De-Lux Tech Talk with Josh Wolfe of Lux Capital
55:33 Share

To be invested with Betterment is to believe in better. Whether it's saving for today or building wealth for tomorrow, they help people and small businesses put their money to work. They automate to make savings simpler. They optimize to make investing smarter. They build innovative technology backed by financial experts for anyone who's ever said, I think I can do better. So be invested in yourself, be invested in your business, be invested in better with Betterment. Get started at Betterment.com.

All right, welcome to the Risk Reversal Podcast. I'm Dan Nathan. I'm joined with Josh Wolfe. He is the founder, managing director at Lux Capital. Josh, welcome. Good to be with you, man. Well, by the way, I'm here with you. We're doing like a home and home. We are here together. This is the third podcast you and I have done together. Last, I think it was December 23. Prior to that, it was July of 22. And both really interesting periods. If we want to go back a little bit and retouch a little bit,

on some of those issues that we talked about then. I definitely want to hit the macro with you a little bit. And I think a lot of the macro right now is obviously tied up until trade and tariffs and China being a really important part of that. I know you have a lot of thoughts there, not just from an economic standpoint, but obviously a whole host of other issues and them being

a primary adversary of ours. I want to talk a little bit about the VC landscape, some of the things that you're focused on stylistically, some of the investing you're doing in and around what might be a generative AI bubble. It clearly,

In the public markets, some of the air has come out of that, but let's hear a little bit from you. What's going on in the private markets? All right, enough about me. Let's hear about you. As a VC, and I had, really quickly, I had breakfast this morning with a VC from the West Coast, and she's asking me all these questions about the macro, like what's going on. And it's funny because I feel like every conversation you and I have ever had, I see you on CNBC a bunch.

You are so well equipped to just not talk about the private markets, but also what's going on in and around the world. And she's asking me what I'm thinking. And, you know, a lot of folks in your business are heads down as it relates to everything else other than investing. So give me the 411 on how you think, I guess, on the way into 2025 and what your expectations were from this administration and then what you're seeing right now. Well, zoom out, first of all, and I think we've talked about this before, but my general lens on macro is that ignorance of it is no virtue.

You need to know macro. And the analogy I always give is imagine that our world is security selection. You're trying to find the best entrepreneur, the best company, which is really like trying to find the best sector. Maybe it's in a particular city. It's the same thing as being a foodie and trying to pick the best dish online.

having selected the best restaurant, having selected the best neighborhood and the best city, all these sort of sub-micro decisions that all of a sudden, and then you're about to take a delicious bite of this morsel and Godzilla comes and just steps on you, right? So these exogenous big macro forces matter. They matter because nobody in any markets, let alone venture capital, is as smart as people think on the way up

because there is an abundant amount of capital flooding into a sector. The number one determinant of returns isn't the hockey stick that was predicted by some consultant or some equity research report. It is how much money is going into a sector. A ton of money going into the sector. If there's insufficient assets and securities and companies that exist, they get overinflated, multiples.

are higher. That's when we get great exit opportunities and there's a robust IPO market. They're now competing with the buyers, typically an oligopoly industry structure, and those people are paying irrational prices for M&A. So you get a boom. On the flip side, when the cost of capital is rising, which is something that we've been contending for two, three years, maybe a little bit early, was going to happen, even though there was a purported abundance of $300 billion of dry powder. And I think when we talked about this last, I said, no, no, it's not dry.

It's going to be wet. What does it mean, wet powder? It means that a lot of funds who were first-time funds were under-reserved. The amount of money that they put into companies, you have a dedicated closed-end fund. How much are you reserving for your follow-on investments? The short answer, not enough. People were not reserving enough.

And when you had an upmarket where you had seed funds and then series A funds and then growth funds and then the Tigers and the General Catalysts and General Lennox and then these pre-IPO funds, everybody was basically just pricing up each other's deals. And that was a good phenomenon because people had FOMO and there was a bun in cash. All of a sudden...

SoftBank is out of the picture. Tiger is sort of out of the picture. Andreessen Horowitz is maybe not as desirable like when they were five years ago in a scarce amount of capital. Some of those big funds in the venture ecosystem are now more focused at gathering and getting really big, by which I mean $80 to $100 billion plus AUM. We're five and a half.

But there are funds that really want to go public. And to be public, you have to have that scale. They will follow the path that Apollo, KKR, TPG, Carlyle all did post last crisis 09 to sort of 12 or 13 when they all went public and created permanent capital. So you have what I've called the megas, which are these big venture capital firms that are getting bigger with the expectation that in the next one or two years, some of these guys will be publicly traded companies.

Then you have the minnows, which are the small tail of subscale funds that were depending on this chain of pricing follow-on rounds that are now screwed because they don't have sufficient follow-on capital. The kindness of these benevolent strangers are not there to price up the rounds. And so you're going to get a hollowing out of those sort of two things and the desirability of founders to pursue them for investors.

So that's the backdrop that I see. Low cost of capital, abundant, great venture time, the 2020, 2021 craziness, the SPAC booms, all that kind of stuff. Now, rising cost of capital, even though there's this purported $300 billion, it's not really there. The natural cost of capital, regardless of what the Fed does, regardless of even what the bond market does.

It's just getting higher. And that leads to some new themes and theses. We'll talk about what you do in a rising cost of capital environment where there's going to be interest in capital. So that's sort of the backdrop of rising rates, rising cost of capital is my view for the next few years. It doesn't matter about the next three months or six months or whether the 10-year is 4.5 or 4.8 or 5.2 or 3.9. It doesn't matter. The cost of capital in our world of venture capital is rising.

What do you make of the notion, like in a period of volatility that we've seen on just the policy front, right? The public markets are mark to market every day, right? And you go through like some really irrational periods. And I think we could kind of say that a lot of 2024 was a bit irrational due to the concentration in the stock market. A handful of names that were all benefiting from really one theme. And it was probably...

really easy in hindsight to now look back and say, we were going to see the sort of deceleration and the sort of rationalization of multiples and the like, especially if you're in this period higher for longer. So how do you think about that in venture? I think that you probably went back in 2022, when we're thinking about it, rates had just gone up pretty precipitously. Inflation was at its high. I think it was like 9% on a CPI. A lot of things changed in the back half of the year.

then obviously chat gpt comes out and you have this two-year monster run in the s p in the nasdaq what is the lag in venture and what do you expect to happen because again i think the policy right now is clear as mud but it's not likely to be impacting i would say

private companies just yet. Does that make some sense? It does. Remember, in a sense, venture is not a lagging indicator on technology. It's, of course, a leading one. ChatGPT and OpenAI, of course, started as a nonprofit and has a complicated structure, which itself presents all kinds of existential risk. Because remember, not only do they have to convert into like a traditional sort of C-corp or B-corp or whatever it might be from this nonprofit status, but they have, and I don't know how to underwrite the existential risk for them, they have an antagonist.

that really does not like Sam, which is Elon. And if Elon were not powerful, you wouldn't really care. But he's, I don't know, maybe the second, third, 10th most powerful person in the world right now. That is not somebody that you want because you're thinking about raising more money and now you're raising money from SoftBank, maybe a good signal, maybe not a great signal. So anyway, that's an OpenAI specific thing. But OpenAI was a private company before it potentially becomes a

public company or the revelation of chat GPT which of course was invented at Google through the transformer technology that they basically really turned into a business or anthropic or any other number of companies in our portfolio from hugging face which is open source or runway ml which is video those companies when we were funding them were all sort of sub 50 million dollar valuations

And today they're four, five, six, $8 billion. And whether they double from there or stay there or go down, I have no idea. But to make an incremental investment today in AI in the private markets,

when it is a widely consensus bet is very hard unless you're backing a team where you have a unfair advantage. We're taking actually one of the early hugging face employees starting a new company where the first check in at a very fair price

And that's only because we had this privileged relationship. That's where you can get arbitrage. Not because you have better information than me, not because you've done better analysis, but because you have an unfair advantage in the trust of an individual who you backed and you were there, as we like to say, we believe before others understand. So on the private markets, that's where you can get edge. Venture will get hit hard because you still have too many companies that are being comped

to a leader which deserves the valuation, and then a long tail of like copycats. So in AI, this is clear. In defense, it's clear. We were founding investors in Andral. Andral was a company that people thought was ridiculous. And Palmer Luckey with his Hawaiian shirts, they thought was ridiculous. And it's one of those classic Gandhi things like first they laugh at you, then they, and then you win. He has won and the old defense guards want to join him. And the secondary demand for

for this company in the private markets today is off the charts. I mean, it is like something I've never seen. Even this most recent round, which $2.5 billion at a $28 billion valuation. And again, here we invested sub $100 million first round.

eight and a half billion dollars of demand for two and a half billion dollars, including us, by the way. We're doing more, but we want to do even more because we believe in the company. They'll be sort of like, I don't know, a billion and a half, two billion dollars end of this year. Maybe they double over the next year, year and a half, maybe another 50% to 100% doubling. And you're looking at five billion top line between organic and acquisition. Put it 10 times multiple on that, you're a $50 billion publicly traded company with margins that are

6 to 10x better than these cost plus contractors that are defense primes that have been taken off, let alone the hype and excitement around the sort of carp led values driven Palantir ascension. So venture will eventually get hit because there are

There's 50 companies that say we're going to be the Andoril of X, but there won't be. Andoril is going to be the Andoril of Andoril. All right. So that's such a great point. I was going to get to this a little bit later, but I think let's just talk about Palantir for a second. In the public markets, this is a $225 billion market cap. They're expected. Maybe they'll do $4 billion in sales this year, maybe $5 billion next year. So you don't even have a growth rate that looks anything. And you just mentioned Andoril is going to do $1.25 billion this year. Close to $2 billion. Yeah, $2 billion.

How does that discrepancy from a valuation standpoint exist in the private markets to the public markets? Because we've seen many instances where the private markets are really the ones with the fat valuations relative to the public markets. Because that makes no sense to me. Because if that company went public, and by the way, at $28 billion, that would be one of the biggest IPOs that we've seen in a very long time. That stock right out of the gate would double.

Like in the public market. Totally. Look, Buffett is not buying these stocks. This is not a value audience that is buying these stocks. It is in part a growth audience. It is in part a narrative-driven audience and a momentum-driven audience. If you think about Palantir, I actually thought back in the day Palantir, when they were trading about a $15 billion private market valuation, was probably going to sell to IBM for $20 billion. And man, was I wrong.

IBM at the time was like $150 billion market cap, I think. And so I thought, sure, they'll give up 10% of their market gap to buy a private company, which has much younger talent, much smarter people, but it's still sort of a mix of cutting edge user interface, consulting in some cases with forward deployed engineers. But I was not a believer that this was going to be like a 50 or $100 billion public trade. Man, was I wrong.

I mean, leadership, execution, like I'm a friend of Carp, I'm a believer. What I think has caught the market about Palantir is a narrative-driven approach

rise in part with defense. And when you look at the pure plays in defense, it's these old sclerotic bureaucratic businesses. And you're like, what's the catalyst there? And then when you hear CARP, there's also a sort of political market premium that's embedded here, which is he is speaking very confidently and very openly about, and he's, I think, generally like left, center left, most of his voting career, even though Thiel, co-founder of Palantir on his board before, antithetical views.

But he is out there leading the charge saying like enough is enough, whether it's on campus stuff, whether it's anti-American sentiment. And I think people are like, yes, I believe in this. And I know people that are on the left, people on the right. And so I think they're getting behind him and the rhetoric and the narrative and the story or the philosophy and the truth that he's telling and are like, I want to be along that. So some aspect of Palantir's ascension is literally the political philosophy or the moral philosophy of CARP that is embodied in this company.

Yeah, that could be your view. And I don't mean to come into your office here. I think it's a bunch of bullshit. But why are people buying it? Because we're in a bubble. I mean, like we're in a bubble. But what causes demand for anything? You know, the difference between...

half a million dollar car and a five, you know, it isn't necessarily the quality, right? It's the perception. And there's something about people's willingness to pay a disproportionately premium, which for, like I said, for any value investor makes no sense. But I believe that there is some portion here where people are voting their sort of moral consciousness. And yes, you could say that that's bullshit. That's not part of a fundamental analysis of a stock. And it's not, it's not trading on fundamentals. It's trading on

you know, vibes. It's trading on narrative. But there is gravity in markets. And, you know, like to me, like this will get cut in half and, you know, people will be very cynical. I mean, I get that. It's not even a pure play on defense, you know, and some of their ex-defense businesses sound less than interesting. You know what I mean? Like, listen, it's- But would you be short Palantir? No, but you know what I would do?

I would be long and drill and short Palantir. That's probably the pairs trade of a century. I don't know about the short Palantir. And I got to tell you, I'm involved with a whole bunch of short sellers on companies that I know are like true frauds and fads. But look, I mean, you've known me talking about Elon and Tesla. Right. We're going to get to that. I could be right about warranty reserves. And did he get away with accounting fraud? It doesn't matter anymore. It is a religion. And

And no amount of reason could appeal to the affinity that people had to be part of this movement.

Like, I'm an atheist. I don't believe in God. But I can't convince millions of people who are pure Jews, because I'm a Jewish atheist, or Catholics, or Mormons, one of the fastest growing religion, or Muslims, that what they believe is not rational. They literally believe it. And- You ever see that Ricky Gervais and Stephen Colbert on Atheist? I mean, Ricky Gervais is like a very outspoken atheist. And Colbert is a- And is very Catholic. Have you not seen the clip? No. Oh, it's so good. I'm going to send it to you. By the way, I'm-

non-practicing Jew atheist too. So you and I should, we should break bread on the week of Pesach. No, I mean, I get all that. I mean, you and I could sit here and argue about that. But these things are religions. They are. Yeah. But the really, I mean, listen, the fever broke in Tesla. It broke. This is one of the worst performing mega cap tech stocks in the last five years. And so, you know, the religion is a belief in full self-driving to robo-taxi to optimist. Like

If you do not believe in any of that, then this is a $100 billion market cap company. The one thing that Thiel and others have said is never bet against Elon, and I got to say.

I still do not believe in the pantheon of great tech entrepreneurs that Elon is a better entrepreneur than Jeff Bezos. Because I think Jeff is more honest. I think he's an extraordinary capital allocator. I think what he's done to compound Amazon, what he's going to do with Blue Origin. Like, I'm just much more in that camp. I want my kids to emulate Jeff more than I would want them to emulate Elon.

But I have accepted that it is a fool's errand to bet against this guy because he will figure out. Like I knew that the XXAI merger was going to happen, right? And then it did. I don't know if all these things get rolled in. I don't know what shell games get played. But betting against him and his savviness or cleverness or whatever it is, it's just really hard. And it's not that Tesla is worth a fraction of what it is. But—

I don't believe that we can effectively, through reason, persuade the millions or tens of millions of retail investors, the Bailey Giffords and other true believers, the Barron Funds, like that –

We're once like these great value investors that just have ridden this and been rewarded for riding it. And I don't see what changes that. Well, it's funny. I mean, I suspect that Tesla will try to buy XAI and they'll make a good argument about it, the infrastructure that they've built. And if you think about full self-driving, they've gone a different way than LIDAR and it's going to be very dependent on AI. And you could make the same argument about Optimus and about, you know, the other business like

- And he'll throw, look, if I had to sit in a room with him, what else could we do? We got robots, we could do life extension. We're gonna cure with neural link, purportedly, autism and blinding. And any real scientist will look at this and be like, oh my God, he's like literally monorail man on The Simpsons, like, you know, hucking his wares. And then people say, well, look at the rockets and look at the cars and like these things are real. And it's a really hard argument. I have accepted that I will still hold these beliefs

But I will not put capital against him. You could make an argument, though, that there's probably 50 amazing entrepreneurs and builders across his suite of companies that would do amazing things because there's not enough time in the day for him to really oversee all this stuff. He said something really interesting on their call in January because he was being asked about how he allocates his time. And he gets that question a lot. But he was like, I go to where the problems are.

And I think that was one of the smartest things, like I've heard him say in a long time. I think he's made some huge miscalculations about the price war that he's been in with the EV, the market, and specifically with China. BYD has just overtaken them in deliveries. They're fully electric or battery part. By the way, arguably, one of the greatest venture investments ever made was not made by a venture capitalist. It was made by Warren Buffett. Yeah, and BYD. Because that investment is just extraordinary. I remember I was at the Berkshire Annual Meeting yesterday,

I don't remember if it was 06, 07, 08, 09, but it was before the crisis. And I got up and that was my question. I asked Buffett and Munger, you're value investors. I'm a venture investor. What do you do in investing in a Chinese electric car company? And they gave a great answer. And the answer was basically like, if you go and visit this guy and you see his operating principles, this is like when he was buying his car and manufacture. This is one of the most insane operations we've ever seen.

Were you nervous asking Buffett a question? No. I mean, it was in front of 70,000 people. I know. That's what I mean. Were you nervous? I mean, that's a pretty big audience there, right? I was less worried about asking him a question. I was worried about the blur of a crowd that was going to boo me or something. But I thought it was a reasonable question. Nobody was going to ask about me.

When you were, I know we're getting off topic, when you were in school, when you were starting out, I'm assuming not as a VC, but like in finance, were you a banker to start out? I graduated Cornell in 99. I'm 47 years old. I went into investment banking. I wasn't even smart enough to collect my first year bonus. But around that time, really, I would say the end of college, maybe the first one or two years, I always wanted to do venture investing.

I had no money. I came from nothing. My mom was a public school teacher, single mom in Coney Island. I didn't really understand the stock market. When I started to get internships through college, working in sales and trading, investment banking, like Muni Finance, and by the way, I just want to share this actually. - Muni Finance didn't do it for you? - No. - I'm kidding.

But I got to tell you, the best advice I ever got was in that job. I was working at Prudential Securities in the muni finance department. I would work a little bit between the public bankers who were basically like financing bridges or hovercrafts or whatever cities needed, you know, for government obligation bonds. And I was working on the trading desk of these trading bonds. And I remember this one guy that summer.

He used to call me pencil neck, which today you get fired for whatever, but like used to walk by and be like, gentlemen, pencil neck. You know, he just like, he was like, do you know why you're here? And I'm like, yeah, I'm here to learn, you know? And he's like, no, no, no. You are here to make my job easier. And I got to tell you, it is the most important advice that anybody that might be listening. It is the advice I give my kids.

For 20 years or whatever it is, you're in K through 12 and then you go to college and you're paying and all these people there just technically serve you even though you're performing to a task and writing your essays and getting 100 on your test and getting A's or whatever. And you think the world is there to serve you. And then you go into the real world. You're like, I'm just here to learn. I can't tell you how many people send me emails.

And they're like, I would love to pick your brain. I would love to. No, the people that get through here are the people that are like, I read this report. I heard you on a podcast. I don't know if you know about this company, but I wanted to send. They're doing something. Try to add value. Yeah, exactly. Most important lesson that a young person can have as they're starting their career. You are not there to learn. You are there to help the person that's giving you the opportunity.

All right. Do you think this is as good of advice? So when I was at school, I was at Penn in Philadelphia and I was just really interested in markets. And I got a job working for a guy named Dick Welsh, who was former Marine, did two tours in Vietnam. And he was, you know, a stockbroker there and old school dude. And he used to come out every day for lunch. And I, every day I was like,

Is he going to remember my name or not? He used to come out with a $20 bill and he'd say, Nathan, will you go down to that place and get me that sandwich I like? And he goes, I hate to ask you. I have the money, but I don't have the time. This is what he used to do. And I was like, that's kind of an interesting philosophy. And what was more important to him was to service his clients. And it was a great experience for me at that age. And it was Prudential Securities also.

- Amazing. - Which is gone. - That was the expectation. The expectation was you were the grunt that we're going to get coffee and do the errands, whatever. And by the way, my internships were never paid. Like I didn't have any money, but it was the opportunity that they were giving me to get my foot in the door. Then I remember like, I don't know, it was 10, 15 years ago. They're like, no internships have to be paid. There's something self-selecting. I understand like people, they shouldn't be like slave labor.

But there's something self-selecting about somebody that's like, I want this so badly. Literally, you don't have to pay. All right. We're going to get back to markets. But it's interesting because we're on this sort of mentorship. I know that you were friendly or maybe it was a mentor of yours, Daniel Kahneman. Yeah. And I know that you're tight with Michael Muebissen, who I love and a great dude. How did those relationships come about and how important were they? Because what both of them do is very different than investing. And I know Michael's

much closer to investing. But there's a lot of psychology involved in their work that I'm sure has been very useful to you. - It's funny because I also tell people today, which is one of the most frequent questions I get from parents or friends or allocators, LPs, what should my, you're on the frontier of technology, Josh, what should my kid be studying? And I always say the one timeless thing truly, which I didn't appreciate when I was younger, is psychology.

is understanding human psychology because we are the same people we were on the place to see. We are the same people with the vanities and the greed and the jealousies and the affinities for people and the desires and the attractions and the alliance forming and the undermining and the cunning and the I know that you know that I know like how many layers deep. You have to understand all of that. Deception, detection of deception, that is not going to change. Human technology can...

and amplify that. You know, I mean, radio can amplify truth-telling or can amplify Mussolini and TV can change the shape of a presidential election and TikTok all the same. The timeless through line, no matter what changes, is human nature. And so I think that that is the most important thing to study and you can study it by reading

cognitive psychology and evolutionary biology, and you can study it by reading fiction and Shakespeare and watching, as I did, copious amounts of TV and character studies and development. You can learn a lot from Mad Men or Deadwood or The Sopranos. Those are all mini Shakespearean dramas told with different characters and different sets and different costumes, but it is the same through line. I mean, you watch...

you know, Hill Street Blues or NYPD Blue or some modern thing today. It's all the same stuff, just different characters, different settings. And every business that we invest in is effectively the same thing. It is some entrepreneur. Hopefully there isn't a coup of the number two, three, or four that work for them. They've got ambitions. I always say that chips on shoulders put chips in pockets. I want to know their personal narrative and their story, what is really driving them when somebody comes in and says, I just want to change the world. Bullshit. No, you don't. That's something that you heard on some

TED podcast or TED talk or whatever. Like, no, I want to know that I was a poor kid in a school filled with rich kids and I resented them because I was smarter than them and I was going to prove them wrong or I was the gay kid or I was the only black kid in a white neighbor, whatever it is. Like, I've got a chip on my shoulder and I've got something to prove. Those are the best entrepreneurs ever. Timeless human story. And by the way, on Kahneman and Mobison, just to answer your question, I met Danny through Michael.

And I met Michael through a shared interest in books and ideas. We were introduced by a common friend that he knew from Connecticut probably 22 years ago. I think I met Danny 12 or 13, and we met at a Santa Fe Institute dinner that Michael had hosted. And Michael was at...

Credit Suisse, where I first met him. And then he went to Legg Mason. He sort of builds like his, like Charlie Munger to Bill Miller's Buffett. And then ultimately to Morgan Stanley. And he joined a group there that are super thoughtful with some great friends that really appreciate this craft of...

the social sciences as they inform the quantitative investing. And then I became friends with Danny and we would do almost like these Tuesday with Maurice. And then I started pulling together a lunch and I would have me and Michael Mofson and Annie Duke and Danny, and we would just talk about risk and probability. People would have paid like 100 G's to sit in the podcast series, actually, where we talked about risk and probability. It was on the eve of Danny publishing Noise, which was his ultimate book, a final book.

And Michael had just published another book and Annie had just published the book on taking bets. And yeah, it was a lot of fun. By the way, you sound like a quant. I mean, like the way you think about markets and risk and all the like here. So interesting stuff. And I'm sure you would say, go read all of Danny's books. And Daniel, excuse me, I don't know him like you do. And Michael, he wrote one a couple of years ago. And by the way, on the quantitative stuff,

I had a really big chip on my shoulder. My parents split when I was young. My parents had a custody battle when I was eight years old. I went to live with my father in California and I was on these gifted and talented classes in New York, in California, two classes per grade room. It was like the inverse. I mean, it was crazy. When I came back, all my friends in fifth grade knew basic probability and I didn't. I felt so deeply insecure. I became obsessed with it.

So that to me was like one of these like silver linings that made me obsessed with low probability, high magnitude events and thinking and odds and understanding Bayes theorem. I didn't understand any of that stuff when I was in fifth grade and I became obsessed with it because I was disadvantaged. Imagining a better future is the first step. Investing that future with Betterment Advisor Solutions is the next. Whether you're launching your own practice, looking to streamline client onboarding,

or just searching for efficient ways to scale your firm, Betterment is here to help. They automate to make tax optimization simpler. They provide support to make administrative tasks easier. At Betterment, they're building innovative technology all for anyone who's ever said, I think I can do better. So grow your RIA your way with Betterment Advisor Solutions. Learn more at betterment.com slash advisors. Investing involves risk,

Performance not guaranteed. Let's go back to something you just mentioned, Hugging Face. And so this is a company that I think there's no shortage of folks in venture who wish they invested when you did, I suspect, in your portfolio. And I was looking at you probably have at least a couple dozen investments in generative AI. And I go back to our conversation in July of 2022. This was something you were talking about then before it became the sort of flavor of the month. So I want to go back to that in a second. But since then and since then,

GPT's release in late 2022, there really has been maybe 15 publicly traded companies that have accrued most of the value based on some of those early investments. Obviously, NVIDIA is up there at the top, but then there was Microsoft. It was Amazon, AWS, Meta.

and Google, now Google has been a bit of a problem here over the last couple years. So my question to you is like, you see all these amazing founders, you see the work that they are doing, they are trying to A, create some new things, but obviously a big part of their goal is to disrupt some existing sort of players.

What do you make of the innovation or the lack of innovation by these multi-trillion dollar market cap companies in the public markets and how they've set their sights on developing models and then how they're deploying them across their clouds and how they interact with their customers and what they expect their customers to use them for so they get some sort of ROI and they get them coming back? Okay, great questions. And we'll spend a little bit of time on AI because there's so many different facets, so many different layers.

In 2016, by the way, I was in a company called Zoox, which was a self-driving car company competing before Tesla had their robo-taxi thing. We would end up selling that for Amazon for a billion, two billion, three. And it became the basis for Amazon self-driving trucks that eventually we will see right-hand turn lanes going 24-7 delivering stuff with a human coming out and just dropping the stuff off.

In that company, we had a bunch of people that were sitting around. We had $25 million invested very early. And I'm like, what the heck are these guys doing? And it turned out that they were training the cars, but not by training them by driving them, but by training them in silico, what looked like they were playing Grand Theft Auto Call of Duty. And I'm like, what are those guys over there? They said, oh, they're these NVIDIA engineers. We have chips that aren't even on the market yet. We're repurposing from gaming into cutting edge software.

simulation. And I was like, this is the soul of the new machine. 2016, I go to this investment conference in Chicago, sort of like the Soam conference here called Invest for Kids. And I'm like, to your point earlier, this is the pear trade of a century.

Short NVIDIA at the time was like $150 billion market cap. NVIDIA was short Intel, $150 billion. Long NVIDIA was $15 billion market cap. To this day, it is the best performing thing that conference has ever seen. And to my LPs and people that listen to me on this crazy speculation that this was going to become the sole of the new machine, it paid off really well. So that was NVIDIA. Yes, they have been the arms dealer. And Jensen would like everybody to believe that everybody needs the latest cutting edge chip.

And those chips have increased in size, in clusters, in price, in clusters. You do not need, and there's this distinction between training where you definitively do need NVIDIA and inference where increasingly there's going to be different architectures that you don't require NVIDIA chips for. I have been quite bullish on the memory makers. This is, I think, unappreciated as much.

SK Hynix in particular, if you look at the big three, you've got SK, you've got Micron, you've got Samsung. Samsung is besought by what I think are like the Intel woes, bureaucratic, slow moving, big corporate. Just Micron is going to be troubled by the fact that they're going to have sanctions for China, et cetera. SK Korean will be able to navigate around that, I think, more effectively. They are coupled with a high bandwidth memory as their main product for the things that are coupled to the NVIDIA.

GPUs, but I believe that there will be a shift to on-device inference where your iPhone is doing 50% of your AI calls on-device, not going to a cloud, not going to a CoreWeave cloud or an AWS cloud or Azure cloud or GCP cloud or an NVIDIA cloud.

It will be on device where it has effectively taken your cached emails, chats, WhatsApp, photos, images, whatever, and are doing local processing. And I think that the memory basis for that is going to be really important. There was a paper that Apple published a year and change ago.

which gave me this clue, which was small models on device leveraging memory. But it's not old memory. It's like next-gen memory. And I think that people are going to be surprised in the same way they thought that GPUs were commodity things that were for PS4 and Xbox that suddenly were like, oh my God, this is the soul of the new machine for AI, that people may feel that same way around memory. So just full disclosure, I'm personally long SK Hynix. We as a fund don't do that. But in our PA, in our family, we bought SK Hynix.

So that's the first layer, which is the hardware layer. The cloud players today, to me, are benefiting from the massive infrastructure spend where what any one individual company does rationally, collectively they do it rationally. Microsoft's building, Amazon building, Google building, Salesforce, Oracle, all these people that are building clouds to host on AI.

The beneficiary of that has been Nvidia and to some extent the memory players, but eventually that is overbuilt and you are already starting to see signs. Microsoft pulling back from project after project. The indications that they were giving to building all these data centers is just scaling back.

This is history. In 2000, you had this with fiber optic cables. It was overbuilt, and then the beneficiaries was the third world, the people who were connected on laying down all this dark fiber that got lit. In 06, 07, 08, you had the first wave of being on-prem, the data centers, where it was either there to being co-located to the cloud, that transition. And eventually, those guys, like the Equinixes or whatever, really struggled. Private equity would come in post-overbuilt, so I'm not super optimistic about

about the core weaves and some of these like neocloud players that are really weird capital structures, a lot of debt, very high concentration with customers, in particular one or two of these things.

I feel like Corweave is going to be like the global crossing of this sort of thing. There's so much hair on that. I think that's going to be a great quote in the future. I really do. So that's one thing. The other thing are the foundation models. Now, the foundation models slash research labs, right, where the effectively open AI, again, we already talked about how complicated that structure is combined with the existential risk of being in Elon's sites, which is not a place that I would want to be. And then you have like anthropic. I think personally…

Anthropic ends up getting bought by Amazon. Google has a stake. Amazon owns roughly 20%. Anthropic really dependent on Amazon compute. A lot of that deal has to do with their ability to do their titanium chips and develop on AWS cloud or tranium rather. And I don't know that that ends up being a standalone company, either of those guys.

Microsoft had effectively controlled, and if you listen to Satya, what he says about OpenAI, now with Masa and SoftBank, I just feel like that's going to end up being a really difficult thing for investors to get out of $300 billion plus valuation today. We, as consumers, have benefited from this. Now, I'll give you the inside sort of riff, which you may have heard even through the BG Squared podcast between Gurley and Gertzner, who are both friends.

Gertner believes, and he's said this publicly, that OpenAI has runaway success. They're the top app in the app store, 20, 30 bucks a month. It's sort of like Gurley and I have debated like, is this more of like MySpace? Are they the first player where it's like, wow, this is amazing. But you mentioned before Google has been struggling.

What happens when BARD, which was a joke, today, Gemini 2.5, is the leader on all of the benchmarking that people do on these kinds of things for both code, not yet image generation, they have video, but...

what happens when they release that for free? I mean, you think back to like Lotus Notes and Microsoft Outlook and how this was all packaged and all of a sudden Gmail comes around and it's free. - Well, they're gonna have to, right? Because they're gonna have to defend that search mode. - Totally. - Right, because again, I don't know about you, if OpenAI or Perplexity for that matter came out with a browser,

I'm dumping Chrome. Perplexity is. They're going to come out. I think it's called Comet. But when I dump Chrome, that means I'm not doing Google searches anymore at all. You know what I mean? So again, and they have a better UI as far as I'm concerned, like perplexity. I don't know if you're watching, but like every day, I don't know if it knows, but that I search for stock tickers. I have stock tickers coming across, you know what I mean, right above the search box and a whole host of other things. So they seem to be innovating. Now, again, this is a wrap.

They haven't built – I know they're working on their own model, but why bother as long as they get Claude as they get a handful of the other models? And by the way, they all have my money. I pay for Claude, the 3.7. I pay for Perplexity Pro. I pay for OpenAI, 4.0 and 0.1 Pro. I don't pay for Grok, but if – or maybe I do pay. You probably pay for, what, Twitter Plus or something like that. But I would. Okay, so this gets to the next thing, which is who's going to be advantaged in this?

I don't believe that the foundation models that have raised tens of billions of dollars and have spent all this money on compute, mostly which has gone to the cloud players and or the chip players are the ones that win and capture value in the end. Consumers win because it's democratizing our ability to do everything from code to write essays to conjure images and all that kind of stuff. I think that the people who win are the people that have longitudinal deep silos of proprietary information. Now that sounds like a shitload of buzzwords, but like,

Big data was the thing that people talked about for 10 years. It didn't really amount to much. I mean, you had some, you know, Snowflake and Datadog and a bunch of other people that's capitalized on that. But by and large, it is the single owned asset. Think about this. Meta. Meta has a longitudinal, meaning long time, siloed proprietary asset. Apple cannot search my WhatsApp chats.

but they have all my WhatsApp chats. I can search through that interface. They have all my Instagram posts. They have all my likes. I'm not on Facebook, but effectively the same thing. Twitter, X, XAI. Elon has everything.

every tweet that I've ever posted, everything I've ever liked, everything I've ever retweeted. The repository of proprietary information in that silo is really valuable. Bloomberg with their longitudinal repository, including all the acquisitions that they've made with normalized data and M&A and MergerStat and all this kind of stuff, super valuable. Pharma companies that have their unpublished, unreported clinical trial results. Anybody that has a long internal database

that goes back in time and is really deep, is going to be able to apply the most cutting edge and best near performative, meaning nearly as good, open source models. And so I think open source plus Google that ends up with some sort of enterprise trusted domain where people are comfortable putting their stuff in and not worried that either the models are going to be training on it or it's going to leak out into somebody else's search are going to win.

So that's where I think value started with the chips, then the clouds, then at least on equity value accrued to the foundation models because it created this, oh my God, this is amazing. I have this magical capability. That is going to get eroded, I believe, by free from either open source or Google or somebody else that comes along with a different business model. And then ultimately the value will accrue to the people that are using these models that are effectively, I don't want to call them commodities, but are effectively free on data that is impossible for somebody else to get.

All right. So that actually speaks to Apple. If they had a strategy as it relates to Intel, Apple intelligence, it's on device. It's their whole privacy focus. They have a ton of data. And if they're able to keep it on device and you own that, it will create a better experience. They have one and a half billion iOS devices globally. It seems like

if they could ever get their act together, this would be kind of one of the sleeper plays as far as deployment and being able to monetize it among their users. Do you agree with that? I do. I don't know if they will be able to do it. I've got close technical friends that are like Apple has made wrong bet after wrong bet in this domain from Siri, which sucks to iOS is great. I still find it to be magical. What's interesting is the very direction that they took on privacy is the very thing that prevents me from doing a search on my WhatsApp chats on the Apple device.

And so one interesting theory is that the really important thing about AI is in the tech infrastructure layer between how all of our apps communicate and ultimately how our devices communicate. And this to me is really going to be the two big battlegrounds. APIs, which is what you need to do to be able to call for an Uber and be able to connect to your payment system, all the things that are sort of the back-end plumbing infrastructure.

Because there was a company, I'm not going to name names, that was working on hardware to be able to bypass the need for accessing an iPhone or an Android phone.

that wants to monetize your attention, but is thinking about new hardware ways to do that. Imagine glasses and wristbands and these kinds of things, okay? They're trying to figure out how do we get APIs to connect to Uber? And somebody was like, why do we have to negotiate these deals and figure out the software and the code for this? Can't we just pretend to be a user and let effectively AI act like you may have seen in operator or computer control, respectively in OpenAI or Cloud, and let it pretend to be a human and just get us an Uber? Mm-hmm.

And so I think that there will be this weird layer where instead of it being behind the scenes code and API negotiations, it will be AI acting like an agent to connect all these services on the top. And you won't know, is it a human or not?

It's actually funny. I had a moment when I was testing operator early on for an open AI and it went through some website as it was searching for something and it got caught because it had to, and you can take control of it, sort of like a steering wheel in like a self-driving car or something. And it said,

Can you confirm you're not a bot? And it couldn't do it. So I had to actually step in and type like the little code to say I was a human. But it'll get there. Do you take the over on that? If I said to you, yes, that's going to be like a behavior that consumers or enterprise, business, that's going to be a big angle, right? The agents. Are you taking the over on that? If I said it was two years when this is going to be something that is widely deployed.

I think that this MCP protocol that is sort of like the next version of APIs where services are going to be connected through this, I do believe that you will have agent-to-agent communication for many things that today you're picking up the phone and calling a customer service agent. Low-level stuff first. You're calling a travel agent. You're talking to somebody at Delta. You're calling the bank. All of that will be just agent-to-agent.

You will literally tell your agent, these are the five things I need to do. It will say, this is my plan, just like you can see today as it goes through reasoning and is like trying to parse your query and come up with like the best search so that it can go and access the information, which is very impressive that many of these groups are transparently showing you the chain of thought when they're doing that. It will do the exact same thing and say, so I just want to confirm what you're trying to solve is getting this ticket refunded. You're like, yes, boom. It'll figure out the way to do that.

And it will do it by talking to probably 80% other agents that are on the other side because that's good for customer service for a company that doesn't have to employ a person to do that for you. And so you will have agent to agent handling all this low-level stuff. It's going to be so frustrating for a while. I mean, which is that frustration will be the thing. It's why Alexa never took off. It's why Siri never took off. Because there will be a ton of people competing to make the best product.

protocols, agent to agent things. There'll be adoption of this. But to me, we will look back and be like, in the way my kids were like, you used to have to quarter, you used to have to carry a quarter into New York City from Brooklyn to make a phone call. But yeah, and you had to wait online to like get it, you know, like it's just insane. I think there's a thing. I have 21 and 19 year old daughters. I don't even think that they know that we had to do that. By the way, you've ever seen the clip, by the way, of,

- A rotary phone presented to like a Gen Z kid and they're like, I have no idea, like what do you do? Like to turn it, it's crazy. So I think that in the same way, there will be two wow moments for younger generation. One is gonna be, wait, you used to have to literally call and go through a phone tree, press one for this and sit on hold. Like, are you kidding me? You're like, oh yeah, no, we used to have to do that. And your mother would yell at me because I was losing my temper or whatever, you know. Now, the second thing is gonna be, wait, you used to have to type on a device

and actually like put inputs because like our investment in control labs, which was a brain machine interface, non-invasive, where you could just gesture in free space, you will just gesticulate and gesture to your computers. And it will feel like you're directing it, orchestrating it without having to know any code. Even the people that are coding today,

They're not actually, I mean, there are people, of course, that are elite coders that are coding, but some of the best coders are like, I'm vibe coding now. I tell it to make this app. And then I'm like, no, no, change the menu bar to red. Like, that's insane. That's magic. Right. No code. This is going to be a big, big deal. No code. Very underrated Pearl Jam album, by the way. I want to go back to NVIDIA for a second, and then I want to just kind of finish up maybe with some of the innovation of some of these large publicly traded companies that we've been talking about a little bit.

So if NVIDIA, and you just mentioned about inference, right, and your belief on memory and how that's going to be on device, every one of their customers, whether it's OpenAI, it's Meta, it's Amazon, it's Google, it's Microsoft, they're probably, you know, 50% of their sales right now. Those sales have been growing, you know, 100%, 100%, now 50%. You know, I mean, it's astounding. This was a company that was doing $20 billion in sales four years ago. They're expected to do $200 billion next year.

But at some point, all of the there's been no competition. So there's been no price competition. Right. And all these folks are kind of really sick of this company having 77 percent gross margins. Right. So if they're all working to develop their own chips with a Broadcom or Marvell or at some point, the NVIDIA story is kind of done. Do you agree with that or not? But again, I would not bet he interests Jensen. And with the market cap and the cash that they have.

What could they buy? What could he reinvent? What could he go into? I don't know. But that only works in a Trump administration as far as from a regulatory standpoint. And I don't believe, I think Trump is going to fuck all these guys, every single one of them who gave the million dollars to the inauguration personally in court. They all lined up on the dais at the inauguration. I think one by one, he's going to screw up. Just waiting to see what happens with this meta FTC. Here's my high conviction take. Yeah.

Nobody knows nothing. We have no idea. I remember going- Do I sound too confident? No, but you're right because you're being rational. I mean, if something can't go on forever, it eventually stops. Was that Stein? Whoever said it. But I remember going into this tariff weekend and Scott Besson is a friend. Love him as a human. I think he's smart and rational. And I think he's a great adult in the room. I remember going into this weekend and close friends, partners,

people I respect were reading everything you could possibly read, like 300 articles. And I had other friends that were reading nothing. And I came to the determination that come Monday morning, they will know the exact same thing. They will be super versed about everything that's going on and have no idea what happens on Monday or Tuesday. They have no idea that there was going to be a 90 day delay on tariffs. And whether this is some great grand bargain that packages the TikTok deal and fentanyl and

our trade deficit. And when it's like, my God, as the Trump supporters would say, he's playing 4D chess or people around him are, or this is the greatest policy blunder where he's effectively saying we got the cards and China's like, yeah, but we make the cards. I have no idea. I think China has so many more cards than I think they were willing to admit. They were 100% ready for this. Yeah. 100% ready for this. We basically put a tax on

on the American Costco consumer in an attempt to basically tax the Chinese factory in Guangzhou I have but all right but that's a question I have for you okay best since a friend yes you have a lot of respect for him adult in the room yes he is being intellectually dishonest because he knows everything that you just said about tariffs the history of tariffs dealing with a country like

China, the levers that they actually have to pull, the readiness to do it. I have a lot of problem with that. I have a lot of problem with like a Tim Cook who was a DEI ESG warrior and then cynically dumps all of it because of a whole host of reasons, maybe fear of retribution and how they can get exclusion in a trade war and that sort of thing. So I think that's so pervasive. And I think that's an issue that this country has to figure out.

before we could actually deal with the EU and deal with Canada and deal with Mexico and deal with China. I'm going to give you the two-handed view, right? They always say that whoever the president was, I want a one-handed economist because he's sick of somebody saying on the one hand this, on the other hand that. I believe Besson in particular is extraordinarily intelligent, extraordinarily smart, very rational, erudite, a true student of markets, a true allocator of capital, understands China extremely well, understands Japan extremely well.

I don't know before he got into this if he understood Trump, but I think he does now. I would rather Scott have to do some of what he's doing because he has to be effective with Trump to affect a long game. That's one view versus, oh, suddenly he's like sold his soul. No, I am much more of the belief that he's going to have to do and say and perform in a certain way.

And I'd rather him do that and be in the seat that he's in to make long-term decisions than for him to dissent and disagree and immediately be decapitated and put, I don't know, Lutnik in there or somebody. So I'd rather have competence in there that has to navigate –

on the field with the game that he's playing. You could have been saying this in 2017 and 18. I mean, that's what everyone was saying, and it ended so badly. So I don't know. Because they worry about the guy who comes in after him. Is Navarro putting in? But I actually think what's more useful is those people to stand up

and talk about things in a manner in which they've spent their whole lives professionally, right, operating within whatever universe, whether it was H.R. McMaster back in the day or, you know, Rex Tillerson or whoever.

And then the other thing, so that's the professional aspect, but what about your morals? What about how, you know what I mean? And I just find this is just a redo of what we went through. I don't know how we got to a political podcast here. No, but we got to it because it's all connected to markets, and the reality is, for better or worse, I've accepted 50% of the country

Trump, maybe more than 50%. I didn't vote for Trump because I can't stand in front of my kids and say, this is a man whose integrity or ideas I wish you'd aspire to. In the same way that I can respect many things that Elon does, including being, I think, the greatest capital market fundraiser in the history of our civilization, truly. Mm-hmm.

But I respect Bezos more. And I tell my kids, like, that's who you should follow. I couldn't vote for Kamala because of social values where I felt like she was pandering to a far left that was really against some of the things that I stand for, particularly around Israel and Judaism and anti-Semitism and letting these things flourish and sort of this. Okay. I voted Bloomberg Romney.

My vote didn't count. I'm in New York. You could say that was cowardly. But I stood up for my principles and I said that these are two people who I think would because they truly were billionaires or couldn't be bought where Trump was sort of a billionaire but maybe a fake one for a while or whatever. And I just felt like these would be – But you did that based on your code, your moral code and your ability to kind of look at all your stakeholders in your business here, right, and then your family. And the fact –

that none of the folks in the administration who know better are doing that. And none of these CEOs...

from Wall Street to Silicon Valley are willing to make those same stance. Now they have stakeholders, they have employees, they have investors, you know, they have partners and all. It just seems like we got to a very dark place in a very short period of time. And I feel like this time around, man, and to quote Slim Charles and the wire, I'm sure you're a big fan. The game's the same. It just got more fierce. And you know what? There is no resistance anywhere. And so that's what makes me very nervous right now. Um,

I do think the game is the same and it is this timeless aspect of human nature. But I think the difference this time is when you look at like what Zuck did, you know, because he was really in the crosshairs. Again, Elon hates him and wants him to go down, right? I mean, even some of the reason why I don't think that they wanted TikTok to be banned was to see that the benefits would accrue to Meta and Instagram and whatnot. I think...

that his swift recalculation, reformulation of his board, putting Dana White. - Dana White, oh my God, let's go back. - But those were all things that were sort of like very, and getting rid of, was it Nick Clegg or his political track? These were all things that arguably,

In the duty of running a company for shareholders, were the right rational thing having identified a risk, which was a politically retributive leader? Yes, it would be amazing if we had a leader that my favorite president of all time is President Jed Bartlett from the West Wing. Guy's amazing. I pine for him. Right. But then we always look back and I remember thinking what an idiot Bush too was or whatever. Man, would you love to have George Bush right now?

Or Romney, McCain, like they were all things I was criticized, but like, I'd pine for that. Or even Obama, people couldn't stand. Anyway, I think that what's different this time is that the speed with which companies got in line

on behalf of their shareholders is interesting. At the same time, I look at what's happening with law firms and them taking knees and people quitting and they're voting their conscience. The universities, on the one hand, I believe that the universities got wildly out of control and they did not have leadership or trustees that were stepping in hard enough to return these places to institutions of civic learning and research and whatnot. On the other hand, I don't like the idea of the government coming in and basically stripping people of billions of dollars and the people that are suffering are scientists. Well, what does it say if the most...

companies the world has ever seen led by some of the wealthiest

people, men, we've ever seen. Law firms, universities. I mean, the list goes on and on. They've all folded. And if you think about that, and it goes back to the lack of resistance, I mean, if those organizations kind of had the sort of front that they had in January of 17 or a good part of 17 and 18, I think we have a different outcome. But there's just no, he doesn't care about norms. He doesn't care about laws, right? What I would say is that those CEOs are realists.

They see the game on the field and they know that if they are not playing the game that is there right now, they are losing. And you can say, well, why do they have to play that game? Then you can go through some whole causative thing. I would say, why are we playing that game? Because Trump beat Kamala on one simple thing.

Whether you agree with him, his principles, his ethics, whatever, he appeared as and was or acted as a stronger leader. Most people do not understand the policy nuances of tax this or this, and she's talking about this, and you could sweep it and say, you know what, they went too far with the pronouns and the trans. But it came down to people wanted a strong leader, and he is a strong leader. So they are realists because—

because people used to say, take him figuratively, not literally. They are taking him literally. They are seeing the retributive justice that will occur if you do not play ball. They're seeing that he has a lot of power. And I don't know, I truly, hand to my heart, don't know in 90 days if you will see 90 deals. It's not going to be with China. China is basically saying, and this is their game theory, we are definitely not backing down.

We are definitely standing up and they're going to have to negotiate. Now, I'll go back to the Besson thing for a second. He broke the Bank of England. One view is he understands the PBOC, the People's Bank of China, probably better than most. And so how weak are they in defending the currency? And so I just would accept as a realist, I have many of the same moral views that you do, which is why I will tell my kids, emulate this person, not that person.

But I also am a realist that I have no idea what's going to happen in the next three or six months. And these things are way more nuanced and complicated than the way that I think that they ought to go versus the way that they will. And as investors, you know, we got to figure out how will they go.

All right. I have no idea. You and I could do this for hours, and I think we should leave it there. I'm sure some of our listeners are probably already checked out because of me, because of me, not because of you. I love these conversations. I really appreciate you coming on and doing them, and I hope we can do more, Josh. Same, man. Good to see you. Thanks.