The Exchanges

Q This is off schedule. This is like for true pros, but this has been so good. Okay. The hardest things in life or the heaviest things in life are not iron or gold, but unmade decisions. What unmade decision do you have that weighs on you most?

A I don't know. I can think of several little things that I would have done, but I don't know. We had early acquisition interest in Rockmelt. And I think that, that, that was probably something that I should have like leaned into more in retrospect. So like, that's, that's an example. And that would have been a very different, like, But in the scheme of things, like, looking back now, it's twenty-twenty-four. Would it have changed anything? Probably not. It would have been fine. Like, I, it would have sold the company two years earlier, or whatever, and maybe would have made more. And, you know, this is the whole, um, what's that children's story about the, the horse rider, and the soldier, and the conscription? Have you heard this story? It's a really amazing story. So, like, the, the summary of it is, is like, there's a, it's like ancient China, and there's a, there's a draft. And so, um, and the first part of it is, there's a family in rural China, and they get a horse. And, uh, This kid finds a horse and the villagers are like, oh my God, he's so lucky. It's so lucky. It's so lucky. And the wise man is like, we'll see. And the kid's riding the horse and breaks his leg. And then the villagers are like, oh my God, it's so unlucky. It's so unlucky. It's so unlucky. And the wise man is like, we'll see. And then there's a draft, a military draft, and the draft people come to this …

AI assessment note: “We had early acquisition interest in Rockmelt. And I think that, that, that was probably”

Q Does that mean there's still great value in open AI?

A I think that's a good question that, that is, um, really interesting because if you think about open AI and anthropic and, and meta and Google, Um, and you know, and then there's a whole bunch of others coming, XAI and Mistral and, um, and so forth, SSI now. Um, you know, I think the foundational model war, like, benefits us all in a way. Like, it's just, it's really, really good for, for consumers and people around it, because it's just, like, they're pushing the state of the art so much. Um, in terms of value accrual, like, you know, I think for Benchmark, like, we have a, we, we have no foundational model investments, uh, thus far anyway. Um, and so, you know, that could be very bad or very good. I don't know. Um, but we don't. And then two, we have a set of, like, infrastructure investments, which I think are, are really interesting. So, um, we have Cerebris, which is a semiconductor and systems company, um, for AI that we invested in and led their, or co-led with Foundation their Series A in 2016. So we've been working on it for eight years, um, as an example. Um, we have, uh, companies like Fireworks, which are an inference service and others kind of of that infrastructure software layer. And I think those like, you know, again, they're growing very, very quickly, like astoundingly quickly, um, and doing really cool things. But at the same time, you kind of have to ask li…

AI assessment note: “we have no foundational model investments... we have a set of, like, infrastructure investments”

Q How do you dig into that as an investor today, evaluating whether a founder has thought that through comprehensively enough for you to get comfortable on that?

A I'm not looking for an answer or a right answer. What I'm really looking for And trying to figure out is, has the person thought about it deeply and is constantly learning or constantly applying and adding new mental models to their framework, um, to figure out what the right path is and they're navigating it. It isn't like, hey, I'm a boat captain and I'm looking and like, this is where we're going to go. You start going and then conditions change and you get more information and you have to kind of constantly change. And And so what, what you're actually trying to evaluate isn't, they're not going to have all the answers and that's okay. You can have theories and hypotheses and then evaluate and change as, as time goes on and you get more information. And I think that's really what you're looking for versus, oh, this is like, this is how it's going to work. And that, that actually, that can be bad in its own way.

AI assessment note: “What I'm really looking for And trying to figure out is, has the person thought about it deeply”

Q terms of, uh, sector. There are very different companies there from Confluent, Contentful to Cerubris. And I spoke to Bruce Dunleavy before the show, and he said that bluntly your breadth of sector mastering is completely unparalleled as an investor. And he asked a great question, I thought, which was, what is your learning process for entirely new categories? And how do you break it down and learn so fast?

A I am not a sector specialist and nobody at Benchmark is. And I think the, the fundamental idea with Benchmark is there's a small group of people, a small group of partners who are all equal. And right now it's five of us, but sometimes it's four, sometimes it's six, but it's basically four to six. Who cover technology. And the challenge with that, if you kind of think about it, is we can't be sector specialists because the, the sectors that have the most disruption and things are changing the fastest, like are changing. That part's constantly changing. So you have to be as, as a group of investors, you have to be, you have to be moving. You have to be looking at new stuff because that's where the disruption is happening. And so So then the question is, so like, I can't be a sector specialist. I can't be a, a semiconductor specialist, or I can't be, um, an open source specialist. Of course, we each have preferences and things that we like and don't and lessons that we, you can apply, but like it, but it, but we're, but it's not a specialty model. And I think about this a lot, and we should talk about it in the context of AI, but you can't have that. So then the question is, well, what can you evaluate on? And I think this is, this is it, this is it for me, which is, you can say, hey, This is an extraordinary entrepreneur. That's a, that's a, that's an evaluation that, that you c…

AI assessment note: “we can't be sector specialists... So then the question is, well, what can you evaluate on?”

Q How do you dig into that as an investor today, evaluating whether a founder has thought that through comprehensively enough for you to get comfortable on that?

A I'm not looking for an answer or a right answer. What I'm really looking for And trying to figure out is, has the person thought about it deeply and is constantly learning or constantly applying and adding new mental models to their framework, um, to figure out what the right path is and they're navigating it. It isn't like, hey, I'm a boat captain and I'm looking and like, this is where we're going to go. You start going and then conditions change and you get more information and you have to kind of constantly change. And And so what, what you're actually trying to evaluate isn't, they're not going to have all the answers and that's okay. You can have theories and hypotheses and then evaluate and change as, as time goes on and you get more information. And I think that's really what you're looking for versus, oh, this is like, this is how it's going to work. And that, that actually, that can be bad in its own way.

AI assessment note: “What I'm really looking for And trying to figure out is, has the person thought about it deeply”

Q But even now, actually, you could argue that say like in an Uber lift case, which is exactly what I was thinking, the insight I think would be that broad is better than narrow in terms of market expansion and being everywhere is more important than honing one. There's always insights. Do you see what I mean?

A Yeah. I mean, I think, I think you could say like that, you know, and that's a good case of violent, violent execution by a determined team. I think there's also a, and so I think there are insights there. As you said, there's like little execution detailed insights in, in your point. And, and so it's not like it doesn't have to be some pie in the sky insight. Like that's like, oh my God, you know, like AGI is coming in Q three of 20, 25. Like that's not the kind of insight that we're talking about. We're talking about Like actual things that are nuances that help like that build a company. And so, yeah, they think they come at different levels. I would also say there, and you know, this is a difference and I've been fortunate to work with Pat, um, on a board and, um, and, and work with the, the larger squad team on a number of companies. And. You know, there is a difference also in stage, right? They're growth investors, like at least Pat and his team are growth investors and we're early stage. And so like, there's also like a, a stage difference in terms of how we evaluate

AI assessment note: “I think there are insights there. As you said, there's like little execution detailed insights”

Q What about then the biggest surprises about the move into VC?

A I don't know about surprises, but I would say I think from the outside, it seems Easy. I think there's a perception that you like sit back and all of these people come to you and pitch you and then you're just kind of like picking and it's like, it's a laid back and easy job. And I think what's really clear to me, and I knew this going in, but is emphasized even two years in is to do this job at the highest levels. It's actually extremely challenging. There's a ton of hustle required. You have to get out there. The best stuff is very competitive and So I think there's a perception that it seems easy, and it may be that it's easy to skate by, but I think to do it at the highest levels is actually very challenging.

AI assessment note: “I don't know about surprises, but I would say I think from the outside”

Q What about then the biggest surprises about the move into VC?

A I don't know about surprises, but I would say I think from the outside, it seems Easy. I think there's a perception that you like sit back and all of these people come to you and pitch you and then you're just kind of like picking and it's like, it's a laid back and easy job. And I think what's really clear to me, and I knew this going in, but is emphasized even two years in is to do this job at the highest levels. It's actually extremely challenging. There's a ton of hustle required. You have to get out there. The best stuff is very competitive and So I think there's a perception that it seems easy, and it may be that it's easy to skate by, but I think to do it at the highest levels is actually very challenging.

AI assessment note: “I don't know about surprises, but I would say I think from the outside, it seems Easy.”

Q Does that mean there's still great value in open AI?

A I think that's a good question that, that is, um, really interesting because if you think about open AI and anthropic and, and meta and Google, Um, and you know, and then there's a whole bunch of others coming, XAI and Mistral and, um, and so forth, SSI now. Um, you know, I think the foundational model war, like, benefits us all in a way. Like, it's just, it's really, really good for, for consumers and people around it, because it's just, like, they're pushing the state of the art so much. Um, in terms of value accrual, like, you know, I think for Benchmark, like, we have a, we, we have no foundational model investments, uh, thus far anyway. Um, and so, you know, that could be very bad or very good. I don't know. Um, but we don't. And then two, we have a set of, like, infrastructure investments, which I think are, are really interesting. So, um, we have Cerebris, which is a semiconductor and systems company, um, for AI that we invested in and led their, or co-led with Foundation their Series A in 2016. So we've been working on it for eight years, um, as an example. Um, we have, uh, companies like Fireworks, which are an inference service and others kind of of that infrastructure software layer. And I think those like, you know, again, they're growing very, very quickly, like astoundingly quickly, um, and doing really cool things. But at the same time, you kind of have to ask li…

AI assessment note: “In terms of value accrual... we have no foundational model investments... I don't know.”

Q This is off schedule. This is like for true pros, but this has been so good. Okay. The hardest things in life or the heaviest things in life are not iron or gold, but unmade decisions. What unmade decision do you have that weighs on you most?

A I don't know. I can think of several little things that I would have done, but I don't know. We had early acquisition interest in Rockmelt. And I think that, that, that was probably something that I should have like leaned into more in retrospect. So like, that's, that's an example. And that would have been a very different, like, But in the scheme of things, like, looking back now, it's twenty-twenty-four. Would it have changed anything? Probably not. It would have been fine. Like, I, it would have sold the company two years earlier, or whatever, and maybe would have made more. And, you know, this is the whole, um, what's that children's story about the, the horse rider, and the soldier, and the conscription? Have you heard this story? It's a really amazing story. So, like, the, the summary of it is, is like, there's a, it's like ancient China, and there's a, there's a draft. And so, um, and the first part of it is, there's a family in rural China, and they get a horse. And, uh, This kid finds a horse and the villagers are like, oh my God, he's so lucky. It's so lucky. It's so lucky. And the wise man is like, we'll see. And the kid's riding the horse and breaks his leg. And then the villagers are like, oh my God, it's so unlucky. It's so unlucky. It's so unlucky. And the wise man is like, we'll see. And then there's a draft, a military draft, and the draft people come to this …

AI assessment note: “We had early acquisition interest in Rockmelt. And I think that... was probably something”

Q terms of, uh, sector. There are very different companies there from Confluent, Contentful to Cerubris. And I spoke to Bruce Dunleavy before the show, and he said that bluntly your breadth of sector mastering is completely unparalleled as an investor. And he asked a great question, I thought, which was, what is your learning process for entirely new categories? And how do you break it down and learn so fast?

A I am not a sector specialist and nobody at Benchmark is. And I think the, the fundamental idea with Benchmark is there's a small group of people, a small group of partners who are all equal. And right now it's five of us, but sometimes it's four, sometimes it's six, but it's basically four to six. Who cover technology. And the challenge with that, if you kind of think about it, is we can't be sector specialists because the, the sectors that have the most disruption and things are changing the fastest, like are changing. That part's constantly changing. So you have to be as, as a group of investors, you have to be, you have to be moving. You have to be looking at new stuff because that's where the disruption is happening. And so So then the question is, so like, I can't be a sector specialist. I can't be a, a semiconductor specialist, or I can't be, um, an open source specialist. Of course, we each have preferences and things that we like and don't and lessons that we, you can apply, but like it, but it, but we're, but it's not a specialty model. And I think about this a lot, and we should talk about it in the context of AI, but you can't have that. So then the question is, well, what can you evaluate on? And I think this is, this is it, this is it for me, which is, you can say, hey, This is an extraordinary entrepreneur. That's a, that's a, that's an evaluation that, that you c…

AI assessment note: “So then the question is, well, what can you evaluate on?”

Q Do you not think it helps provide a lens of focus to narrow your examination of where to spend time?

A You know, we very, very openly and regularly talk about, um, and, and have done like things that are just, you would say like, Hey, that's way off. Like that's a fifty million dollar check for, 10% ownership. Like, you know, it doesn't, it's not something that's not the core model, obviously, but the flip side is I look at, you know, you mentioned the Levin X. That's like an amazing company that we're super lucky to be part of. I think about Brett Taylor, Sierra. I think about Lynn's fireworks. Um, you know, I think about, um, you know, I go through and I like to look at these companies and I'm like, Okay. That's a, I like that AI portfolio. It's like, it's a bunch of infrastructure software companies. It's a semiconductor company in Cerebrus. It's a, um, and it's a, it's a few application companies as well. And like, I think, and they're all, it's a good kind of set. And so the foundational model rounds and some of those things have gotten like really, really large. Um, but you kind of look at some of the things that are happening on the ground in the early stage in AI, and it's like, yeah, it's totally doable, totally manageable.

AI assessment note: “You know, we very, very openly and regularly talk about, um, and, and have done like things that are just”

Q To what do you stand with AI? Do you think we are overestimating what we can do in one year and we're all getting ahead of our skis?

A You know, one of the beauties of this, one of the beauties of this in our model, like I think about if I go back to 2010 and 2011 for a second. You know, in that time timeline, that's when Snapchat, Uber, Twitter, maybe 2009 to Instagram. Um, like that's when we did the series A's and like Instagram, Snapchat, um, Uber. And whatever, a weird round in Twitter, which was a round, a rule breaking round, by the way, like at that time, it was crazy. The round, the round that Peter led in Twitter at that time was like technically a series C or series D. It was like at like 200 pre because the company had its history, right? With, with ODO and everything else. And so it was, it was a rule breaking round. It's a good example of exactly what we were talking about earlier, which is like, Yeah. You kind of have your like norms and then every once in a while you just have to be like, throw it all out and just do it. And like, and, and, and that was a good example. But you think about that, like that body of work, which is, which was obviously, um, tremendous terms of returns perspective and, and like fast forward to today and you're looking at the game on the field here and, you know, we have to kind of ask ourselves like, Hey, are there extraordinary opportunities and extraordinary companies getting built here? And if so, like, You, you, you just gotta do it.

AI assessment note: “are there extraordinary opportunities and extraordinary companies getting built here? And if so... you just gotta do it.”

Q He's amazing. I have such a man crush on Peter. I haven't told him, and so it's lucky that this isn't the podcast. But, uh, my question to you there is, is that not dangerous? Should a partner not be the counterbalance, not the Duracell battery to your energy?

A Well, I think both are true. My partners have over my 10 years at Benchmark. They have kept me out of countless companies that like, it's amazing. You asked the sector question earlier. We were talking about it is like, there's, I spend a lot of time trying to like understand chemistry Um, my chemistry with an entrepreneur and, and try to figure out, like, am I gonna love working with this person? Do I believe this person is a learning machine? Um, you know, and, or not. So I, I spend a lot of time on that. I spend a lot of time trying to believe, like, do I think that insight is cogent or not cogent? Um, like, the insight that they have and, like, does that, does it hold together? And, and I spend a lot less time on, like, the sector specifics because I just feel like I, I, like, If I'm an F on a sector, like if with best effort, I can get to a D plus, like that's not good enough. Like, and so I just rather like not, and I actually think this is, this is maybe contrarian and total aside, but like, I think this is why the memo writing culture at a lot of firms gets you in trouble because you put a lot of information You, you basically are, are, it encourages putting a lot of information that is like third and fourth order stuff Into document as if that is impacting your investment decision, where like most of these investments, there's really like one or two questions that real…

AI assessment note: “Well, I think both are true. My partners have over my 10 years at Benchmark.”

Q To what do you stand with AI? Do you think we are overestimating what we can do in one year and we're all getting ahead of our skis?

A You know, one of the beauties of this, one of the beauties of this in our model, like I think about if I go back to 2010 and 2011 for a second. You know, in that time timeline, that's when Snapchat, Uber, Twitter, maybe 2009 to Instagram. Um, like that's when we did the series A's and like Instagram, Snapchat, um, Uber. And whatever, a weird round in Twitter, which was a round, a rule breaking round, by the way, like at that time, it was crazy. The round, the round that Peter led in Twitter at that time was like technically a series C or series D. It was like at like 200 pre because the company had its history, right? With, with ODO and everything else. And so it was, it was a rule breaking round. It's a good example of exactly what we were talking about earlier, which is like, Yeah. You kind of have your like norms and then every once in a while you just have to be like, throw it all out and just do it. And like, and, and, and that was a good example. But you think about that, like that body of work, which is, which was obviously, um, tremendous terms of returns perspective and, and like fast forward to today and you're looking at the game on the field here and, you know, we have to kind of ask ourselves like, Hey, are there extraordinary opportunities and extraordinary companies getting built here? And if so, like, You, you, you just gotta do it.

AI assessment note: “are there extraordinary opportunities and extraordinary companies getting built here? And if so, like, You, you, you just gotta do it.”

Q Do you not think it helps provide a lens of focus to narrow your examination of where to spend time?

A You know, we very, very openly and regularly talk about, um, and, and have done like things that are just, you would say like, Hey, that's way off. Like that's a fifty million dollar check for, 10% ownership. Like, you know, it doesn't, it's not something that's not the core model, obviously, but the flip side is I look at, you know, you mentioned the Levin X. That's like an amazing company that we're super lucky to be part of. I think about Brett Taylor, Sierra. I think about Lynn's fireworks. Um, you know, I think about, um, you know, I go through and I like to look at these companies and I'm like, Okay. That's a, I like that AI portfolio. It's like, it's a bunch of infrastructure software companies. It's a semiconductor company in Cerebrus. It's a, um, and it's a, it's a few application companies as well. And like, I think, and they're all, it's a good kind of set. And so the foundational model rounds and some of those things have gotten like really, really large. Um, but you kind of look at some of the things that are happening on the ground in the early stage in AI, and it's like, yeah, it's totally doable, totally manageable.

AI assessment note: “it's not something that's not the core model, obviously, but the flip side is”