00:06:35 · Eric Newcomer
[00:06:35] S1: I want to dig into each of the categories for a little bit. AI infrastructure in particular. There's clearly a wave right now of these inference companies, whether it's Base10, Fireworks, Modal, and then a lot of companies around the inference layer. I guess the argument against them is just lots of revenue growth, but then people wonder about the revenue quality. That's one piece of the infrastructure question. The other is sort of the, I don't know, heartburn from the crypto era where it's like there was a lot of inference investment in something that didn't really take off. And I guess do you worry with agents that if we build sort of the infrastructure layer for something that hasn't been built out yet, there's the risk that it never fully gets built or the world that you expect doesn't come exactly that way. So I'm curious for your perspective on both
00:07:29 · Aydin Senkut
[00:07:29] S0: those Yeah. Mean, I think the way to look at inferences, the moment AI became interesting is when it transcended from one token in, one token out. When you think about how we think, like the way that brain works, there are some things that you take extreme shortcuts and some things where actually like spending literally all of your neurons are like focusing getting it right. And the big difference to crypto in my mind and the reason why inference is really important is now all of a sudden you're getting real jobs done and like you're making real decisions. And that honestly, like if you think about what human beings like use their brain for, like we make decisions. And even when you're not realizing it, when you're like seeing something, hearing something, smelling something, each one of these things have like hundreds, if not thousands of processes that is not completely oblivious to you. And all of this stuff like includes like making some decision and like, oh, like I'm going use this signal and not that signal. So in some essence, inference is like a key unit of how our brain work. And now we have managed to scale it. Then you think about, well, human brain, even with its ingenuity, can do X. And now we can be in a world where that X can be 10X or 100X. And the reason why that's really important is that can produce uncapped value, right? There might be like a lot of things we do that previously we couldn't do or we could do it but we could not scale it. So unlike crypto, that's why I feel the whole inference thing is not a toy. And trust me, like I'm speaking, I was an early Google employee and I'm like, God, please let this be a $10,000,000,000 company, became a $5,000,000,000,000 company. And that's just given human beings information, access to information, equalize it, scale it. And now what we can do is if we basically change it and say now we can allow human beings like make better decisions or even do their work for them. It's just a new era. But the reality is if you also ask me, do you ever envision a world where Google is going to spend $100,000,000,000 in CapEx a year? And I'm like, what are you talking about? Like that is never going to happen. This is a company that like...
00:09:30 · Eric Newcomer, Aydin Senkut
[00:09:30] S1: That's going to ruin the beauty of the business [00:09:32] S0: that it does happen. This is a company that didn't even want to do advertising. So there is real... I think the scary thing is you cannot not spend that money. I feel like we're in some kind of like survivor game and like the winner is going to take a lot of value but like five, 10 people are spending the money. And at the end I'm not sure if all five, 10 players will the economic returns will be equally distributed. So there will definitely be some changes. But look, every time we get afraid of momentary stock market changes or whatever, we're to see some shock in the economy, whatever, what follows is a 10x improvement. 2008, 2,009 people are like, Oh my God, this is the worst thing. Stocks lost all the values. But then that became a great foundation and some of the most interesting and valuable companies got created. So I'm an eternal optimist and I do think there is some really exciting stuff to come there. [00:10:28] S1: Are you investing in agent infrastructure in particular companies to allow some sort of agent economy or are you
00:10:28 · Eric Newcomer, Aydin Senkut
[00:10:28] S1: Are you investing in agent infrastructure in particular companies to allow some sort of agent economy or are you [00:10:35] S0: looking I for more am. The middle and the way that we're thinking about it like I think all you have to do is like I don't even think you need to be like the most incredible expert. You just have to imagine a world where like instead of just humans now you have like millions, billions of agents and millions, billions of robots. What changes in that world? And so a lot of it is around developer ecosystem and it used to be just people that can code computer science degrees. Now you have citizen developers and the agents themselves. And it's awesome because these agents, there are so many ways that Citizen developer. [00:11:08] S1: I don't know if I've heard that phrase before. [00:11:09] S0: Citizen developers isn't like, look, think about all the replets and lovable of the Right. [00:11:13] S1: No.
00:11:14 · Aydin Senkut
[00:11:14] S0: Get People that didn't even like can barely speak English or their language is now coding and creating apps. How cool is that, right? It's democratization. Also I feel like every great Felicis company is democratization of something. Canva is democratization of design. Shopify was like democratization and scaling of entrepreneurship and commerce, make it available to people like citizen entrepreneurs. And now we have citizen developers. But I think more importantly than that, one of the things that fascinates me imagine I don't think people are appreciating the potential of commercializing the agents because the agents don't know the world the way we do. Think about all the years that brought you to here, how many things you learned you don't even think about. You know what a university is. You know what a country is. You know like a location that's a city or any type of simple thing. Like you know colors whatever versus the agents have to learn the context and everything from scratch. In a world where search was only consumed by humans, now imagine a world where search is no longer consumed the way humans do it but via an API by billions of agents that will do like a million more searches compared to like a single human. How exciting is that world? Then all of a sudden you're like, Wow, we created this amazing abundance in terms of AI but now security is the biggest threat. Well, how do you manage a world where even the smallest bad actors have insane AI tools in their hands? Now this is going to be really important. If you thought identity management, whatever, that was a big area. It's going to be like a thousand times more important critical bottleneck with what's going on. And so every time you have this chaos and unexpected situation, we don't know how this is going to turn out. It is the best foundation for some of
00:12:59 · Eric Newcomer
[00:12:59] S1: the biggest disruptions to happen and surprise us hopefully to the upside. I wanted to turn to the second bucket, which is the, don't know, biology, health. Where do you see the most opportunity? There are companies investing directly in drug discovery. That's obviously like a hard, slow business. Like where do you think there is opportunity in that sort of, I don't know, the do not die business? Like if you were going to fund that today, what are the approaches that you're most bullish on?