, and that seems insane. But if the 20 goes to a 200 in one year, and then 600 the next, and three billion the next, all of a sudden that looks extremely cheap. And so our job is, how do you find the things that are on that curve?

AI assessment note: “We think about valuation last. It's the last question we try to answer”

Q Because of the expansion of TAMs and outcome sizes, can you be thoroughly elastic on entry price even at the real growth stage? Or does Price elasticity constraints significantly with increasing enterprise value.

A Ultimately, price always does matter, right? I think some folks will say, ah, price doesn't matter. I think price does matter, but I think it matters least. You, of course, could make the argument of, oh, Lucas, well, you do it five. Why not six or seven or eight or nine or ten billion? What if it was twenty billion? What if it was thirty billion? There does come a delineation point where you feel like the returns are going to erode such that you would pass on an opportunity. But I'd say, by and large, if you're the one instigating these rounds and you're the one that's preempting these rounds, you can kind of help figure out what the right price is for a company at any given moment. And I do think you want to think about it last because, again, These generational companies, it's almost never too late for them, right?

AI assessment note: “There does come a delineation point where you feel like the returns are going to erode”

Q Maybe. No. That's how we think about it. Do you have any internal monikers or frameworks for like?

A I think the more simplistic way that we think about it, and again, this is not a hard and fast rule and it's It's more qualitative than anything is if I invest in this round of this price and the company executes, do I want to put more at a higher price? That's the litmus test. It's to say, all right, say I invest in a company at five billion and it does super well this year. Is this a big enough idea? Is this generational enough? Is this transformational enough? Is the founder amazing enough that if in six months they wake up and say they want to raise it 10, that I'm going to want to do that.

AI assessment note: “if I invest in this round of this price and the company executes, do I want to put more at a higher price?”

Q What have you changed your mind on in the last 12 months?

A The size of outcomes. This is really simple. Like, 12 months ago, I wasn't as convinced that we were really going to be able to address labor, um, and that this, like, token machine concept that, you know, human inputs were going to become machine inputs. I wasn't all the way there. We were still kind of in an assistant world versus an agent world. I've become fully convinced on this. A lot of it is due to using a lot of the tools like Cloud Code myself and just really feeling this. But that, my opinion, has really changed on that in the last year. I think the outcomes this generation in technology are going to be so much bigger than the outcomes from the last generation.

AI assessment note: “12 months ago, I wasn't as convinced that we were really going to be able”

Q How do we determine value in this pool of reductive market caps?

A I think it's really, really hard right now is the short answer, right? This is the debate that we have all the time inside of our building, right? Which is, you know, you take a design tool, for example, you can make an argument that that design tool is super well positioned in a world of AI because they're going to integrate AI into all the design process and generate so much more value than before. But then you could say, well, I just create all my designs in ChatGPT now, right? So why would I even need this design tool? And I think that's the argument that you're going to have on both sides of this at all times. I think the things that you're going to want to look for, the leading indicators that you're going to want to look for are, is the revenue still continuing to grow sequentially? Is net new AR still continuing to climb? What's happening with the retention dynamics of these businesses? And I think, like, the more you can see that, the more, um, the better you're going to feel. But the reality is, for the next three months, six months, nine months, we're not really going to know what's really happening in the world, right? Because things are happening so fast, and all of the earnings that happen are retroactive. Right? So you can only see into the past that way. So I think that's why you're seeing people basically walk away from the sector.

AI assessment note: “leading indicators that you're going to want to look for are, is the revenue”

Q then are saying that risk adjusted, that is the best place you believe to put your capital, which is where I get stuck. I'm like in an ecosystem where there's so much opportunity. I understand that you can get there, but is that really the best place to put my money over the 10 other homes where I don't have to double, triple, and then do a somersault into Kenya?

A No, it's really fair. And again, it's, it's something we think about a lot. I think the The two things you want to consider when it comes to that, right? And it's one of the reasons why, again, we love having a flexible mandate. We are not tied to just being able to do A Series B at 300 post and like that's all we can do is because we can have almost this like rowboat that rose up and down the river. And anytime we see something opportunistically that we think is the best risk adjusted opportunity at that moment, we can invest. And then the second thing we're really looking for, right? Take that round as an example. And one of the reasons why we go back to this big idea test is I want to believe that if that company works, that its best days are ahead of it and I can continue to invest. One thing that, um, Jeff Horing from, from Insight always says is the, you know, the, the best round is the double down round. And so by getting access to that company at a certain stage, if I think it has a shot at being a hundred billion dollar company, that round may not actually be the best round, but it gives me the opportunity to double down and make an even larger investment where more of my capital is going to be deployed over long periods of time. And again, it's important why this market structure is changing, right? I now, because companies Are staying private longer. These platform c…

AI assessment note: “that round may not actually be the best round, but it gives me the opportunity to double down”

Q Maybe. No. That's how we think about it. Do you have any internal monikers or frameworks for like?

A I think the more simplistic way that we think about it, and again, this is not a hard and fast rule and it's It's more qualitative than anything is if I invest in this round of this price and the company executes, do I want to put more at a higher price? That's the litmus test. It's to say, all right, say I invest in a company at five billion and it does super well this year. Is this a big enough idea? Is this generational enough? Is this transformational enough? Is the founder amazing enough that if in six months they wake up and say they want to raise it 10, that I'm going to want to do that.

AI assessment note: “if I invest in this round of this price... do I want to put more”

Q to invest as much money as possible as the company becomes more expensive, which is one of those kind of counterintuitive statements. Do you want to kind of spray earlier, and spray is a derogatory term, and I didn't mean that rudely, but like constrain capital effectively, and then double down very aggressively, or do you want to aggressively get ownership and then focus on constraining as time goes on?

A Yeah, we're, we're much more the latter. And I think there are two, there are two dynamics around this. One is, our view is there are very few companies that generate the disproportionate value in technology, right? If you look at the private markets today, take the whole private market ecosystem, 20 companies have generated 80% of the enterprise value. 20 companies, 80% of the enterprise value. Of all the private companies that exist in the world. And four companies, right, have generated 65% of the enterprise value. Four companies. And so, what really matters is being in those companies. Those 20 platform companies that are generating the disproportionate amount of value. And then your next question is, all right, well, how? Like, this wonderful framework. Like, We all would love to be in all of these platform companies, but like how? And the answer is, from our view, you can't do this prey and prey at the early stage or the early growth stage. The reason why is you may be in the wrong horse, or you may be in the wrong market, and you may be investing your time wrong. Because there are very few companies, we need to make very few investments, even at the, even at the early growth stage or the growth stage. We can't afford to be in the wrong horse.

AI assessment note: “we're much more the latter. And I think there are two dynamics”

Q But they can destroy the economics. Is seed still a business when you have mega fund entry with different economics in the way that we do?

A I mean, I think it's gotten harder for two reasons. One is you do have this mega fund dynamic, but the other thing is we're in a different world than we were five years ago, right? Which is, In general, people are coming out of the gate with bigger check sizes and bigger valuations, right? And that just raises the risk like dramatically over time, right? So I think that's, those are the two dynamics that are, that are really at play is it's harder for a seed fund to buy 20% today or 10% today or five percent today than it was a few years ago because of this dynamic. And that has to do with a lot of different things. One of them is in a SaaS world, You didn't need that much capital. You started up, you kind of get going, whatever. In this world, businesses tend to be more capital intensive, right? They may be actually more durable at scale because of this. Makes it harder for the next entrant to come in. But the reality is they're harder to start. They take more capital. And that has led to some of these, like, very big ballooning seed rounds. I think that makes it harder to be a seed investor in today's world. And again, why? Having a flexible mandate where you can row up and down that river and not have to be there is really a nice place to be.

AI assessment note: “I mean, I think it's gotten harder for two reasons.”

Q Do you think a good investor at A can be a good investor at D? A lot of LP mindsets are like, no, early stage is different to growth, and that's very different. I think Josh, who's a dear friend at Thrive, has proved that actually that's not the case, but other people still very much hold that true.

A I don't think it's impossible, but I do think it is very hard. And I think that's because the type of frameworks that you use, the types of things that you see are very different at different scales. Being able to read a balance sheet actually does matter for a pre IPO company, right? Like that really matters. But seeing thousands of founders, thousands and thousands of thousands really matters for seed because what else do you have, you know, to go off of? And so I do think it really matters. I don't think it's impossible. I think there are some funds that have done it exceptionally well. But I think it's why you see, for us, right, we, um, we as a fund, we, I actually think the public market skill set and the private market skill set is also different, and so having different folks that are focused on different things is really important because there are different parameters, different things that you see all day, and there are other people that you're competing with in all of those different segments that make it really tough to be the best at everything.

AI assessment note: “I don't think it's impossible, but I do think it is very hard.”

Q What have you changed your mind on in the last 12 months?

A The size of outcomes. This is really simple. Like, 12 months ago, I wasn't as convinced that we were really going to be able to address labor, um, and that this, like, token machine concept that, you know, human inputs were going to become machine inputs. I wasn't all the way there. We were still kind of in an assistant world versus an agent world. I've become fully convinced on this. A lot of it is due to using a lot of the tools like Cloud Code myself and just really feeling this. But that, my opinion, has really changed on that in the last year. I think the outcomes this generation in technology are going to be so much bigger than the outcomes from the last generation.

AI assessment note: “The size of outcomes. This is really simple. Like, 12 months ago, I wasn't”

Q previous software revenues were. And now we have, you know, this transience of technology superiority, which sounds really kind of wanky, but like technology cycles have changed so far. So Gemini is better, and then Claude's better, and then OpenAI is better. That your durability of revenue seems to be more questionable and transient than ever. Should we ascribe value to revenue in the same way that we used to?

A Yeah, I think, I think you're absolutely right. I think it's changing, and I think it changes during every architecture shift, right? This is the really critical part of technology, right? As you moved from on-premise technology to SaaS technology, as you moved from the internet to mobile internet, right? You had a potential for all of the companies in the prior generation to completely evaporate. And the big question is, can you find the companies that have the talent density, that are the most forward-thinking, That are willing to reinvent themselves over and over and over again, which is so hard. And those are the companies that you want to find out. I think of a great example every time I think of this point, which is Databricks. Right. If you talk to Ali from Databricks, um, we've been an investor since 2019. I think one of the things that we've seen from there and even before, I mean, I looked at the company when I was at Kleiner Perkins, um, I've seen this company for a very long time is his ability to reinvent that company over and over and over again and ride multiple S-curves from being, you know, basically an ELT data transformation layer to, you know, running and training models to being the center of all data in The enterprise. Those are like multiple S curves that he's hopped and multiple times he's reinvented the company. And I think it's not revenue growth that …

AI assessment note: “I think it's not revenue growth that you want to chase. It's that.”

Q What was your biggest lesson from working with Mary?

A I mean, so many lessons. I think the biggest lesson is she has this, and it comes from her background of being at Morgan Stanley for a really long time. She has this incredible analytical bent of being able to see things and see stories and numbers that other folks don't and being willing to lean against the grain whenever she feels things. Um, and she's able to tell and tell these incredible stories with data and understand what's happening in the world based on data. I'll give you one example is I remember my second week at Kleiner, um, I didn't know how to model. I came from, I came from inside. I could barely, I could barely model. I was great at talking to founders, but could barely model. And I found myself You know, in the middle of modeling exercise with Mary and just getting absolutely destroyed. And one of the things she taught me is like, that is actually really important. Being able to express a company in a few, a complex company in a few lines in Excel and tell stories with data is like an incredible skill. And she has this knack of being able to look at like cell in 95 and know there's an error. And so that's what I learned is to be, like, highly analytical, very detail-oriented, and to tell the story with the data.

AI assessment note: “that's what I learned is to be, like, highly analytical, very detail-oriented”

Q Pretty much. I put my feet on the ground. I say margin matters. Um, but I start to question whether margin actually does matter in the early days. If your company is rocking, you're spending on inference, and that is a sign of good usage and love. Does margin matter?

A Yeah. I think the same business principles that have applied to businesses for the last three decades in technology are the same business principles that matter today. Margin matters, but that is new, but it's nuanced. Margin, I would, I would add an addendum to that. Margin matters at scale. The best businesses in particular infrastructure, whenever there's a technology wave happening and an architecture shift, Some of the best businesses, not all of them, but some of the best businesses have had horrific margins early. The hyperscalers. The hyperscalers were low margin early. Those are the best software platform businesses in the world, right? Snowflake and Databricks. Very low margin early. A lot of people passed on those early rounds because they're, ah, in SaaS, you have to have 80% gross margin. Look at Snowflake. It's got 20, you know, like margin matters, but early it can be a misleading indicator, right? Especially when an architecture shift is happening. The reason why in AI, and I'll give you the the bull case on this, right, is the reason why margin might not matter early on in a company's life in AI is the cost curve is coming down so fast. Say my inference margin is 10% today. It may have been negative a quarter ago and super negative two quarters ago, but the token costs are coming down so fast. Maybe I'll be, if I'm an application AI company, I'll probably be ab…

AI assessment note: “Margin matters at scale... early it can be a misleading indicator”

Q But Well, respectfully, they're not great companies. They're good companies. And in a prior cycle, they would have been funded, and they've been funded well. But now, are you really going to jump out of bed for a ten million growing to twenty five million?

A The short answer is, I don't know. I don't know what's going to happen to those companies. I don't know what the terminal value is. I don't know what the exit pathways are, right, with private equity in the space, in the place that it's in with the public markets where it is. All I know is I have a lot of conviction and I see a path in the style of investing that we do. I don't, I don't know how to comment on the other part of the markets, right? There's this notion that like the, the triple, triple, triple, double, double, double is dead and these companies suck and all this stuff. I don't think that's true. There are great companies. You can drive real margin from them. They make incredible businesses. It's just not our strategy, right? And I think in today's world, The reality is in a SaaS world, the triple, triple, double, double, double was a thing. It was an incredible metric. These businesses were incredibly repeatable, very comparable. Now we exist in a world where if you have a product that the market likes, it is going to absolutely yank you into that market, right? It's not going to triple at the earliest stages. It is going to scream, right? And I think you re you really see that, right? And so those are the companies. And again, It's not like we think these companies are all bad and this and that. It's just. Our strategy is to find those and to work with those comp…

AI assessment note: “They make incredible businesses. It's just not our strategy”