Transcript
0:00 Have you met anyone in your travels out here that you would say is like way more bullish than you? And if so, what do they believe that you don't? I mean, essentially everyone out here is more bullish than me, man. and then when I talk to people, whether it's people at the labs, whether anyone in this ecosystem, like I'm like bearish relative to essentially everyone. So, legendary tech investor Gavin Baker just spent the last week in Silicon Valley with one goal and one goal only.
0:35 Find something negative. Find just one piece of actual evidence that maybe all this is finally starting to slow down. But when he sat down with Patrick Oshanesy on the Invest Like the Best podcast, that's not at all what he had to report. in fact, just the opposite. And I think what he heard gives us a really interesting glimpse into just how differently the people closest to AI view the future. It's almost as if there's this growing cognitive dissonance between those following Frontier AI closely and everyone else.
1:14 Let's get into it. All right, so just for a little bit more context here, Gavin actually went on this trip to Silicon Valley during that slight pullback when semiconductor stocks were getting absolutely crushed. This was also right around the time Leopold Ashen Brener's fund reportedly got caught overleveraged in the sell-off and was essentially forced into a deal during his wedding. But again, Gavin's entire goal was to find just one negative thing. He can actually pinpoint some hard evidence that maybe the market was seeing something he wasn't. And well, here's what he found.
1:56 >> And I have it like my main kind of mission out here this week >> was like pressure test. >> Yeah. >> Yeah. >> Find tell me something negative. Yeah, >> you know like you know the question I asked you have you is there one negative quantitative metric you've you've heard has been what I've been asking everyone >> the main thing people are saying is the anthropic like the third party data suggests that the anthropic like curve started to go off of its trajectory a little bit that's like the only thing that I >> I think I think that's I think that may very well be true but then you have open AI and open source massively accelerating >> yeah the complex you look at the sum, it is net accelerating. May I don't know that it looks the same. I think it may have accelerated like I think open source is a little bit of a you know they talk about dark matter in the universe like open source is kind of dark matter to the public markets. You know, it's hard for public markets to measure it, >> but like if you just track what these inference clouds are saying >> and you know, these are people saying things on podcasts or people saying things in meetings. They're not, you know, audited financials, but like demand is clearly accelerating, which makes sense because you had this huge capability leap with GLM 5.2 and Kimmy K3, which I think we're going to see continue. I think you're going to see Nvidia bring Neotron steadily closer to the frontier. It has been a very like it's been a humbling challenging month and but just it's also like wow I've kind of pressure tested every assumption the underlying fundamentals are improving and stocks Nvidia is actually as we record this at its lowest forward PE of the last 10 years. crazy.
3:46 >> The only time the cibbies have been cheaper were liberation day deepseeek and that was those were kind of Vbottoms. >> and that means to you just that the market thinks they're significantly over earning. >> Yeah, the market 100% thinks they're significantly over earning and you we need to be humble. >> Maybe they are. >> Maybe they are. but like my kind of mission out here this week was to look for negative data points as hard as I could. And normally you come to Silicon Valley and you know there's a mixture of like okay here's here's something negative here's something positive d on balance it's positive you know tech it creates value over time but I haven't been able to find one that is like a quantitative metric. So yeah, after basically spending the entire week trying to poke holes in his own thesis, Gavin says he still couldn't find the negative quantitative data he was looking for.
4:46 But this is where things get a little weird because if you know Gavin, you know he's already extremely bullish on AI, especially after a pullback like the one we just experienced. He literally says that looking at these markets right now makes him feel like a foolish optimist. And yet once he actually started talking to the people in Silicon Valley, he somehow found himself on the other side of that equation. Check this out. >> Have you met anyone in your travels out here that you would say is like way more bullish than you? And if so, what do they believe that you don't?
5:21 >> I mean, essentially everyone out here is more bullish than me, man. I like I mean, you know, I read this thing that Doresh wrote and I was like >> the 3x compute price thing or whatever. >> Yeah. Well, he was I forget what it was. >> No, no, it was like 15x or something. >> Yeah. But no, but just basically that you know, renting an H100 for a year would cost $250,000, you know. >> and that's 15x the current spot or something.
5:46 >> Exactly. >> Like wow, you know, that was just like not >> in my book. That wasn't in my, you know, forget my like basian probability space of expected outcomes. That wasn't even in my considered but dismissed his totally unlikely outcomes, you know, and then that guy is, you know, he's very dork. He's very smart guy. He's very plugged in. And and you know, and then he pointed out that like, hey, the, you know, something like I think he just said margins on compute are going up.
6:20 the amount of compute is going up and inference margins going up. And if you multiply those three, that's how you're getting this crazy acceleration in the sum of the labs plus open source. Although obviously open source, the margins on open source are not really going up. But I mean, >> so everyone's marvelous. >> Yeah. Yeah. I like, you know, I just I look at what's happening in the stock market and I feel like a foolish optimist >> and then when I talk to people, whether it's people at the labs, whether anyone in this ecosystem, like I'm like bearish relative to essentially everyone, which is just a strange state of affairs. So the obvious question then is what are all of these people seeing?
7:09 What is making the people closest to the frontier so confident about where this is going? Well, Gavin actually gives us a few clues throughout the interview. And probably the most interesting came when Patrick asked him what he was hearing about the technology itself. Because apparently there's one major breakthrough in particular that has a lot of people talking. something that Gavin himself called his biggest technical takeaway from the entire trip.
7:42 >> There seemed to be like a lot of people seem to feel like they are very close to solving continual learning and sample efficient learning which we've talked about before. And it is possible that if those are solved that you know could that be like a temporary like kind of like discontinuity you know in demand if instead of you know having to like I think somebody told me that the like I was trained on effectively 20 billion tokens and then it's like these models are trained on 300 trillion tokens and if you know you can trade something on 10 trillion tokens and then let it out into the and learn sample efficiently.
8:26 Yeah, that that doesn't sound good for trading demand but like trading has a percentage of semiconductor demand to compute is going to asmtote to something not approaching zero but very small but I would say that is the most kind of interesting and you know who knows if it's long horizon or short horizon you know SSI says that they're going to come out you know with their their model in August you know there's this whole generation of new labs that are focused on this >> and this would be good for the This would be amazing to be clear. Yeah, >> this would be awesome for the world.
8:59 >> Yeah, we all want we want this for >> Yeah, we want this. It would be amazing for the world. And it's just it's hard for me to believe that that would actually be negative for AI infrastructure demand. But again, trying to be really really open-minded. I I would say that was probably like the biggest like whether we call it scientific or technical takeaway, >> but it's just, you know, it's also like >> you just don't know.
9:22 >> Well, yeah. also like Nvidia is heavily involved with all of these startups. >> Okay, so there's a lot to unpack there. First of all, Gavin is saying that multiple people he spoke to seem to think we're getting very close to solving continual learning or at least making these models dramatically more sample efficient. This could mean that far less data may eventually be required to train these models while continual learning would allow them to keep adapting and learning from experience after the initial training. And when you combine those two things together, you start getting something that at least in some ways looks a lot more like how humans actually learn. Because if you think about it, one of the biggest differences between our human brains, assuming you're a human watching this video, and today's AI, is just how ridiculously sample efficient and energyefficient we are. We're not only just a little bit more efficient. We're more efficient by many orders of magnitude. I mean, the human brain only runs on around 20 watts. And yet somehow we've had people like Einstein, Isaac Newton, Da Vinci, and so many others.
10:42 Even just what the average human can do on a diet of ramen noodles and leftover pizza is beyond what any AI system today could ever come close to. So what this tells us at the very least is that biological intelligence operates under radically different efficiency constraints than the systems we're building today. And so if AI researchers can figure out even just some of what makes that possible and then potentially even replicate it, well, that would be huge. And perhaps Ilia's company SSI has actually made some real progress on that front. And perhaps we'll even see it this month since, as he said so casually, they're dropping a model this month. Now, this is where we start getting into that split between the majority of regular people and the people actually building AI at the frontier. Because while people at the frontier are already talking about what comes next, what comes after today's training paradigm, in reality, we haven't even come close to fully deploying or developing the technology we already have. According to Gavin, the number of people seriously using agentic AI today is almost unbelievably small.
12:04 And yet, even so, we're already running into a pretty major constraint. So, and then you know that's happening to like a cutting edge of public companies and then you have this whole wave of AI natives and like they are leaning into this so hard and they're not hiring humans. They're just really putting it mostly into tokens and so they're not slowing down. And then you have companies on the east coast of America who have like barely adopted AI companies you know broadly speaking on other you know not in the coast who maybe are has cutting and then Europe who's like just trying to figure out how to regulate AI you know >> before using it.
12:48 >> Yeah. So just like there's kind of these differential differential kind of waves of adoption all happening at the same time. But the thought I can't get out of my mind is like I think I said it maybe last time but just Yesterday like I don't know 500,000 people in the world 250,000 maybe are using agentic AI and we're in an acute compute shortage that's you know there's seven or eight billion people on the planet. what happens when we go from 500,000 >> to 1% >> to 100 million you know to 500 million >> and then I do think it's it it is interesting you know a lot of people are just like okay well you know I I do think it's like helpful to post on X to see the push back and a lot of people are saying well you know where fundamentally is the okay we accept your argument that hyperscalers are a dirty compute reprices their operating cash flow is going to accelerate and maybe we could fund this but like who where's that operating cash flow going to come from where is the customer and kind of definitionally it has to either come from you know faster economic growth through productivity kind of Satcha's comments like either we're going to start growing 10% or we're not or labor substitution and for sure I think in a lot of these AI natives you're seeing labor substitution but not because They're firing people. They're just not hiring nearly as many humans. You know, the gross profit dollars per FTE and, you know, A16Z, Iconic, a bunch of companies that have done this work. You know, they're, you know, they're they're vertical. U particularly relative to past generations of startups.
14:34 >> So, yeah, it's easy to forget just how few people are actually using AI to its full capabilities. If you think about it, as he said, we're already in a massive compute shortage, and yet almost no one is even really using agentic AI. Sure, a lot of people are coding with AI and using things like codecs and clawed code, but how many of them are actually giving these systems complex tasks and letting them run autonomously for hours or even days at a time? Probably not very many. Just the sheer cost and compute required to do that is already out of range for many. And that's not even including the technical knowledge required to actually set these agents up and use them effectively. I mean, again, most people probably don't even know this stuff exists yet, which is just insane to me. So, when Gavin says maybe 250 to 500,000 people are seriously using Aentic AI today, honestly, that doesn't even sound that crazy anymore.
15:40 But what happens when that 250 to 500,000 becomes 250 to 500 million? Well, all of that compute eventually has to translate into real economic value somehow in order for it to make sense. And as Gavin points out, that basically leaves two possibilities. Either AI drives massive economic growth, or it starts replacing a meaningful amount of human labor. And realistically, it'll probably be some combination of both.
16:13 But either way, if adoption gets anywhere close to those levels, we're going to need a lot more compute, which means a lot more chips, a lot more power, and ultimately a lot more data centers. And ironically, this is where Gavin sees one of the biggest threats to the entire buildout and really his entire thesis. >> What's the worst thing that could happen in AI? Is it regulatory? Is it some sort of like I >> I think regulatory has to be the biggest risk. I mean, it's the most obvious risk. And so, that was kind of one reason I was excited to be here this week was to just like I I want to be scared, you I like I don't I don't want to feel like a lunatic, you know, watching these stocks relative to you know, get more cheaper thinking the expected forward returns are going up, you know, while you know, it feels like the on the ground fundamentals have like pretty materially improved in July relative to even June. But I still ca come away thinking like you know regulation it just has to be the biggest risk like you just can't ignore New York making a data center moratorum and just like we are we're living in this weird postfactual postlogical political world and you know and I mean I think the AI industry it has done a terrible job of PR are and I do think >> I think it at least realizes that now >> maybe if not fixed it it realizes it.
17:57 >> Yeah. But like kind of the narrative in Washington, you know, the the the political narrative, you know, I think amongst a lot of ordinary Americans is like data centers, >> they're going to raise your electricity prices, they're going to take all your water and then they're going to take your job. And the reality is like given the deals that are being cut now, when a data center goes in, electricity prices actually generally go down for everyone around there because of behind the meter deals and this is that like data center pledge that kind of Trump asked people to to sign. Generally the data center developer used to be they just had to build a like you know whatever they had to get the police department and the fire departments like you know new trucks and new cars and you know new body armor or whatever. Now it's like, well, we're going to build you a hospital, a school, a new police station, and a fire station, and we're going to lower your power bills. How does that sound? And by the way, the jobs are ongoing because it turns out that you kind of need these plumbers, electricians, you know, HVAC contractors, and this is like data centers are like are in a lot of ways the best thing to happen for bluecollar wages in my lifetime.
19:09 So yeah, whether you agree with everything Gavin said there or not, I think this is where that disconnect we talked about earlier becomes really obvious. Because data centers are almost the perfect physical representation of that disconnect between those following AI closely and everyone else. To someone like Gavin, a data center represents everything that could come from AI. more compute, more capable models, scientific breakthroughs, economic growth, potentially even an abundance of goods and services. But to someone who doesn't follow this stuff every day, that exact same building could represent something completely different. higher energy demand, billions of dollars being spent on AI that definitely isn't going into their pockets, and this looming fear that whatever is running inside might eventually replace them. And I'm not talking about just their job. So, it's the same data center, same building and everything, but two completely different ways of viewing it. And apparently that gap has become so wide now that some of the people inside Silicon Valley don't even realize how differently the rest of the world sees all this. Some can't even seem to understand it. And so like that story needs to be told along with you know like there are you know we we we we we heard us we we heard a story I think we talked about it last time about how AI is increasingly really saving lives curing rare diseases like we you know I think I can't remember if it was I I think it was at ASCO this year you know the kind of vibe you know the the vibe was like hey we've this is the most scientific breakthroughs we've ever seen at a single conference And for sure some of that is due to AI.
21:06 And so we need to like tell those stories like you know if you have a sick child you know a sick parent a sick loved one like AI meaningfully increases the odds of them recovering >> like and we just we we need it's everybody needs to tell this and I think people out here it's all of this is so blindingly obvious to them >> that they they assume everyone else already knows >> they they can't yeah they can't process that this is a true but wildly divergent view from most Americans.
21:49 and so like I think the industry really needs to tell its story better >> because this is like New York it just feels like is the first of many. And even in some of these deep red states that are super progrowth, they're just like, "Hey, you guys are not doing a good job telling your story. Then we can't we can't tell your story. If you tell your story though, we can retell it, but like you're the experts."
22:16 you know, if you like like something I've if you do not speak your own truth, no one else will. >> So, yeah, I honestly kind of relate to what he's saying here. The potential of AI just seems so obvious to me and like it's almost inevitable at this point. But I could also see how many, again, especially those who aren't following AI as closely, are extremely skeptical, especially when the AI companies themselves keep talking about how their agents are escaping and hacking companies, or how their models are too powerful and dangerous to release. I mean, put yourself in the shoes of someone who doesn't spend every day following this stuff. You're basically being told that AI could replace your job, consume enormous amounts of energy and water, and potentially even become dangerous enough that the companies building it are worried. Meanwhile, you're not hearing nearly as much about how AI could cure disease, accelerate scientific discovery, solve math and aging, and create enormous amounts of wealth for everyone. And part of that is just how our brains work. We naturally pay more attention to threats than opportunities. That probably served us pretty well when there were actual predators lurking around us. But today, that same instinct gets exploited every single time someone needs you to click on a headline, video, or post. So, it's not exactly surprising that a lot of people aren't the biggest fans of AI.
23:51 That's not to mention either that humans generally don't like change, but at the same time, it's also not surprising that the people seeing the frontier advance firsthand might have a completely different perspective. And I think that is really the most interesting takeaway from this entire interview. Maybe the people Gavin is talking to really are seeing something that the rest of the world hasn't caughten on to yet. Or maybe Silicon Valley has become so convinced of its own vision of the future that it can't see outside its own bubble. The truth is probably somewhere in between. But either way, that gap seems to be getting wider. And I can't imagine that's a good thing for the AI industry because ultimately it doesn't really matter how powerful these models become if the rest of society doesn't actually want this future Silicon Valley is trying to build. You still need communities willing to accept the data centers, governments willing to approve them, investors willing to keep funding the buildout, and ultimately billions of regular people willing to actually adopt the technology. And so if this AI boom were to stall or the bubble eventually were to pop, Gavin's argument suggests the biggest threat wouldn't be the technology itself failing to live up to its potential. It might simply be people deciding that the cost of getting there isn't worth it. At the end of the day, AI doesn't exist in a vacuum. No matter how quickly the technology advances, even with RSI on the horizon, it still has to exist inside our economies, inside our governments, inside our society. And if the gap between the people building this future and the people expected to live in it keeps getting wider, that could become a bottleneck itself. And I don't think many people are really talking about that. So yeah, let me know what you guys think down below. Do you think Silicon Valley is seeing where all this is headed before everyone else, or are they the ones getting ahead of themselves?
26:06 And as always, thanks for watching. If you enjoyed the video, please consider hitting that like and subscribe button.