The Exchanges

Q podcast on YouTube, Spotify, Apple, or wherever you listen. The link is in the description to sign up. You more formally bucket this out, um, In terms of the decade of data, AI as the new platform, and decentralized infrastructure as database. Could you break down each of these and maybe provide a portfolio company or two to better, you know, emphasize and, uh, just, you know, expand on that?

A Yeah, for sure. So the decade of data with the modern data stack, postmodern data stack, this is the idea that every company or every product is producing data and they need to make some, something useful out of it. There's raw material like an ore, and then it needs to be refined to make steel, let's say. Um, and so over the last 10 years have been many companies in space. We were lucky. I was lucky to be on the board of a company called Looker that Google bought for a 2.7000000000, and we backed the The, many of the original team of that company in a business called Omni. And, uh, that business, if you think about like classic BI, there were four companies in, in the year 2000 that were building BI. They reached worth about two billion and they controlled it. They locked it down. It was totally centralized. Very, very difficult to get access to report. A company called Tableau that Google, that Salesforce bought for about sixteen billion, came out of Stanford and said, everybody can have access to BI. And so there was this pendulum swing from centralized control to decentralized control, and then we invested in Looker, which was more centralized control, but on cloud data warehouses. Omni is a combination of the two. We're thrilled, thrilled to be partners with Jamie and Colin and the rest of the team. Um, in that AI is a new platform. We look for three things when we invest …

AI assessment note: “So the decade of data with the modern data stack, postmodern data stack”

Q So where exactly are we in terms of AI adoption? There's been lots of profile funding rounds over the last few years. Are companies actually delivering any value? Where are we at?

A I think one of the biggest adopt, various adoptions of AI is that is really expensive. The ROI in most cases is not there, but like you, so let's start with the consumer. Let's organize it. In the consumer world, I think AI adoption is taking off in a really big way. You look at like, uh, 18 to 24 year olds, something, the Verge ran a survey, 75% of, of those people default to generative search, uh, as, as a place to go. So that, that's already happening in a really meaningful way. I think you have like, uh, OpenAI announced three hundred million. I think MAU on ChatGPT. So you're, you're kind of approaching the billion user number in consumer AI. I think we'll definitely surpass that this year. And that's awesome, right? It means that Google search model is totally up for grabs. The ad dollars there, complete destabilization of the SEO, SEM market. That's really exciting. A lot of market cap, uh, is now becoming loose within the enterprise. I'd say it's earlier still. Uh, because the use cases within the enterprise are more challenging. Why is that? Well, the range of acceptable outcomes within the enterprise are much narrower, right? Um, I remember I generated this image of a Kitty cat on a fire truck, and it was a Tonka truck. It wasn't a fire truck, and the kitty cat was not, it was a little, you know, five toes or whatever it was, and that's okay for a consumer use case. M…

AI assessment note: “In the consumer world, I think AI adoption is taking off in a really big way.”

Q And, okay, so uh, going back to construction of the fund, you brought in a really interesting individual at the founding moment of the fund. So you brought in Lauren Demuse, former Palantir exec, To help build an institutional grade firm from inception. How did you meet, and what practices did Lauren introduce and build in the firm, and how does this support the data driven strategy?

A Lauren and I have known each other. We worked together at Google a long time ago, and, um, we just stayed friends, and I've watched her career blossom, and as you mentioned, she was a Palantir. She worked on the healthcare practice there, and she's no stranger to architecting very sophisticated data systems, and so the, the goal was for someone who really understood technology, really understood data architecture, to join the firm and manage the intelligence team. And, uh, she's been phenomenal. We architected, or she architected, uh, the system that we're currently using today. And, uh, and I think, you know, couldn't be more excited that, that she's on board and, and all the people who work as part of the ERI has really come together.

AI assessment note: “We worked together at Google a long time ago, and, um, we just stayed friends”

Q you're definitely a Swiss army knife of product testing because I've followed your blog for many years. I feel like every time I read it, you're doing something else, you're trying something else, and it's really fun to ride along. I'm just really curious, like, internally, like, what is the operating system for that? What is the stack? And what are the different programs and tools that you regularly use?

A Yeah, I was, I mean, um, I was using a command line email client in the terminal, so I'm not using Gmail, but using this really, really old piece of software called Neomat for a long time, um, because actually during COVID, I wanted to learn how to use the command line, and so, okay, why? Uh, well, I remember, I remember meeting the Dropbox team at the Seed when they came out of Y Combinator, and they had commercialized Like a Linux command line utility called rsync that allowed you to synchronize the status of two different folders. And they ended up building, you know, whatever, a five, ten billion dollar company on top of that single protocol. Um, and I think there are a lot of brilliant programming ideas that exist within the terminal that had been invented 30 or 40 years ago that are constantly surfacing and being exposed to, to, uh, different people. Like superhuman email client, uh, is a great product, and it has a lot of what was built into Mutt, which is a command line email client, or Neomutt, which is a new version of it.

AI assessment note: “I was using a command line email client in the terminal”

Q I know I should have these memorized, but I'm gonna run through the hyperscalers names and some more. So we have Google, Microsoft, Meta, Amazon, Apple. We now have XAI, OpenAI, Anthropic. As we move towards a more exhaustive compute environment with inference booting up, advanced agentic ecosystems, where do you see the power dynamics between hyperscalers and these, you know, independent LLMs?

A I don't know if it was a pun or not, but the word power is incredibly important. Um, so electricity is absolutely critical. Uh, somebody, I was talking to a friend of mine who spends a lot of time and energy, and he was saying, I was asking him, like, compared to an electric car, how much does a GPU consume? This is about a hundred X. So if you're building a data center, if you're building a power plant for a bunch of electric cars, typically around like five megawatts, if you're building a power plant for significant data centers at five to 30 gigawatts. And so the, the major constraint, one of the major constraints today is just energy. So you see the reinvigoration of like Microsoft tried to start, um, Three Mile Island again, the, the pressure on the nuclear regulatory, uh, agency in the United States to approve modular nuclear reactors. And so I think that that's critical. Just truly the energy will be absolutely essential. The next is just the CapEx. I mean, you have like Microsoft and Google and Meta each deploying something like 70 to eighty billion, uh, in capital expenditure, just buying GPUs and building data centers, uh, in just next year. And they were, I think, you know, 60, 50, and 45 this year so far. So amassing those GPUs. Microsoft is still compute constrained. Uh, and they project, I think, all the way through twenty-twenty-five, they'll still be needing to …

AI assessment note: “the major constraint, one of the major constraints today is just energy.”

Q we are one to three years away from something like AGI, which could plausibly replace human labor, um, for some set of tasks, and these tasks can range from like healthcare to, you know, operating a bakery shop or, you know, maybe some more like Computer work. Um, do you think this is a Jevons paradox, uh, situation where efficiency gains lead to more consumption? Where, where are you thinking?

A Yeah, I think it's really exciting. Um, I think, uh, I mean, one of the, one of the interesting questions is like, what is AGI, right? It's like, is it talking to an AI that's just as sophisticated as a high school student, in which case most models are 80% as smart as a high school student? The open AI newest model has PhD level knowledge on many different topics, including like math and programming. And so there's like the knowledge part, you know, many different kinds of intelligence, knowledge retrieval is one, and then there's the reasoning part. I think once we get to a place where The AI is no longer like an intern, but is more like an employee, then we'll start to see some pretty significant productivity gains, and one of the things that's missing there is long-term memory, right? Molly, you and I are talking, and I learned something about you, like, I don't know, your favorite kind of food is Italian, and you really like, uh, penne. I'm going to remember that because it's memorable, and an AI system will forget, like, your goldfish's memory, and so, Uh, we're working on that kind of long-term memory, which will be absolutely critical to having sort of longer term workers. But I don't think, I think the technology will be here much sooner than anyone can take advantage of it because all of us have to look at the way that we work, the way that we answer emails or make ou…

AI assessment note: “one of the interesting questions is like, what is AGI, right?”