The Exchanges

Q a little bit about, um, benchmark. I think it's interesting for people as well to, Get a sense of, I mean, you, you guys, you know, practice the, the, the, the, the art, certainly, not the science, obviously, you know, in a different way from precisely and Andreessen Horowitz. I mean, you're pretty much the opposite ends of the spectrum. So how do you, how do you think about that?

A Well, I, I start by saying I think it's great that there's variance in our ecosystem, that it's not just a monoculture, and, and, and Driesen Orwitz, as an example of a firm, I think has, um, invested in things that, that they can scale, uh, which I, I think, you know, if you look at the list of services, there, there's a, there's a good argument that you can scale BD, you can scale PR, you can scale access to CIOs. Um, our firm actually has a very different structural kind of Uh, makeup, and I use this analogy. It's not totally fair to them, but like a jazz band versus a marching band. Um, it's not fair. I, I don't, I'm not sensational to say it that way.

AI assessment note: “like a jazz band versus a marching band”

Q you do spend a lot of time on there, like, areas that sort of, at least conceptually, you find more interesting. I don't know, this whole thing around, you know, machine learning and AI and the application layer of all this infrastructure. Is that, is that interesting? Or you invested in CityMapper as well, I believe, which is a, Super interesting company, which is a data application type company, so.

A Yeah, so I think we've seen that There's a chance to create, you know, differentially, radically better product experiences with data at scale, combined with machine learning. And, you know, I love Stitch Fix, which is just an example, that's an application in many ways of, of, of machine learning, predictive analytics on top of, ah, on top of large data sets. The, the thing we, we struggle with is, um, and I think you, you, you've covered this in the past, but it's worth just putting a spotlight on it. You have to get the data. So, in an application company, it turns out, I think, that the at-scale providers, companies like New Relic, or people who are already moving terabytes of data, or Uber, they can apply the technologies for machine learning, um, and, and, unfortunately, this is the, this is the issue I think we sort of face as investors, um, they can actually hire the best people, they can afford to pay the most, it's worth most of them, because they have the best, you know, the highest marginal return on that, that, that, you know, data scientist, and so, you see this, this sort of imbalance of power Uh, and, and machine learning is the game of kings. You know, it's not, deep learning or deep mind is part of Google. Like, as an independent company, like, they could, Google can pay them more, so that's where they end up. So, the, the application issue is, how do you crea…

AI assessment note: “I think we've seen that There's a chance to create”

Q so is that right? So you, um, so a, You know, a cockroach, for example, um, Spencer who spoke at the event, uh, you know, a few months ago, so, so it's literally, you, you don't have a thesis around, you know, the best of both worlds in SQL and no SQL converging. It just happened that Spencer walks into the room and he's incredible, so you invest in him?

A You know, it's, he's in the room, so I have to pretend like I had a very deep thesis. Um, uh, indeed, we've been looking for a SQL interface on top of a, Um, distributed transactional key value store, and, and then, and then he came and I saw it all in, no, I mean, come on. So, here's what I sensed, is that Spencer had worked on some of the most important, I think, foundational work of, of the internet, and broadly computing at Google, and what Spanner and F-One did was, you have to be an idiot not to see it, I mean, it's, it's, it's the scale, and you wrote about this, I think, in an extraordinary blog post, and, and that the work is being done at Facebook, to a degree Twitter, Um, and I had a, I had a strong intuition, I think we had, collectively as an industry, had an intuition that those technologies were not going to be trapped inside of the big internet companies. So, which, but what you needed was not a Google dweebish engineer like what you see in, in some things like, ah, well, I'm not going to pick on Kubernetes, um, but, but, ah, you needed as a leader, an entrepreneur, and, and, and Spencer had left the comfort of Google to pursue his vision and his passion with the startup, and much like, you know, Travis at Uber or, or Evan at Snapchat, the first one doesn't always work. And it's like, I love backing company two and three, and some, and number one, you get lucky …

AI assessment note: “he's in the room, so I have to pretend like I had a very deep thesis.”

Q in the, in that category, so from Hortonworks, New Relic, um, and through your, uh, partners as well, so there's Confluent, uh, Domo, like all those guys, so you guys are, and, of course, um, so, uh, So even though you, you, you may not like the term, uh, is that, is that an actual, um, focus of yours, or is that just a different flavor of enterprise, um, infrastructure?

A Yeah, you know, it's, it's, uh, I'll come back to Excel, not that people care, but Excel and benchmark. Excel, I was taught that, that, okay, we need to have a prepared mind. You do extraordinary work, much better than anyone I, when I was at Excel than we did. Uh, and really thinking about a segment, a category, and its coherence, and, and that would lead you to be more, Um, Louis Pasteur said, this is what I was taught, chance prefers the prepared mind, or luck prefers the prepared mind. So, I came to benchmark in, um, I didn't know that I agreed with that, uh, and my partner said, that, like, don't you do that shit here. Throw the crystal ball out, you can't predict anything. What you could do is you could recognize when lightning strikes, and if you get good at that, you'll get behind some momentum. But more importantly, you should, you should be recognizing character types that you want to work with. A little bit about the, the nature of the people. So, so I, I actually got into Elastic Surge through, um, a collision of accidents, uh, investing in Spring Source. I was, I was an early investor in, in Open Source and JBoss, which led to Spring Source, which, so then the, the problems of what we now call big data were attracting a next generation of entrepreneurs. And one of the great things about, about our world as, as investors is that, um, Um, we had to actually follow, i…

AI assessment note: “in hindsight looked like they were deeply strategic”

Q So speaking of open source, so a lot of the investments that, um, you made in the enterprise, Hortonworks, Elastic, Docker, Cockroach, um, so those are open source businesses. What, what are your, um, what, what do you find open source companies attractive from an investment standpoint?

A So I, I started a venture business in 99, and my partner Bill, if he was here, would say, yeah, just like right now, you know, nuclear winner, and, and, Money disappears, and you know, all of a sudden, you assume a bunch of things in the background that just shifted an instant, and, and so by oh one, ah, two thirds of the companies that, that we were involved with were getting shut down, and, and you left behind this residual of, wow, we got way out ahead of ourselves in, in putting dollars into sales and marketing, and not into engineered product, and, and it was a wasteland, literally a wasteland, of oversold product, you know, companies that, that we forget now, um, Because the lists are so long of, of, of the bullshit artists. So, I, you know, I had this belief that, um, the internet collapses time and space between an author and, and its consumer, and, and, or their consumer. And the software distribution model, when we make an investment, it's still somewhat the case that if we put a, for every dollar we put in, something like 60% of that goes into selling the technology. And that nobody sold anybody Facebook. Nobody sold me Google. Nobody sold me Twitter. Right? There was a, a dynamic of adoption of habit formation, and that created an, ah, and then ultimately created a virality in the usage of the product, which I think you can actually apply to open source. So, so, ah,…

AI assessment note: “virality in the usage of the product, which I think you can actually apply to open source”

Q data world, but, um, want to start maybe at a high level, and, you know, particularly in light of, um, your, your incredible track record about, you know, how you invest. I mean, there's a bunch of entrepreneurs here, and the number one question is always, well, how do you, how do you pick? I mean, is it the people? Is it the idea? If that's a people, then how?

A Um, yeah, you know, I, uh, I've been doing this job for, Almost enough time that I can't do it anymore. In the sense that, you know, there's a, one of my former partners at Excel said, when you see the problems, and you're not naive anymore to the opportunities, and you don't start with the question of what could go right, and if you do, as I said, almost 20 years, you start to see a lot of things that could go wrong. Um, and, and the instincts that I taught, was taught early on, I, I still worship at today, which is that I, I look first and foremost for the same thing that a founder looks for when they, Start a company, which is that, that moment where you start to dream about this possible world or this possible product or experience, and it, it terrorizes you. You can't sleep at night. You, uh, it brings, it elevates your, your, your energy levels, your blood pressure, everything, and, you know, that emotion requires a high degree of naivete, which is something that's really essential in our business, and I find that there's a little more naivete on the West Coast than there is on the East Coast, But that's changing. You know, MongoDB was, was a company, my partner, who's now retired, who sees more problems than opportunities, and I say that in jest. There was always a hundred reasons why a company wasn't gonna work. So, so what I look for is that first and foremost, that so…

AI assessment note: “what I look for is that first and foremost, that sort of feeling”