Things Andrew Says on Podcasts

VCs don't do that, right? And that's an industry-wide problem. It's also a lot of people have false confidence where they'll see these updates. And I don't know, like the trend you saw kind of in 2022, 2023 were people entering the venture capital ecosystem raising funds who didn't have investment management experience before. One of my beliefs is that you have to see the dark side of this industry to be able to know what the good side has been. I have sat with founders over drinks and had to say, look, we're liquidating your company. We're winding this down. I've seen late stage companies go to zero. In our own portfolio, I've seen great success too, right? You have to know the highs and you have to know the lows. If you don't know the lows, it's very easy to think that it's a high all the time because of the market. But those lows always catch up to you. And that's the lesson you learn with ventures. Time is cruel and the market is indiscriminate. And a lot of things that you think will go well won't. So, but that's why I think a lot of people keep cashing in companies when they really shouldn't against their own judgment.

No, I mean, so we used to coach tastes and flavors at Caltech, right? That's the first unit you do, the Caltech cooking classes. You would teach about tastes and flavors. And the number of people who gaslight themselves into thinking things taste good that do not is really high. Now, unfortunately, and why I've struggled as a food investor, I do better. So I will tell you, I have one investment in alcohol that's done really well so far. Where I struggle with food is I have weird taste in food. I eat a lot of weird food. I was really into it. Tell me, tell me.

I mean, I think everything's been eaten up by AI investments. I think it's almost impossible to justify a SpaceX valuation at 1.6 trillion. It's not, and that money normally would go elsewhere if it right there.

Yeah. I mean, look, I think venture is a high-risk asset class. And especially when you get into biotech, the outcomes tend towards binary. And for better or for worse, late-stage venture firms tend to underwrite biotech in a lot of the deeper science plays as if the risk has been written off when it often has not. And so you'll see these massive rounds and massive valuations for things that still contain a fair bit of risk. I'd look at, you know, Motif is a great example of that. Another one, like Relativity Space, for example, is a perfect look at saying, okay, look, there's promising technology. There's great science here, but it doesn't quite have the same stability. And venture at a growth stage, your investors are optimizing for something a little bit different. They still want that high upside. But if you're taking these massive bets, you know, you have to have a portfolio that you can construct around it. Oftentimes, people don't manage the portfolio construction game very well, which means that there's not cash available to keep things afloat when things go wrong. And if you manage that poorly or you don't underwrite it well to start, and there are technologies, I mean, I'm not going to say that on the podcast. There are one or two businesses which have…

Which leads me to believe this is going to be a hardware battle, not a software battle. I think there's probably a place for an OpenAI and an Anthropic in that world. However, I've come to the new conclusion that I think Apple will be the big winner here because their hardware is already the most advanced, especially for on-device. I mean, my MacBook runs most of your good open source models. And as those models get better with smaller context windows and cheaper, Apple, and as their hardware increases in position too, they're incredibly well positioned to beat everyone. So that's my hardware guy, too. Yeah, I do a lot. I do a lot more hardware. I don't do a lot of like LLM-based AI. I don't think I do anything that you would call a rapper in any capacity. I do think data will be incredibly important in the next 10 years. I continue to have that belief, proprietary data to do traditional AI. And that was always my belief when I set out to build the firm is that the LLM AI wave catalyzes the adoption of a bunch of AI technologies that we really should have adopted 10 years ago, but didn't.

give the two-second version, which is we're on a bubble. I don't know when the bubble is going to burst. I don't know how it's going to burst. I have said for a while that we're spending exponentially more money for a logarithmic increase in performance. I think you're seeing that firsthand. There hasn't been a step change in performance. Although, I'll be honest, and one thing I did not predict, tool use did improve the models a lot. And the ability to use- What do you mean by that? Just like web search integrations, the integrations that OpenAI has with, like, I, God, I love Codex. This is not part of the sponsorship here. I use it for literally everything. It integrates with Adobe Premiere. Like, I've used it to added videos. And because they've just spent so much time doing the computer use, the integrations, it does increase the power a lot. Now, from a strict API side and from a strict model quality and response side, it is a maybe linear improvement over a year, if even. They're spending hundreds of billions of dollars now. So I've had that thesis for a while. And then similarly, and this is, again, not something I necessarily predicted, but had something because open source has caught up a lot more, which…