
Loading summary
A
All right.
B
Josh, didn't you read your own audiobook?
A
Amazing show. Yeah, of course. Can you imagine somebody else reading my book?
B
Didn't your brain feel like mush after doing that, though?
A
I did. I did it. It took me six weeks.
B
Six weeks?
A
Yeah.
B
See, I did it in two days because I.
C
Can y' all switch headphones?
B
Oh, sure.
A
I don't know if you heard any of my audiobook, but I really.
C
Yeah. Why are.
A
I Perform that thing like Mick Jagger. Like, I was like.
D
I'm like.
A
Yelling into the microphone takes a. Like. It's a performance. It takes a lot out of you.
B
Yeah. I was dead afterwards.
A
Mine is so autobiographical. It could never be read by somebody else. It would sound ridiculous. You know what I mean?
B
I actually enjoyed the process, but it was a lot.
A
I hated it.
B
I like how it was done.
A
I like how it came out. Yeah, I enjoyed it after it was done. I like. I happy with the products, but I don't know if I want to do it again. You working on your next book yet or.
B
I need a break.
D
Kind of. Notebooks for you?
B
For myself, not yet, but I kind
C
of want to do one now.
B
But all your white papers, put them all together. That's a book.
C
There you go.
D
Have you spoken to him or not? Supposed to. Haven't reached out to you.
C
I should probably.
D
They must have.
C
Yeah. Yeah.
A
Why do you want to. Why? Devil's Advocate. Why do you want to do a book?
C
Because all the cool kids have one.
A
Okay. Do you think people read books?
C
No, but it's a. It's a cool thing to have done.
D
He just. He. His book got published yesterday. Why are you talking shit?
A
No, I'm not. I've written four books. I'm not talking shit, dude. I'm reading his book right now.
D
Do you think people read books?
A
I don't think people read books anymore. I don't know.
B
So you did a lot of podcasts for yours. The percentage of people. I did not do. The percentage of people who I'm on a podcast with that read my book is probably 10%. I can tell if they read the book or not.
A
Yeah, I don't think anybody reads.
B
And I don't. I don't. I don't, like, fault them for that.
A
I don't even think people read articles. I think people read headlines, and soon they're only going to watch videos about articles.
D
I read so many headlines today, I'm exhausted.
A
I mean, I. Look, I. I love reading. I just. I don't know how many people read. Are reading Books these days? I'm not sure.
B
I've seen surveys buy books.
A
It's not the same thing. Yes, people buy books because they want to read them, they truly want to. But I think our brains have been rewired by algorithms.
B
This is why audiobooks are so great.
A
Yes, I agree.
B
You can have mine in the background.
A
Well, he's an audiobook. Michael's an audiobook.
C
I love it.
D
I've never seen a bigger audiobook either. But you know what? I should listen to your book. What if I listen to your book?
B
Yes. It's like I'm talking to you. Put you to sleep at night.
A
Yeah.
D
I'm going to pause and interrupt you.
B
Put your.
D
Disagree with you.
B
Put your sleeping mask on. And by the way, someone. Michael's got a nice shirt on. Someone last night said, hey, I've been noticing Michael's dressing better. Did you start dressing him?
A
He's so rich now. That's what's going on.
B
Wow.
D
Appreciate that.
A
He is dressing better. He is.
B
That's a compliment to say that I dress you.
A
It's not just that he's dressing better. His taste is improving. It's good. Cause you could have bad taste and buy a lot of expensive clothes that don't look good. He's like put together. Like when you see him at an event.
D
I'm extremely fashionable.
A
No, he is.
B
Who is this guy?
A
I know he is.
D
Who is this guy?
A
This guy. He's so liquid right now.
D
It's a bull market.
A
So liquid.
D
You can't trust nice in a bull market. Right, Kai?
C
That's right.
A
All right, so guys, this is going to be a good one. The market is so horny right now. Full on full. I mean, the market is just.
D
Did you watch dtf?
A
Yes. How It's. I liked it. It's slow, though.
D
It's so good.
A
No, it's a slow burn. I did like it. I didn't.
D
So you got. You got the full on reference?
A
Yeah, I. I just watch it.
B
Weirdest show ever.
D
That. That little smirk of yours makes me feel like you watched it.
A
I like that. It was weird. It's not what I expected. It's definitely not what I expected.
D
They're so. When they're just talking to each other, just signing like just.
A
Yeah. What's that actor's name?
D
The Stranger Things guy?
B
David Harbour.
A
David Harbour.
D
Big Knicks fan.
B
It's good.
A
Yeah.
B
All right.
A
Like that. I watched the Hulk Hogan documentary.
D
Phenomenal this.
A
This week. That's what I mean.
D
That's what.
A
That was a very good One Netflix, four episodes. It was so good because I think because it's like, so my childhood, I was like, I don't know, eight years old for WrestleMania. So, like, that was it for us. That's the only thing we cared about
B
was at the Silver Dome in Detroit.
D
How many guys or how many people in the last 75 years do you think were on the top 10 globally most famous, like, most recognized name list?
A
10.
D
There's only been 10 for the last 75 years.
A
You said how many people are in the top 10?
D
I'm saying it's a. Shut up. It's a small list. It's a list that doesn't turn over a lot.
A
You know what I mean? No, no, no, no. He's one of the most famous athletes who ever lived.
D
People.
A
100% people. Yeah, maybe people. One of the most instantly recognizable people on the globe. Right?
D
So.
A
So there's that part of it also. He had not, like a rise and fall. He had a rise, a fall, a rise, a fall. It's like maybe four or five cycles of everyone loves him, everyone hates him.
D
It was cinema.
A
I mean, it's a really. It's a really great doc, and it was very well done.
B
He was in Rocky iii and he
D
inspired me to get some equipment.
A
So Vince McMahon told him, if you go do Rocky 3, you're fired. Like, he's like, no, no, no. You have to be at an autograph signing somewhere. And Hulk Hogan's like, oh, yeah, brother. Yeah. Let me tell you something, brother. He's like, no, I have to go to la. Sylvester Stallone called me and said I could be in the movie Thunder Lips. Vince. Vince said, if you go, you're fired. Can you imagine? Good luck. Yeah, good luck with that. So I thought it was really well done. Big, big recommendation. So what else are you watching, Kai?
D
We're on a podcast.
B
Yeah.
A
Wait, what else. What else are you watching right now? I need a new one.
D
Playoffs. I'm watching basketball shows. There's a new show on Apple with the guy from Matt Reese. What's it called? Anybody else watching that?
A
Oh, the guy from the Americans.
D
Yeah.
C
Yeah.
A
What is it called?
D
Widow's Bay.
A
What is it about?
D
It's hard to. It's basically they're on a. They're on a Nantucket, like, island off the coast of Massachusetts. And if you leave the island, it's, like, stupid. It's like.
A
It's Hulk Hogan in it. It's a little Shutter Island. It's not. I'm totally out. No, Hulk, no Hulk, no. Hulk. Oh, Margo's got money troubles. Did you try that? I watched the first one. It's. It's. Don't watch it with kids.
D
Oh, I won't. I've been telling this to Ben. I. I've been trying to watch your Friends and neighbors. I'm on episode two. My kids are so Kobe's going to sleep later. He's like, going to sleep like 9, 9:15.
A
Can't watch that with kids.
C
That's not.
D
No, the point is, I'm sleeping by 9:45. I just don't have the bandwidth. I just go straight to sleep. Can't stay up.
A
That's a good one. That's a good one. Did you watch that? Your Friends and Neighbors.
B
That's a good one. Rich people in Connecticut.
A
Yeah. And now they're on season two. That's a good one.
B
John Hammond, Amanda pilot.
D
The reason why I like this show, I was telling Ben, even though these characters are obviously so unrelatable in terms of the way that they live their life, I don't know anybody like that. The characters themselves are relatable just in terms of the human element of these people. Even though they're, you know, gazillionaires or whatever, they talk about real shit. God damn it, Kai, say something.
A
Michael's trying to get you to chime in.
B
Are you a non TV person?
C
I have a one year old at home. I watch two shows. Coco Melon and Sesame Street.
A
There we go.
C
So unless you want to talk about Elmo.
A
Coco Melon, to talk about.
B
Oh, my God. It goes in your head over and over and over again. Every day, right?
C
Every day.
B
Because they want to see the same thing over and over.
C
At one point, I got my daughter into K pop Demon Hunter, which actually has decent music, but that was.
D
That was.
C
And she kind of got over it. Now she's back into Coco Milk.
B
So is she walking?
A
She is, yeah.
B
Okay.
C
She's a very good walker.
A
Oh, that song from K Pop Demon Hunters could drive a grown man to the brink of insanity.
D
I like it.
A
You like Golden?
D
Let's talk intangibles.
A
Yeah.
C
All right. Now you got me.
D
Let's turn Khaed.
C
Now I'm here. Powering up.
A
All right, let's go, Johnny. All righty. Do it.
B
Episode 242.
A
All right.
D
Whoa, whoa, whoa. Stop the clock. Here's a word from our sponsor. What growth strategy are leading reas using that most firms don't? Segmentation. Some clients needs are sophisticated and require deep, ongoing planning. Some clients needs are simple like those in the wealth accumulation stage. The smartest firms know planning shouldn't look the same for every client, but the experience should always be exceptional. Now it can be with Betterment Advisor Solutions. It's the platform built for segmenting your book and streamlining these smaller and simpler accounts. The onboarding experience is automated and paperless. The portfolio management is streamlined and tax efficient. The client experience is consistent and modern. And the impact isn't just felt by your clients. It's felt across your entire practice. Imagine a back office that's humming. A team that's thriving in a service model ready to scale. Betterment Advisor Solutions. Your biggest regret will be not doing it sooner. Learn more@betterment.com advisors.
A
This episode is sponsored by Clearbridge Investments. Amid rising geopolitical tensions and continued market uncertainty, investors are looking for stability. Even before recent developments in the Middle east, stocks backed by real assets were gaining momentum and can offer more predictable cash flows as volatility increases. Position your investment portfolio for wider equity participation with fundamentally driven Clearbridge active equity strategies. Clearbridge, a Franklin Templeton company. Go to clearbridge.com to learn more.
C
Welcome to the compound and friends. All opinions expressed by Josh Brown, Michael Batnik and their castmates are solely their own opinions and do not reflect the opinion of Redholtz Wealth Management. This podcast is for informational purposes only and should not be relied upon for any investment decisions. Clients of Ritholtz Wealth Management may maintain positions in the securities discussed in this podcast.
A
Episode 242 of the number one investing podcast on planet Earth. My name is Downtown Josh Brown. First time listeners, first time viewers. Thank you for joining us. This handsome gentleman, well dressed gentleman to My left is Mr. Michael Batnik, Co host of the show. Say hello Michael.
D
Hello. Hello.
A
All right, we have two returning guests and I'm super excited for this episode because we are in a market that is literally on fire right now. We have so much to talk about. Ben Carlson needs almost no introduction. Ben is the head.
B
Michael, do.
A
Yeah, do this, do this down.
D
I don't do a turkey neck, do I?
B
Come on.
A
All right, Ben. Ben needs almost no introduction, but we'll give him one. Ben is the head of Institutional Asset Management at Rith Wealth. He is the co host of the wildly popular Animal Spirits podcast with Michael Batnik. How long has that been running?
D
We started November, November 2017.
B
2017.
A
Oh my. Coming up on the Time Flies. And he is the author of A Wealth of Common Sense, which is a blog that for more than a decade has been required Reading for anyone serious about investing. Ben is a cfa. He spent his entire career managing money for endowments, foundations and long term investors. He is also the author of the brand new book Risk and Reward which we're going to dig into today. Congratulations, Ben. No one will read it. No, I'm just kidding. This is a huge accomplishment for you because not just because how prolific you are in terms of publishing books, but the blog, three days a week. Ish.
B
The blog helps me write the book
A
so that I totally understand as a blogger turned author. But still, it's a lot of work. It's very impressive. Congratulations. All right, and another returning guest, Kai Wu. Kai is the founder and she. Founder and chief investment officer of Sparkline Capital and investment management firm applying state of the art machine learning and computing to uncover alpha in large unstructured data sets. You have two ETFs we allow to
D
say, we're allowed to say, we could
A
say it, but you can't confirm or deny that they exist. Is that how it works? Somebody told me the ETF rule of compliance rules, it's literally the most insane shit I've ever seen. So you can't talk about your own ETF on your own podcasts or blogs, but if somebody interviews you for their magazine or website or whatever, then you can engage.
D
So I can ask a question.
C
I can respond to your question.
D
So Kai, you are the founder of Sparkling Capital and you have two ETFs, Itan and DTAN. Can you confirm or do that?
A
Do not say anything until I check if you can. Okay? Okay. And you must answer in the form of interpretive dance. So I don't, I don't know if you want to say. All right. The AI trade came back to the front burner very quickly after Liberation Day. It basically became the only game in town. I said on TV today, I really don't think there's anything good going on in the economy other than the AI Capex boom. I think it's basically taken over anywhere that you see economic growth. You could trace it back to something that has to do with a trillion dollars in Capex spending directly related to this AI buildout. And everything else is kind of boring and, or bad.
D
But I don't want to gloss over the fact that the AI trade was d like the Oracle blow up as a result of what Sam Altman said on Brad's podcast. Yeah, was, wait, wait a minute, people got super spooked and there was a lot of questions about, hey, wait a minute, how are they paying that five year, $300 billion contract to Oracle. Oracle stock fell 60% and then time passed a little bit and the models improved a lot. And every other day we're hearing that Claude is at a new milestone in terms of the revenue. So I think the market is reacting to changing fundamentals.
C
Yeah, I think what happened was Claude Code stayed the market.
B
Yeah, right.
C
I mean people say there's two ChatGPT moments, there's a release in November 22nd of the consumer product and then there is Claude Code and the Opus model that came out November, December last year. And it just kind of sparked a
A
whole wave of developer Sonic 4.6.
C
That's right, yeah, yeah. And yeah, I mean just massive adoption by, by software developers and that's just, you know, kind of completely change the narrative.
A
The revenue growth that Anthropic announced was like the most insane. People's jaws dropped because we all understand it's growing, but I don't think people really processed the sheer tens of billions of dollars that are, it's, I don't want to say out of thin air. No, it is, it feels like it's, I mean it's not even a public company. So listen, do the numbers.
D
So John, you have chart four for a second. So Oracle stock is trading at a multi month high because there, look at, look at Oracle because hey, wait a minute. The numbers that we're hearing from, from these alums are actually pretty insane and maybe the contracts are money. Good. So this morning I listened to Patrick o' Shaughnessy interview Krishna Rao, the CFO of Anthropic. And Patrick tweeted a list of surprising and mind boggling stats in this conversation. Here we go. Net dollar retention is over 500% on an annualized basis. Anthropic's flow.
A
What's net dollar retention?
D
Basically how much money you are retaining and growing from your customers.
A
This is for anthropic.
D
So 100% is like all right, great, nobody canceled in the aggregate, but you're not growing that, that those contracts. 500% on an annualized basis. Anthropic's first dollar of revenue came in March of 2023. March of 2023.
A
Yeah.
D
Over 90% of code inside Anthropic is written by Claude Code. The head of tax is the heaviest token user on the finance team. Run rate revenue went from 9 billion to 30 billion in one quarter sequentially. Quarter over, reportedly on pace for 50 billion by the end of next month. Cowork is growing faster than Claude Co did at the same point in its life. And Signed. He said this on the podcast. Signed. Two double digit million dollar commits and a 20 minute Uber ride to the podcast. We've never seen the ship before.
A
What do you, what do you think? Are we like overly excited about this or is this like.
D
You can't possibly be overly excited. It's insane.
C
I mean, here's the thing though. If you annualize at that rate for the next two years, Anthropic, the biggest
D
company in the world, you'll have the Fed's balance sheet.
C
Yeah. So it's got to slow down and, but it's impressive.
B
Yeah.
C
I mean, certainly impressive. Kudos to them.
A
Where is this money coming from? It's got to be coming, it's got to be spending that would otherwise have gone somewhere else. It's, it's not like new money is materialized to contribute to a $50 billion revenue run rate for this company is the answer.
D
Everywhere?
C
Yeah, everywhere. I mean tech, tech startups, enterprise industrials, everywhere. Other hedge funds, you know, multimillion dollars.
B
There's a lot of big financial firms that are now totally on board. Like I have talked to some huge blackrock in those kind of places are all in on this stuff.
A
BlackRock launched. I was at the New York Stock Exchange today. This, the Cerebras ipo, which we're going to talk about in a second was going public on the NASDAQ. But on the New York, I think BlackRock had a Digital Realty Trust IPO. I think that either they were celebrating it today or it actually happened today. I don't even, I don't even know. But like to that point everybody is, everybody is all in on this. And again, it's the only economic growth story that there is right now. There's not a second one, Unfortunately. Cerebras raised five and a half billion. They sold 30 million shares plus four and a half million more, which I'm sure they'll do. The green show valuation out of the gates. What, what, what was. Wait, we don't know. We don't know.
D
The bankers priced it at 185, I think. Right. That's a number.
A
Yeah.
D
It opened at 350. It's now 326. So is this a hundred billion dollar market cap?
A
I think that's what, I think that's what it would.
D
And they did, they did a, I think, I think they did $1 billion or they're on a billion dollars now. Obviously we can all do the math. That's absurd there.
A
Why can't it trade 300 times revenue
D
but So I don't know if I don't know anything about this business, I'm not going to comment on the valuation. They're very. Their revenue is growing rapidly.
A
Order book was oversubscribed. 2020 X they said at the open of trade they had five orders. Five orders to buy for every share available like. And so who is that?
D
Who are those orders coming from?
A
Retail. Everyone is in already. This is so every, every institution already owns it because now we have IPOs. Before we have IPOs. They did a series H round. You know how much fun we Fidelity. Fidelity is in the stock.
C
How many letters do we have?
A
Fidelity is in the stock in size prior to the ipo. So everyone's already in it. So it's only retail. This is the biggest US tech IPO since Snowflake. They raised 3.8 billion in September of 2020. It didn't go well after the first day. The stock promptly collapsed. But whatever, it's still, you know, the company's fine.
D
Actually, hold on. Snowflake also went public at 100 plus times revenue.
A
Yeah. So the story here is they have a master agreement with OpenAI for 750 megawatts of inference capacity which the only person in the room that knows what that is is kai expandable to 2 gigawatts by 2030. They actually were trying to go public a year and a half ago. They pulled the S1. The concern was customer concentration. People didn't want to buy a semiconductor company that effectively had like two or three really giant companies. Nobody cares about that anymore. So that's the number biggest risk in the S1 is that OpenAI contract. And I think they have a big deal with the UAE risk. So yeah, I don't even know why people are worried about that.
D
What, wait, what does the company do?
A
They're trying to compete with Nvidia inferencing. So they're not doing tiny little nano chips, they're doing big wafers. There's a technical term and the idea is that the memory sits right next to the compute so that there's no latency. And this is what they say is the ultimate chip for inferencing. So explain. Can you explain it any better or what did I miss?
C
I mean, look, that's the high level. I mean I think this is exactly what you'd expect to happen in a time like today, Right? Nvidia has these huge profit margins grown like a weed at huge scale. What are you going to do? You're going to compete with them, right? Their margin is Just so attractive.
A
Somebody has to.
C
Exactly. So whether it's this company or the next one, or even Nvidia's customers, right? Google and Amazon, everyone's trying to get in the chip game because that's where the money is. So this is very, very much in line with where I think we are in the cycle.
B
So how do you talk about something like this without sounding like a cheerleader? Someone asked me yesterday, what's the difference between now in the dot com bubble? And I said, the biggest difference now is probably that the financial media was a bigger cheerleader back then. But I don't know how you as an analyst can talk about what's going on in these growth rates and what these companies are doing without sounding like a perma bull and a cheerleader. It's impossible.
A
Well, the earnings growth is there, so that's how you do it. Sean, put this together for me for the show today. I'm not saying, like, the semis are cheap, but they're 27 times forward earnings, and that is inclusive of the fact that the SMH is up 220% since Liberation Day. So, yeah, the stocks are up huge, but so is the earnings growth. And a 27 multiple does not rhyme with 1999.
D
All right, I have another side, the other side of that. Not anything. What you said is correct. So Nvidia, the stock has gone sideways from August of 2025 to today. So that's a long sideways digestion considering that the market's been generally pretty positive. And then, of course, it broke out as we're speaking, had a massive move over the last couple of weeks.
A
Put the chart.
D
Over the last seven days, Nvidia has added $900 billion in market cap, which is as large as McDonald's, Disney, Boeing, Uber, Starbucks and Royal Caribbean combined. This is seven trading days.
A
What are these? These numbers almost don't mean anything anymore.
D
All right, so in terms of thinking about Nvidia and the valuation that it should trade at, Nvidia is too big to get a market premium. Next chart, please. It's one thing for Nvidia to trade at 50 times forward earnings or 6 times forward earnings when it's earning $20 billion in, in net income. So you could look at the chart on the right and so say, oh, the 4p has gone down. It's gotten cheaper. It's like, dude, come on, give me a break. It's $223 billion of what they're expected to earn in net income over the next 12 months. It can't possibly Trade at a premium. It is too large.
A
The valuation. What would the valuation be if it traded 30 times earnings?
D
So glad you asked. Josh, look at the next chart. So chart goat Matt Kettle's working hard today.
A
This is awesome.
D
All right, so check this out. So right now, at a forward PE of 25, Nvidia is at a $6 trillion market cap, give or take a little bit less, and it's 8.5% of the index. The reason why I say it can't possibly trade at such a large premium is because it would swallow the index. If it was still trading at 45 times forward earnings, it would be $10 trillion and it would be 15% of the market. It has to have a size discount.
C
Yeah. There'll be two companies, Nvidia and Anthropic.
A
Right, Right.
D
So I think, I think the way that the market has been treating Nvidia, letting it digest, even though the revenue and the net income and the margins keep going up to the right, I think it makes total sense, perfect sense.
B
So last week or two weeks ago, it was like Google almost reached Nvidia's market cap. Now Nvidia has a trillion dollar lead
A
on them because there's enough trillion dollars more than shares outstanding plus the gain. You know what's so crazy though? This stock, to Michael's earlier point, sat at 180 for six months. You could have bought as much as you now. Where is it now? 230.
D
So you were saying this to Brad. I thought you made a really good point. So it's 235. It was there for almost a year. You could have picked it up, but
A
I guess it was under 200. You could have bought as much as you want.
D
But you know what, I understand why investors didn't, because in October when we were in Austin, it reported a monster beat. A monster, monster, monster. Be like an LOL type beat. The stock popped up 4% and ended up closing down 4%. And it was very understandable for investors to say, all right, I guess the trade's over. And it was until it wasn't.
A
Is it irrational to have Nvidia be a 10% position in a, in a, in a portfolio?
C
Seems like a big position, but is
A
it irrational given the size and scale of the business and how important it is to everything else?
D
Kai last week out of park was saying Nvidia is a sector. So back to Josh, if you view it through that process, the category of
C
accelerated computing is a sector, but they have competition. Right. As we discussed, Cerebrus Just IPO today to compete against them. We have their biggest customers, Google and Amazon, also competing against them. So the history of this stuff is cyclical. Nvidia was one of the most cyclical companies for a long time, until more recently. I think you have to take that into account. This goes to the question of, yeah, obviously the market's up big, and it's up big on the back of the fundamentals, earnings that have increased. The question is not so much are valuations extended, prices extended relative to earnings. It's more how sustainable are these earnings to the extent that they're all kind of downstream of one phenomenon, which is massive capex by the hyperscalers into building AI data centers, which feeds the chip companies and feeds the power companies. It's all one trade. And so that's the big question, right, which is at the end of the day, what matters is what will the enterprise adopt? AI will, you know, right now all these CEOs, the answer is yes. I think the answer is yes.
D
No, you know, the answer is yes.
C
We know the answer is yes.
A
Well, Otherwise, where does $30 billion in quarterly revenue, in revenue run rate for anthropic come from? That's not.
C
Well, it comes from, you know, folks fomoing in, right? From like business leaders being told, hey by their board, hey, if you don't adopt AI, you don't digitally transform, then you're fired. So I think right now there's, you know, obviously an arms race by corporate America to say, hey, we're on top of the ball, we're doing stuff. Now the question is if they will they actually generate meaningful, like, revenue boosts from this adoption. If so, then they'll keep paying and if not, they'll pare back and say, hey, that was, you know, that was an interesting experiment. We're on to the next thing.
A
We had the company that's built our data lake here yesterday. They're called nvent. They're geniuses. And I wasn't like there for the whole meeting. You know me, I just pop in and say, say the most outrageous thing. So I walk in and he's sitting there with my president, with like, you know, my vice president, like going through all. And I'm just like, guys, let's cut to the chase. Are we doing AI? They're like, yeah, we're doing AI. I said, all right, awesome. You know, that was my contribution. But it's sort of a joke. But it's sort of not like every business leader in every segment of the economy at every company just make sure we're doing some AI shit. Like that's what we have to do. Everyone else is doing it. Not just do it, but like, let's do it in a way where I can go back to my board of directors or I can go to my shareholders and say we made X dollars because we're doing this or we saved X dollars because we're doing that. That's the pressure that every leader feels right now. Which is why none of this seems like optional spending. Right.
C
It seems like, I think people talk about this AI bottleneck and this kind of the fact that anthropics compute constrained, but I think that there's this, as you point out, kind of FOMO in corporate America. And then you couple that with this idea that these businesses, anthropic and OpenAI are in a competition with each other as well. They try to try to lock the market up.
A
Right?
C
This is the kind of classic Uber playbook we want to be. We think it's winner takes all, winner takes most. We want to win market share. And so therefore they subsidize token costs. Like token costs are below what it should be. These companies are not profitable yet they're willing to run at a loss because they want to capture market share. Right. So it's not a true price signal. The fact that we are compute constrained now doesn't really mean anything. So I think we should kind of look past that and ask the question of do we actually think that technology will be useful to driving ROI for businesses? And if the answer is yes, you know, how much can the labs and the hyperscalers capture of that versus their customers?
D
What do you think?
C
I think the answer is yes. I think that, you know, having, you know, I was an early adopter of a lot of these tools. I've been, you know, I trained my own LLMs starting in 2019, 2020. I think this stuff is for real. I think the technology is tremendously useful. Obviously it's not perfect, you know, certain things I wouldn't trust AI for. But in terms of like, you know, kind of lower level stuff, you know, basically the history of disruption is this, that whenever new technology comes out, you want to start by giving it kind of the low end tasks where mistakes are forgivable. And I think at first AI was a good analyst. Now I'll give you an example as a quant. So I'm a quant. I code a lot, I run models.
A
We still accept you in this.
C
Yeah, you can tell my quant, right? I mean, I threw a blazer on to try to disguise it.
B
But we know he's like the bund with a quant. That's. That's a first thing.
C
I'm a Brooklyn quant.
A
See.
D
Yeah.
C
No, so I used to employ analysts whose job was to. I'd give them as a research director, tasks. Hey, I want you to go run this experiment. See how this factor would have performed in this market. Cloud code or Codex as well, are perfectly capable of doing a lot of that analysis. Hey, Codex, here's the API key. Go to this database. Here's a schema, Here's a script. Can you mimic this? I want you to study X and it'll come back and you know, we can iterate together. So I think like it's starting to climb up the ladder, taking the place
A
of you having a back and forth with another quant who you're assigning things to. Now you're talking directly. Like the user interface is like it's, it's. It's call and response.
C
That's right.
A
You're saying things to it. It's saying things back that prompts you to say the next thing. It's conversational.
B
Yeah.
C
It's like an iterative process.
B
So the question I have for you and I have like the cognitive dissonance in my brain has been firing for months. Like I know the history of this stuff, but I also know that sometimes these things are different. So like your background fascinates me because you worked at GMO with Grandson and Chancellor and so I listened to your recent. So you're. You got a great new podcast called the Intangible Economy. You talked to Edward Chancellor, who's the author of one of my favorite history books of all time, Devil Take the Hindmost. I love that book. I recommend it all the time. And you were talking to him about capital cycles and you're a very forward looking person. You're also a quant who knows market history. How do you deal with this, understanding history that. Hey. And Chancellor was basically saying, listen, every time this happens, we spend too much money. These Capex cycles, they follow a pattern. This is gonna end in tears. Versus I think Templeton one time said, listen 20% of the time. It really is different this time. So how do you have. Cause I have those competing thoughts in my brain all the time. Yeah.
A
That nagging suspicion that this is going to look like every other capex boom that's ever happened, where they'll go way overboard with spending and we're all going to pay the price for it in the form of stock prices.
B
How do you handle that?
A
Well, make us feel better about it. Yeah.
D
What's your secret, Kai?
C
Look, I think two things can be true at once. I think a technology can be transformative and it may also be a bad investment. The question is, as investors, for us, is less, is less. Will AI. Is AI a bubble or not? That's too simplistic. It should be. Where will the value accrue through the value chain? Will it be the model providers? Will it be the chip makers? Will it be the users? Chancellor's framework's really interesting. So he's a capital cycle theorist. He studied the booms and busts around electricity, the railroad, the canals, the dot com boom. And in every case, aside from one telephone being the one example, one exception because it consolidated into monopoly. But in every single other case, what's happened is all the capital comes into the sector on the supply side to build out the infrastructure needed to run the new technology. But they kind of get over the skis. Too much money comes in, demand may materialize, but perhaps too slowly. So there's an air pocket and then there's a shakeout where prices fall, capacity is unused, and the ironically the guys who invented and built out the infrastructure end up going bust. Right. Hundreds of railroads, hundreds of auto companies, all the telecoms.
D
What would you say to this? That companies know so much more about how to run a business than they did 100 years ago. We have such through sight into how much, what the supply demand imbalance might look like. It seems hard to believe that one day we're going to wake up like, whoops.
C
I think, I think the problem is it's behavioral. I think it's companies can't help themselves. You have like Zuckerberg, you have, you know, the CEOs, all these companies saying, I'd rather go bankrupt than lose this race. Like it's, they, I think they're, it's like peer pressure. They have to. Right, because you know, if you're running company X and you know, you're like, well, well actually if, if Sam Altman and open, I get AGI and I'm not even doing anything, I'm done. Right? Like that, that's the end of my business. So I have to compete. And so once Sam goes all in, everyone else has to go in all into. It's this game theory, this the prisoner's dilemma.
A
Is he the horse?
C
He's the ringleader?
A
Yeah, it's still him.
C
I think so.
A
Okay. I heard Bezos make some sort of a comment where he Thinks AI will be transformative and beneficial for society. But he also sees it as an industrial bubble and most of the spend. I forget if he said most or a lot of the money being spent will end up being wasted.
B
How about this? What would surprise you more? This is a huge bubble that pops and AI is still successful. Like the dot com bubble. Everything we wanted out of the technology.com bubble, everything they wanted and more happened, but we had to live through the dot com bubble to get there. That's option one. Option two is no, this is a perfect handoff. The baton goes from spending to ROI in use and we're off to the races. Like which one would surprise you? More of those two questions.
C
I just think so many things have to happen, right? For the baton to be handed off perfectly.
D
What a hater.
A
Yeah, I mean, why are you so
B
bearish at this scale? It's never happened before, ever.
C
We've never managed to perfectly execute the handoff.
B
I also think that this cycle, this 15, whatever year cycle, has had so many things that have never happened before. But I'm willing to at least have an open mind that it could.
A
Yeah, right.
C
Look, I think there's a chance that this all works out. I think the downside risk is probably not favorable if you're an investor in the infrastructure side.
D
All right, you want to get nuts? Let's get nuts. John Chart 6. All right. Jason Gabfer tweeted. Look, I know none of this stuff matters anymore, but my God, this will be the fourth time the S&P 500 has hit a record high while 5% of its members fall to 52 week lows. Here are the other three dates. July 1929.
A
Not great.
D
January 1973. December 1999.
A
Yeah. So we have a lot of stocks on the 52 week low list for a market that's for a Dow 50,000 party. And you know, without even looking, you already know it's anything housing related. And then it's a lot of software shit and equal weights of a lot of. And a lot of consumer restaurants, a lot of consumer stuff.
D
Home builders, Home Depot.
A
Is that both? Is that K shaped economy but like on steroids? Now does that bother, does that bother you about the tech rally? That it's not as broad as you'd like to see it or do you not get up about this?
C
In a way it's kind of two sides of the same coin, right? Semi Israeli and software sells off. Right. It's almost the same trade these days. I don't think it should be we can get into why that is the case. But you know that that is a narrative today that the more powerful is is AI, the more disrupted is software.
A
Can I tell you something? Yesterday I was looking at software charts. I think they're going to zero. I really do. I know they want all, but I am starting to believe that they are newspapers.
D
Hold on, this is an absurd claim. So when you say they're going to zero, be specific.
A
Like a lot of publicly traded software companies are going to be zeros. Like not like zero, zero, but never, never coming, never coming back. And I'm hoping it doesn't happen to the big ones. But I looked at like 20 software company charts. Every single one of them is either at a 52 week low or close. There aren't even like up weeks. Some of these charts, it's like 25 straight weeks lower close on Friday than the open Monday. Yeah, it's, it's, it's actually unbelievable. There have also been zero takeovers.
D
So does a company like Adobe, let's use them as an avatar. Do those companies make it?
C
I think so. I think so. I mean, I don't know about Adobe in particular. I mean, here's the question, right, which is, so why are these companies down so big? They're down so big because people assume that their moat is code. And with Codex and Claude code, code's basically free. Right. Any one of us here can effectively vibe code and replace Adobe Salesforce, these products on our own. Okay, so if they have no moat, then they should go down. I guess the question I would ask is whether code was ever the moat for these companies. Right.
A
Like I think even distribution was the moat.
C
Yeah, I would argue distribution lock in.
A
Right, but so then why can't OpenAI and Anthropic mimic that distribution? They certainly have the capital they're trying to.
C
Right. So that we saw the OpenAI Development Company, this announcement this week. They're partnering with a bunch of PE firms to essentially launch like a army of four deployed engineers, basically consultants to go into your company and help recommend
A
you use the product.
C
Yeah, use our product and here's how you use it, et cetera. They're certainly trying and I think it's interesting because, you know, I had this view for a long time that the luminaries on AI were kind of like these naive people that they kind of thought if we build it, they will come, let's just build a really cool product and then the enterprise will buy it. I think what they're. What they've learned correctly is that you have to sell too. You have to push this into the enterprise. Like businesses are very slow to move, they're very slow to change. I think this is an important recognition. They've also done partnerships with, you know, the old school consulting firms, the accentures of the world. And then we saw Google is hiring a bunch like a thousand or something.
A
Yeah, you have to get to the decision maker and you have to push them over.
D
But the market seems so convinced of it that these companies are, to Josh's points, metaphorical zeros. Like Adobe is the earnings per share and the 40Ps are still at all time highs. I'm guessing that Salesforce is not, has not seen any material contraction but the market is like all right, we don't care, you're down 6% anyway.
C
Yeah, I think Adobe is trading at a p of like nine.
D
It was up to 10. Yes.
C
Yeah.
A
And like Salesforce, that's an automaker pe. What we're saying is that, what we're saying is that those earnings are not going to show up next year. That's the only answer.
D
A software company with high margins trading below 10.
B
So this, the stock market is like the economy in that we've had these rolling recessions in parts of the economy, but it hasn't brought the whole economy down. And the stock market is the same thing where we're having these losers get separated. So the question is, could we have this whole thing with AI without bringing the whole market down?
A
Well, the semis are gaining more in market cap than the software companies are losing. So it's a net positive for the nasdaq. It's a net positive for the S and P. To that question, like yeah, we're sort of seeing 25 fairly large companies disappear before our eyes. But then there's 25 other very large companies that have just become gigantic like top 20s and P market caps.
C
This chart here shows that the, the SOX index of semiconductor index. These companies are now 23% of the S and P. Right out of nowhere.
A
That's a baton being handed off.
C
23.
D
23, yeah, 23 coming up.
A
So the market, so the overall market is not suffering, but there are gigantic companies whose market cap is literally vanishing. So I say like zeros. I don't mean like there's no more earnings and revenue. I just mean like these are stocks that might never come back again. They could just. And I.
D
Those are take privates at some point.
A
But that's what I'm asking. Where are the deals? Nobody thinks these are cheap enough to lbo.
B
It's way too quick for deals, don't you think at this point?
A
So but if we think there's $3 trillion in dry powder amongst private equity, private debt, whatever, like we know there's a lot of money they want to put to work. Take one.
C
But they're, they're already over subscribed in software, they're full. Right.
B
So right here, buy our private credit fund and also we're buying Adobe.
D
Yeah, yeah, why not?
A
You imagine, imagine the media coverage of that.
B
So, so you mentioned like the earnings side of this thing and we've got a million charts in here that shows how great earnings are doing. Michael, why don't you like pull ups the NASDAQ ones that we, Michael and I talked to a guy from NASDAQ for analysts the other day.
D
All right, let's, let's do chart 14.
B
And like when you look at the earnings, this is the thing that like I said, this is the most logical melt up of all time because it really doesn't look as much like 99 when you think about like the way that these companies are.
D
So in 1999, in the fourth. So at the top, at the top of the bubble, ten and a half percent of the NASDAQ 100 had negative margins. There was no companies that had margins between 50 and 100%. And it's just completely lopsided. So today, today 20% of the NASDAQ 100 has margins between 50% and 100%. 50% of the index is between 25 and 50. That compares with just 24% from 1999. Back then most of the companies had a margin between 10 and 25% or 0 to 10%. These were not the same businesses. Now Dan Greenhouse made a fair point. Dan basically said the biggest myth is that the tech companies today are real. And in the 90s these were like just fugazi companies. And he said that's not true. We had Cisco, Microsoft, Oracle and at the time Cisco was doing $19 billion in revenue. Microsoft is 22. I think if you inflation adjust this. So Microsoft is doing 23 inflation adjusted, that's 44 billion. Cisco with $1938 billion inflation adjusted. So Cisco is doing what OpenAI is doing. Like these were, these were massive companies. It's just that a lot of the rest of the index was pure shit.
A
Well that's the thing. We had pre, forget about profit margins. We had pre revenue companies. We had thousands of IPOs that we don't have right now. Like that's not what's happening.
B
John, gimme 16.2.
A
So I have 16.2. Holy shit. Are you sure you want to whip this out?
B
Target did this for me. So this is NASDAQ 100 earnings growth. For 20 years, this thing has printed 14% annual earnings growth. So you had Mauboussin on your podcast. He always talks about baselines. I can't back this up. I'm just guessing there's no way we've ever had a period like this with earnings growth this high for this long before. So in terms of baselines, the question is, could this continue? Could we see earnings? So yes, because the Nasdaq is up 20% per year or something. Over this, over this period is AI again, back to the baton handoff thing. Can AI keep this kind of earnings growth going?
A
Robotics could do it. You want another decade, you want another decade of 15% revenue growth? How about 10 million robots? How about a million humanoid robots and automating literally every inanimate object in the entire world? That's what's coming now.
C
Golden era of biotech.
A
Yeah, well, that could do it too. We live to 120. Yeah, that'll move the needle. So that's how you get 50, I think. I was talking to my friends who are not professional investors. I said maybe the worst thing you could do is, is have too much cash at too young of an age. Not that the market won't at some point have a horrific event where you wish you had more cash, but like that's going to get bought really fast all over again. Because if we have another, if we have like anything non recession, any kind of market crash, right? That'll get bought up in two seconds. We'll get another V. We had to. Yeah, we'll get another V. You know why? Because people see the degree of change happening and the speed and the acceleration of automation and I think they realize they can't not own the chips, the robots, the AI. So I'm not saying like the market's not going to fall. It definitely will. I'm just not convinced we're going to have like a two year bear market with this much innovation happening.
D
So I remember back in March of oh, this year, this was two months ago, Micron was at 470, had another earnings report that we all laughed at. The numbers were like, wait, what? And then Micron swiftly went from 470 down to 310 in March in like 10 sessions it fell 30%. It went from 310 to 800. Yeah, in a month or two.
B
Right. If this, if this cycle turns for whatever reason, it's going to fast. John, give me chart 16 real quick.
D
Point 3.
B
This is the melt up. Matt did this for me. So the Nasdaq in the 90s, like we're the Nasdaq 100 over the past 10 years it's kind of approaching that. But after the 90s the Nasdaq fell 83% and it took like 12 years to break even. That's the kind of thing we're not going to get again.
A
I don't think so.
D
I don't think so.
B
There's no way you can, I don't think we can have fallout for that long.
D
Here's why. Chart 15.9. So in the dot com bubble, 32% of the index constituents were trading between 60 and 100 times earnings. 34% of the index was trading over a hundred times earnings and 10% of it was unprofitable. Today 60% of the index is trading between 20 and 40 times between 20 and 40 and 20 is trading between 40 and 60. We have 10% of the index that's trading at stupid levels. But we're not, we're just not valued the way that we were in the 1999. So yeah, the market get cut in half, it always can. But it's going to fall 80%. That would be very hard to believe at this point.
A
What do you worry about in this, in the midst of like everything that's going on? What do you, what do you think is the big risk? Maybe even if it's an obvious risk, what do you think is the thing that we should be focused on?
C
I still think that AI is a risk if only because it's the entire market. Right. Like the S and P is you know, 33% MAG7. You add in these chip stocks. 50, 60%.
A
Yeah.
C
Of the passive index, which we're all told that that's like diversified is one theme. So like yes, I mean as things, things continue to be going well as they have today in the past month. That's, that's fantastic. But I just think there's a lot
B
of concentration also such a big part of the economy now that if this turns, Michael and I were trying to figure out what that mean. Like what would, how far would they have to pull back on the spending. But you'd, you'd probably get a recession and a bear market together. It wouldn't just be one.
A
There's a, there's a different answer to that though. We all assume that this ends badly because of a hiccup in the CapEx story. And people have been saying that now for three years. I've said it. All right, so everyone, everyone. But like the other risk is that this works really, really well and, and the job displacement just hits harder and faster than any of us expect. And that becomes an economic risk, which I, I don't know if that affects the nasdaq. I don't even know if the NASDAQ and the actual economy have any relationship whatsoever left. But to me, it depends if anthropic
C
and OpenAI are in the NASDAQ at that point.
B
Yeah, I had a family member who said, I'm worried about my job because of AI. You know what I said? Buy stocks.
A
Well, I've been saying that for 10 years.
B
Right.
A
Just don't own the damn robots. I mean, I don't know, I don't know what your alternative, but like, do you, do you think that that is underappreciated risk? That the job loss like materializes and it's bigger than people thought?
C
I think it's a possible risk, but I actually think that that's an overrated risk. I think, you know, the average person, you know, because they've been watching sci fi movies, you know, over indexes on that as a possible future relative to what I think is actually attainable. Now, never say never. Things can happen that, you know, we can't predict, but I'd say that that is, you know, probably not, not the first thing I'd be worried about.
A
Electricity. You worried about that?
C
Not having enough of it?
A
Yeah.
C
In the short term, could that stop
A
this capex boom from booming? Yeah, that'd be the fundamental constraint. So we've heard about chip shortages. You understand, we short compute. Could the, could the strain on the grid be the thing that diminishes our chance of hitting these earnings expectations?
C
Yeah, look, there are a lot of bottlenecks potentially in the value chain. And you know, it's weird because if you step back, I almost feel like it's a good thing. Right. So like talk about the job loss. It's a governor, it's a governor. It kind of slows down the bill. I like that, like if things happen too fast, the government, you know, regulations are too slow to adapt. The job market, you know, retraining employees and restructuring companies happens too slowly. So in a way, the best case scenario is one in which AI ends up delivering abundance. But it takes like many years for that to happen. The scenario that we worry about is one in which that happens overnight and then everyone's laid off and it creates like a doom loop.
B
So how about like getting again everyone knows at risk. So you got a chart here, John Do 28. So you have talk about how the Mag 7 is going from intangible asset light to asset heavy. What if it's just a RE rating and they go all this spending and you becoming more of data centers and it's physical, it's not intangible anymore. What if we just get a RE rating that way of valuations, that is
C
my base case actually. So I don't think these companies are going to go bust.
A
Sorry, are we rating lower?
B
Yeah, just valuations have to be lower because you're more capital intensive.
A
So let's use an example. Let's take an Alphabet. We loved this business for so long because of how un capital intensive it was. Right. Basically they invented search 25 years ago and they've been eating, they've been dining out on that innovation and the margins were crazy. And then cloud computing, amazing margins. Now obviously nobody's calling AWS and Alphabet Overall or Microsoft, nobody is calling these companies Cap Light the opposite.
B
So because of your, your research on intangibles, you're saying that's your baseline now
D
John, throw that chart back on. Kai, speak to, speak to what we're looking at here please. So we're looking at the Mag 7 Capex to Rev Capex as a percentage of revenue.
C
Capex to sales ratio. Okay, yeah. So this one, yeah, to Josh's point, like the, the Mag 7, why are they the Mag 7, why are they, why do they have their own acronym is because they just have delivered amazing ROIC return on invested capital over the past 20 years. 15, 20 years.
A
Unprecedented.
C
Unprecedented. Never seen any, never be seen before at such scale.
A
Right.
C
Like yeah, okay, it's possible to build asset light businesses that are small but at the trillion dollar scale. I mean no one thought it was possible. And the way they've been able to do that is through these intangible assets, through leveraging brand human capital, network effects, network effects in particular for Google. And that was a great thing while it lasted. And so here's the irony is that these guys are on the forefront. They invented Google, invented the transformer. They're now kind of in a way engineering their own demise because they invent this new technology. What does it do? Well, it changes the rules of the game. Entering the AI revolution, the way it worked was there was a handful of digital services and these businesses kind of like carved it up into their own fiefdoms. All right, you get searched, you get social, you get shopping. And it was a good cozy little oligopoly to have. What happened is with AI now the perception at least is that it collapses all the markets into one. That now it's all about who gets agents first and who wins that market. If you win AI, you win everything. And that's ignited this game theoretical prisoner's dilemma situation where you have all these companies saying, wait, this is an existential risk, I need to do this.
A
They're still spending, therefore if they're spending,
C
I got to spend. Ideally they would all moderate their investment, be incremental, continue to make AI not a disruptive but a sustaining innovation for the incumbents. In this case though, however, if Altman's going to spend the trillion dollars, then you got to spend a trillion dollars too. And so I think what's happening is that they are again, they're better run companies, no doubt than the companies in the 90s. That being said, like they just can't help themselves. It's rational, based on game theory for them to be doing what they're doing individually.
A
We have not seen the multiples derate yet though Max 7s are at an
D
all time high today. The group. So what's the catalyst for rerating?
C
Well, I think the challenge, I think once the market catches up to the perception that, wait, these are no longer the asset light businesses of five years ago. These are utilities. Right. Which is.
A
And very tangible.
C
Yeah, Very capital intensive with massive depreciation. Right. That's the other thing which is. Chancellor talked about this on the podcast. He was saying, look like it's just mechanistically in any of these up cycles what happens is you have companies spending on capex, but capex is a capital asset on a balance sheet that gets depreciated over time. So for a five year depreciation you're only spending for each dollar that's been spent on buying data centers, you're only spending what, 20 cents each year hits your net income, whereas Nvidia gets to record the entire dollar as revenue. And so you end up with this kind of just again accounting based inflation of net income.
A
So Michael and I had this discussion two days ago. We were talking about the CapEx, just CapEx in general, not really being great for shareholders. And this is of course like the biggest capex cycle we've seen. And the example that I, I don't know if it's a good example or a bad example. So I want your opinion on this. The example that I used is like at&t and Verizon. The amount of money that they spent from 1G to 5G, just 25 years of endless billions of dollars. These stocks are the same price they were in 1997.
C
Right.
A
Like, I'm not even, not even exaggerating.
B
Verizon, Josh, you're being paid to wait with a dividend, though.
A
Yeah, the dividend is not even. The dividend is not even great. Verizon, I think it was still called Bell Atlantic in 1998, was $45. It's $45.
C
And here, here.
A
I know there's a yield on it, but like the, my. I guess my question is these companies won. The entire wireless business is three companies. T Mobile, it's Verizon and it's AT&T. And that's it.
B
Right.
A
And what, what did we, what did we win? These stocks don't go up. Great. 5% dividend yields. All right, so it's not zero, it's a bond. So it's not clear to me that this is materially different. I'm pretty sure Amazon, Alphabet, Oracle, I'm sure they're going to win, but what do we win as shareholders? So that's my question for you. What are your thoughts?
C
Yeah, I mean, a couple of things. So first of all, if you look at the capex to sales ratios for many of the hyperscalers, it's higher than AT&T at the height of the dot com boom. Right, Right. So they're certainly in the same category. And yeah, it's a pyrrhic victory. You can win the market, but it's. Is it a market you want to win? Because if your margins are like 60% doing search, why do you want to get.
A
I'll do one, I'll do you one better. You know who won? Apple, like Apple is the beneficiary of the CapEx, the combined CapEx spending of Sprint, which got sucked up into T mobile, AT&T from Verizon. Apple doesn't own any of the networks. They built the best product on top of the network. Right. So it's not that capex bad, capex good. It's that who, what layer, where do the profits accrue. Now we know in wireless the profits accrued to the iPhone.
C
Right.
A
And the iOS ecosystem, we don't know who's going to win, like really win this.
B
Where are you looking for winners?
C
It comes down to barriers to entry, it comes down to competition. Right. If you're in a sector where it's easy for people to compete if your only moat is, oh, I have more money to throw at the problem, well then like, you know, I could call up Masayoshi or whatever and solve, solve for that. So that's not really a true moat. And that's why historically capital intensive businesses have not actually been like the best place to invest. From an ROIC standpoint, what are true moats are the network effects. You mentioned these intangible assets. You know, no amount of money could have replicated Google search at the time. Now it's, you know, arguable that maybe AI will help obsolete that technology.
A
But a lot of my searches are starting on Claude.
C
Yeah, and I think, I think that is a meaningful risk to that part of the business. But again, like, there's the question around the disruption of the existing business. Microsoft is a good example of a company that's, you know, right in the crosshairs here. And then the question of they're going into this new business, will they win? And even so, do they want to win?
A
There are people who think the profitability will accrue to the layer that Palantir sits on, where Palantir basically surfs atop all of these LLMs and is the company that gets paid by other companies to literally tell them how to make best use of all this technology.
B
How about this? So you have this quote in here from Mobis and I don't know where this is from. He said, I think you would have to argue that almost all the profit ends up going to the consumer ultimately, and that's because of competition. So what would that look like if you're saying all the utility goes at consumer.
C
Yeah, I mean, it'll look like the railroads where all the railroad companies went bust and you know, but that being said, a ton of GDP was created by the railroads. I have this chart. Here it is. Chart. Let's see. Hold on.
A
Just don't do 16.5.
C
36. This is from Azeem Azar Exponential View.
A
This is a good one.
C
Yeah, this is a good chart. So the blue line shows cost of railroad construction each year. The green line shows how much money, how much earnings the railroad companies earned.
D
Oh, wow.
C
And the red line shows the contribution of railroads to gdp.
A
How do they calculate that red line? I don't know what goes into it.
C
You got to click on the link.
A
Okay. They made it up.
C
They made it up. But I mean, look, if you take this at face value, what you see is the costs above the earnings. You can see the little spikes. Those are the panics when all the real companies Went bust. And then even after all that work, at the end of the day, yeah, they were making some profits, but they're utility like profits.
A
The users made the money.
C
Who made the money?
A
It was the manufacturer. Furniture in North Carolina, put it on a boxcar, shipped to California.
C
I understand that's right.
B
So they'll be like AI is going to help business formation or something. It's going to make it easy for people to do everything now.
A
Well, so here's what the hyperscalers would argue. They would say, well, we're invested in the LLM layer. Like we have partnerships and like literal, literal investments in some cases.
C
So.
A
And they would say we're going to get a piece of that too.
C
I actually happen to believe that the LLM layer is also at risk of commoditization. Take a look at number 29. So, so what this chart shows.
A
Oh my God, what's going on here?
C
So this is, what this chart shows is like the. Just the Y axis is like how good are the models? Right. Go back to November 22nd when ChatGPT was released. The underlying model was GPT 3.5. You can see that at that point the OpenAI had a huge lead over their competitors. In fact they were like two orders of magnitude high.
A
Now it's a horse race and the leadership is switching back.
C
Yeah, it was Gemini, then it was Claude, then it was GPT 5.5.
B
Llama is a donkey apparently.
C
Llama. That didn't work out too well. Yeah, those efforts didn't work. Then you have Deep Seek and you have Chinese.
A
I don't think one of these players is going to break away in terms of capability. I don't. But one or two of them already have broken away in terms of critical mass, like the amount of users. Like I think we have a Coke and Pepsi brand. I think we have a Coke and Pepsi and Dr. Pepper basically at this point I think Gemini arguably is Coke. Say Claude is Pepsi. OpenAI is Dr. Pepper.
C
Like I think other way around, I think that Claude. Claude is Coke.
A
Maybe Claude is Coke already.
C
I think a lot is. Or maybe Open AI still Coke. Opening I, still Coke. Yeah.
D
I think the Average person knows ChatGPT. That's right. The address doesn't apply.
C
It's more of a.
A
Yes, but the average person is doing more AI stuff with Google and doesn't even realize it. Every search is now an AI workload. So I think it's. I think Gemini is the number one like AI in terms of usage. People could, they don't. They're not doing on purpose. They're getting Gemini results instead of search results.
C
So Google has a special advantage is they have distribution. Right.
A
They're putting Gemini in my email.
C
That's right.
A
Whether I like it or not, it's in there.
C
It's built into an installed base that is significantly larger than what, OpenAI.
A
We don't think that's bigger than OpenAI right now. I do.
D
It could be measured.
A
Depends on how you measure it.
C
Yeah. And it's a question also of like, you know, also the quality of the model. I think right now it's clear that Gemini is in third place. They will likely release a new model next week and we'll see.
D
Kai, I have a two part for you. So what first brought you to intangibles? Like, what attracted you? Because your entire, your entire operation is built around this idea. And then part two, what does all of the spending do to the intangibles inside your portfolio?
C
Yeah, so answer the first question. So I used to work for a company called gmo. We are, you know, kind of like aqr, one of the, you know, pioneers in quantitative investing. One of the bread and butter factors within, you know, GMO and other firms is the value factor. The idea that, you know, you buy, if you systematically buy stocks that are cheap relative to, say, book value and, you know, underweight, those that are expensive, that, that historically has earned excess returns.
B
Did Chancellor say that you helped them update some of their models? Like what? Did he say quality? You kind of helped them update it a little. Bring up to speed to.
C
Yeah, that was one of the things I worked on when I was analysing,
A
like modernize the way they're capturing.
C
Yeah, look, and it's been many years. I'm sure they've done more upgrades over time. And again, not specific to them, but it's been a challenging time for value investors as defined via these systematic factors. I had Mobison on the podcast. He made a cheeky point. He was like, look, the fama French value factor is this academic factor that basically is an index of cheap value stocks shorting or underweight expensive stocks. It made money every year for like 80 years. And then it stopped making money in 1994 when they published their paper. Right. So. So that's the thing.
A
So arguably, like when America Online came out, but you know what I mean, like, or maybe the Internet changed when
C
we all got computers or. Yeah, and so that was the big question, which is like, you know, we were in a tough place because you're sitting there and you're saying, all right, well like this beautiful idea, you know, that Ben Graham coined, you know, 100 years ago with security analysis, that it's
A
buy low, sell high, it makes perfect sense. It should work.
C
No one is arguing that buy low, sell high doesn't make sense. The question is against what intrinsic value. So that was always my contention. It's not that value investing doesn't make sense by definition. It should make sense if you know what true value is. The problem is that the way we were measuring value was just obsolete because so many of the traditional metrics are backward looking. They don't take into accounting stuff, R and D, they don't take into account advertising and marketing, human capital. They're based on these tangible assets that at one time 100 years ago made a lot of sense. But as the economy has transformed from, you know, industrial to information based asset light companies like Google and Nvidia have come to the forefront.
B
So you consider yourself a value investor still?
C
Absolutely, yes.
A
Just valuing things differently.
C
Yeah, exactly. Because look, it's, it's, the challenge is that if we don't have an anchor of value, how are we going to invest? Right? I mean there's, there's different ways to invest, of course, but you know, I do think that being able to kind of marry the two schools, the, the idea of, you know, being a value investor, having some kind of price discipline, you know, buying, buying bargains and knowing when to kind of sell and take profits, you know, is a really important stabilizer for markets, a really important thing to be doing. But we need to update our metrics, right? Think about Warren Buffett and Berkshire. I talk about the story a lot that, you know, he started off as Ben Graham's like actual disciple and he bought Berkshire Hathaway, a struggling textile mill.
A
He was only buying struggling things.
C
And yeah, and it worked okay. But looking back after meeting Charlie Munger, he talks about this, he's like, look, I'm never going to do that again. And then he bought Coca Cola and then Apple. Right. And I think the implicit lesson there is that he added intangible moats to his framework.
B
And Graham would have done that too. He used to buy stocks that were worth less than cash. Like, guess what? That doesn't exist anymore.
C
That's right.
B
He would have updated his, I think by the end he was saying like, he's an indexer essentially in the 70s. But all those guys, they didn't.
D
He retired a. Yup. He said like, yeah, shit doesn't work anymore.
B
Yeah, yeah.
C
I want to say that like 84% of Buffett's investments were purchased with a price to book above one. So in other words, he was not a price to book investor.
A
So he changed his mind over time. Not just, not just because he met Charlie Munger, but also, like, the world also changed.
C
Yeah, the world changed and he had to evolve his process with it because as you point out, Ben, there just weren't enough things to buy other horses.
A
He sort of also was an intangibles investor because he recognized the power of brand Geico. Yeah, he recognized the power of things like American Express and Coca Cola and what the brand meant to the consumer before anyone else was really talking about that. So there's like a lineage from Buffett to what you're doing with your intangibles portfolio.
C
Yeah, so what I'm trying to do is to kind of do a systematic version of that. Right. Like trying to say, I think, you know, anyone with common sense can sit there and be like, yeah, I think brands matter. I think human capital matters. I think this company has really interesting technology. The challenge is how do we quantify that? That. And so for the longest time, quants were in this tough place because we had crisp combustat the traditional databases, which were all structured data. In your intro that you read for me, you mentioned unstructured data. Unstructured data is everything else. The information in patents, in trademarks, in company filings and news and analyst reports on Twitter, all this information contains obviously very valuable information on companies and on their intangible assets.
A
If you can bring order to it,
C
if you can parse it. And I think for the longest time, the challenge was that quants, we had linear regression, we had a few tools in our tool belt, but they were all optimized for a world where structured data was what we had available. Now with large language models and natural language processing, we're finally able to unlock this huge trove of information where I think the task, which at one time seemed unattainable of trying to codify a Warren Buffett style approach is now on the table. Not saying that, you know, I've done it or, you know, anyone has done it yet, but I think that, you know, we're increasingly moving towards a point where through AI and all these tools, a lot of the, you know, kind of intuition that's baked into a fundamental investment strategy can be codified in quant.
A
So how do you quantify brand? Because we're in a market right now where I think the worst stock in the world is Nike and Lululemon. Lulu's number two might be. Lulu might be following it down the drain. Is there a way to. Is there a way to parse the things that might go into a calculation where you could say this, forget about the stock price having lost value, the brand was losing value a year before and like, that was the signal to not be a Nike. Like, can. Yeah, can we do things like that? Is it effective? Is that part of your approach to.
B
By the way, these companies are both down 76% from their highs equally.
A
Yeah.
B
Crazy.
A
Those are apparel specific. But just generally, like, how do you say that a brand is either gaining or losing value?
C
Yeah, I mean, absolutely. If you were a Peter lynch or someone, you'd say, hey, you know, talk to your friends who are Lulu customers. Are you still buying their products? Right. There's a.
A
Not too bad. I'm friends with Peter Lynch.
C
Oh, there you go. Nice.
A
Yeah.
C
You should ask him then, what he would say.
A
I don't think he wants to see me again, but I interviewed him for an hour.
C
That's awesome. He's a legend.
A
And he was telling these great brand stories about walking into the supermarket and seeing like, one brand being sold at the register and another being sold on a shelf. And, you know, like, he. He was doing this in a very analog way. Right, okay.
C
Yeah. And I think increasingly you can start to systematize some of these things. Yeah, right. Social media data is an obvious thing. You can go on Instagram mentions, brand mentions, and like, what's the tone of the mention? Is it positive or negative? Right. Like, is it amongst the right people? Like, there's a kind of in crowd of cool people. Are they. You know, again, you can buy influencers. You got to be careful for manipulation as you do with all signals. You know, companies can juice earnings, you know, whatever, but if you're careful about it. Yeah. If you track all this information in theory, you can certainly capture, you know, where, where the trends are headed before, you know, ideally the stock price.
A
How do prediction markets factor into this? You must be really excited about that data because that is like, I mean, especially things that are very far afield from what would normally be in an 8K or, you know, some sort of, like, official filing. There's a lot of opinions being expressed there. And you can quantify it.
C
Yeah, I think it's a really important and interesting source of data. I'm not using a ton of it, to be honest, yet, I think, because most of the stuff I'm doing is at the company level. Right. So it's Very useful. If you're trying to do macro forecasting, who's going to win the election? Obviously there's a lot of sports betting and crypto stuff which is, you know, less relevant there. But you know, is, is Nike going to make a comeback? I mean that's, I don't know if there's a contract on that. If there were, it would probably not be a very liquid one.
A
I don't even know how come, like, how would you even.
C
I. I've. One thing I've used. Yeah. One thing I've used is Google Trends. Google Trends is an interesting source of data.
A
Okay.
C
You can like look at what people are Googling like terms like if they're googling, you know, the lululemon align like pants or whatever, that's like a potentially good thing. And if that's going down, that's a little bit concerning.
A
You have two ETFs. What's the difference between the two? And which one should Ben Carlson buy? Pitch it. He's sitting right here. Get him to buy it.
C
So yeah, the first ETF is US based, the one I launched in 2021.
A
Okay. It buys a portfolio of stocks itan the intangibles etf.
C
Yes, that one.
A
Somebody else had to say it.
C
Thanks, Josh.
D
Detangibles, what's that one?
C
Developed markets.
A
Yep. So international intangibles.
C
That's right, yeah. And so that the second one follows the same exact strategy, both of which their goal is to buy stocks that are cheap relative to a expanded definition of intrinsic value that includes these intangible moats too. So brand human capital, IP network effects,
A
you know, you know about like how the IPOs are all going to join the major in the indices faster than ever.
C
That's right.
A
You making plans for that new world where you go to space X and it's in the s and P 10 days later probably similar thing will happen with anthropic. Like how. How does your work get affected by that? You kind of have to play along with it.
B
Yeah.
C
If it's in the index, it's in the investment universe. I don't have. It's not a. It's an absolute return fund. So I don't need to have, I don't have a benchmark per se. I don't need to own anything. Right. We're trying to make money over the long run. So to the extent SpaceX is added to the index, it's now investable, which is nice. I doubt it'll be in the portfolio day one. Just given how it's a quant.
B
Do you have any discretion? Are you rules based? Completely.
C
Completely rules based. The idea would be if there's something that looks weird that doesn't accord with fundamental intuition, I'll ask the question of why the model is missing that and try to adjust the model in a way that solves the problem. Not just for this one case, but also moving forward.
A
Are companies like SpaceX and Anduril and some of the things that were. Some of the more exciting companies coming, are they intangible assets or are they tangible assets or are there. It's a. Obviously, everything's a mixture.
C
Definitely a mixture. I mean, some of these companies are a little bit more physical. Right. They're not pure software companies. I mean, SpaceX, they launch rockets, but obviously most of their value is in their ip. And I do think that a lot of their value is just in Elon Musk, in this case, his ability to raise money.
A
It's the ultimate intangible asset is Elon Musk's involvement, his aura. Yeah, right. Hard pivot. Ben's book. Can we. Can we say some words about it? Are you allowed to speak about it? It's not like an etf, right?
B
Yes. There's no compliance rules. All right, I will say, I told Michael the other day, I'm not good at sales or self promotion like Josh, you're a salesman.
A
Is that good?
B
Yes.
A
Okay.
B
Josh could take. Josh could take a ketchup popsicle in 90 degree weather and sell it to a woman wearing white gloves. You could do that.
A
Thank you.
B
That's not me. Okay, so I'll. I'll make one hard sell for the book.
A
Okay.
B
There's never been a better book for charts and tables than this book. I counted because. Chart Kidmad. Help me.
A
Wait, say it again.
D
That's a bold claim.
B
There's never been. There's 52 charts.
A
You have charts in every chapter.
B
There's 52 charts and tables in this book. It's got more charts and tables and data than any book. And chart Cadmed helped me, so I'm giving him all the credit. Yes.
A
Way to not sell the audio version. What are you doing?
B
The audio version comes. No, it comes with a PDF of all the charts. They send it to you.
A
Ooh, that's a good idea.
B
Right?
A
Whose idea was that? Is that Craig?
B
Yeah. Okay, so anyway, that's my. And I got a couple charts. There's one chart I want to run by you guys, so chart 50. John.
A
Let's do it.
B
Okay, so I looked at the worst Days, the worst months, the worst years in stock market history. Okay, so I did one that looked at the worst months in stock market history and most of them are in the 30s, 40s, the 1987 on there.
A
It's in chapter one.
B
Yeah, chapter one.
A
I read this chapter this morning on the way into the city.
B
So I know exactly.
D
I can't believe you asked Nick for a quote and not me.
B
The funny thing is Nick was actually on the COVID and they bumped him for Morgan. So I feel bad for that.
A
Wow. Poor guy. Is Morgan's quote even hot? Let's say he has mastered the art of Ben Carlson has mastered the art of exposing the few big topics that matter most to investors. Let me hear Nick. What's Nick's quote? Nick's not even on the back. Oh man, where'd Nick go?
B
Here's what I want, here's my point. I want to make talking about faster cycles. So there's been six times where the stock market has down 20% or more in a single month. I think you could make the case. And again, a lot of these are in the 30s. You could make the case that going forward, instead of having these massive long drawn out crises like hey, we're down 60% over three years, we're going to have these more air pockets because information moves faster. Where we have these huge down 1, 2, 3, 4 month periods where it's like, oh my God, the stock market was down 25% in a single month. That's where I think we're headed in terms of the speed of these things. I think there's going to be more
A
air pocket speed running the correction because
B
I wrote a, I wrote a chapter on the Great Depression and people keep asking me, do you think it could happen again? And I think we've completely cut that left tail off. So the question is we've cut that tail off because of fiscal policy, monetary policy. Okay, you can't, the risks don't ever go away. What does that mean? I think it could mean we just get faster, more of these things.
D
So buy and hold gets harder, mate.
B
Don't you think that these cycles just speed up? There's no way this stuff is slowing down and AI is going to just add more.
A
That is literally what has, has happened.
B
My contention is that's where we're going is just cycles are going to be.
D
Of course that's your contention.
B
Right.
A
What do you think about that?
D
Let's talk go well hunting.
C
I think that's right. Yeah.
A
Like everything else is faster. Why wouldn't the market's ability to process bad news be faster.
C
Everything's happening faster. There's a lot more trading. 24 hour trading. Like we're all. We're heading towards a world where yeah, everything will be instant on your phone.
A
Right. There's no rule that says how long a bear market has to be. I know the. I know the old heads like to say the average bear market is 13
D
months when we haven't had a true economic recession in a long time.
B
We haven't had his credit cycle 18 years. We have had a credit cycle.
D
We'll have another one.
B
All right, I got one more chart. I want to spike the football in some people's heads. John, throw up chart 53 on Japan. This is the question I get more than any other.
A
Now show Japan.
C
Meaning.
A
Meaning the.
D
The implication is why are people so obsessed with Japan?
A
Because. Because Ben says it pays to be a long term investor. And people like oh yeah, Japanese people sat in down stock market for three decades.
B
So no one. I've never seen people show this from. Yes. From 1990 to 2024 Japan did nothing. It was like one and a half percent per year. You wouldn't know it was. Japan peaked in 1989 in like December 1989. It bottomed in March of 2009 which is crazy horrible. But if you extend it, the returns were so good in the 1970s and 1980s that it was so compressed it had to be bad. So from 1970 to 2024 you got almost 9% in Japan. Long term investing did work in Japan. It was just all those returns were compressed in the first. Yes.
A
So you did 22% a year from 1970, 1989.
B
Small cap stocks in Japan did 30% per year for two decades.
A
So then you do. So then you do 1% a year from 1990 to 2024 which is a lifetime.
B
And guess what? The average worked out. You still did okay over the very long term in Japan.
A
Over the 60. Over the full 60 year period.
D
Yes, the cycle.
B
Right.
A
It's a good way of thinking about it. I think most, I think most people would have preferred if the. If the returns were back end loaded rather than front and load. I do. The book is called Risk and Reward. I know Kai is very excited to read it. You're going to, you're going to listen to it?
D
I'm exhausted.
A
Is Ben going to read it to you?
B
Yeah, we're going to call each other.
D
I'm going to listen to it 100%.
C
I have the audiobook too. I'M going to listen to it in bed.
A
All right, so I'm. I'm reading because I'm old school, but I love it. I. So, you know, I'm, like, one of the biggest fans of your writing in the world, and I make it through the first two chapters, and I'm just like, yeah, man, this is what I need. This is the medicine. Because it's all about. Things will probably be okay. It's. It's unique. Most people writing financial books, it's this. The dollar is going to not be the reserve currency anymore or gold is going to replace. You're just saying, like, no, no, no. Things will be okay.
B
The 10%, and here's how you know the 10% per year over the last hundred years is inclusive of all the bad shit that's happened, right? The Great Depression. That's part of it.
A
Yeah.
B
Right? 1987. That's part of it. 70s. That's part of it. All the bad stuff that's happened is inclusive in the long term. Returns that are still good.
A
We've been through a lot and things are still okay. Yes. And that's a really great message. I love it. We should end there. Guys, did you have fun on the show today?
C
Great time.
A
Yeah.
C
Kai, you brought. Thank you.
A
You're so smart. Why are you so smart? Do that another time. I want to tell people where they can learn more about your research because your research is really spectacular. Your funds are great, but, like, you're a thinker, you're a philosopher, and you test your ideas with data. And I love. Anytime your stuff drops, where do people go to learn more about Sparkline and your work?
C
Well, thanks, Josh. Yeah, you can just go to my website, sparklinecapital.com Sparkline Capital. Okay.
A
And you're active on your tweets?
C
I tweet sometimes. I. I try to respond.
B
This guy's so smart. He doesn't blog. He white papers, Right?
A
He's writing white papers while we're doing blogs. Yeah, no doubt. All right, guys, thank you so much for listening. Great job to the crew. I know you guys worked your asses off this week. John, Duncan, Rob. Amazing. Nicole, Daniel, Travis. Happy birthday to Graham Thomas again, everybody. Katie. All right, guys, thank you so much for listening. We'll see you soon.
Date: May 15, 2026
Host: Downtown Josh Brown
Guests: Michael Batnick, Ben Carlson, Kai Wu
This lively episode dives into the surging market euphoria driven by AI and semiconductor stocks, focusing especially on the sustainability and risks of the current bull run. Special guests Ben Carlson (Author, Head of Institutional Asset Management at Ritholtz, Animal Spirits Podcast) and Kai Wu (Founder/CIO, Sparkline Capital) join hosts Josh Brown and Michael Batnick for a deep exploration into the impacts of AI, capital cycles, "winner takes all" dynamics, market concentration, and valuation frameworks for today's intangible-heavy economy. The group also takes some entertaining detours into media and culture, but ultimately grounds the discussion in what investors should consider as tech fever hits new highs.
For further learning: