The Compound and Friends - Why Demand for Compute Is About to Explode With Alex Kantrowitz

Episode Date: July 31, 2026

On episode 253 of The Compound and Friends, ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Downtown Josh Brown⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ and Michael Batnick are joined by Big Technology founder, Alex Kantrowitz, to discuss: OpenAI and Anthropic’s explosive growth, the threat to traditional software companies, soaring cloud demand, agentic AI, and the diverging strategies of Microsoft, Meta, Amazon, Apple, Google, and other tech giants. Plus, where AI profits may ultimately accrue, and whether today’s massive spending boom can deliver lasting returns. This episode is sponsored by ProShares. Learn more at proshares.com. Sign up for The Compound Newsletter and never miss out: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠thecompoundnews.com/subscribe⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Instagram: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠instagram.com/thecompoundnews⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Twitter: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠twitter.com/thecompoundnews⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ LinkedIn: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠linkedin.com/company/the-compound-media/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ TikTok: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠tiktok.com/@thecompoundnews⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Investing involves the risk of loss. This podcast is for informational purposes only and should not be or regarded as personalized investment advice or relied upon for investment decisions. Michael Batnick and Josh Brown are employees of Ritholtz Wealth Management and may maintain positions in the securities discussed in this video. All opinions expressed by them are solely their own opinion and do not reflect the opinion of Ritholtz Wealth Management. The Compound Media, Incorporated, an affiliate of ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Ritholtz Wealth Management⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠, receives payment from various entities for advertisements in affiliated podcasts, blogs and emails. Inclusion of such advertisements does not constitute or imply endorsement, sponsorship or recommendation thereof, or any affiliation therewith, by the Content Creator or by Ritholtz Wealth Management or any of its employees. For additional advertisement disclaimers see here ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://ritholtzwealth.com/advertising-disclaimers⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠. Investments in securities involve the risk of loss. Any mention of a particular security and related performance data is not a recommendation to buy or sell that security. The information provided on this website (including any information that may be accessed through this website) is not directed at any investor or category of investors and is provided solely as general information. Obviously nothing on this channel should be considered as personalized financial advice or a solicitation to buy or sell any securities. See our disclosures here: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://ritholtzwealth.com/podcast-youtube-disclosures/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ProShares' disclosure Investing involves risk, including the possible loss of principal. There is no guarantee any ProShares ETF will achieve its objective. ProShares leveraged and inverse ETFs have a daily investment objective. For any holding period other than a day, your return may be higher or lower than the Daily Target. These differences may be significant. Carefully consider the investment objectives, risks, charges and expenses of ProShares before investing. This and other information can be found in their summary and full prospectuses on ProShares.com. Read them carefully before investing. Distributor SEI Investments Distribution Co. Learn more about your ad choices. Visit megaphone.fm/adchoices

All right. So did you hear about this kid? Leopold. No. I wasn't even going there. Okay. We'll get to him. I don't know his name.

He has an, I guess, an AI company called Lemon Lime. Yes, the tattoo guy. Okay. You heard this? No. All right. So it's a kid in San Francisco. He's like a YC kid.

And he threw a party. and he said anyone who gets a tattoo gets a job interview. Just the interview, right? Not the actual job. Not even the job. In the interview. He's got a tattoo artist there and people are getting tattooed. I don't know if they're getting the company logo.

I think it was the logo. Oh my God. Not only that, he took that and used it as a moment to share how great the company was because they provided these free tattoos. Permanent. That was LinkedIn material for him. Actual tattoos. Yeah.

And he thought, I mean, this is at least what I'm getting from the discussion. So I could be wrong here. But from all indications, because he's already come out and apologize, from all indications, I read the apology. If you went to the party and got a tattoo, that got you an interview. What did he apologize? What did he apologize for? For the whole thing.

He linked getting a tattoo at this party to getting an interview, basically like almost preying on the desperation to get a job these days. And people did it. And people did it. So imagine you. Imagine, imagine them, first of all, just think about the people. I understand desperation, but imagine the mentality of somebody who sits down in front of a tattoo artist. And it's like, yeah, do it.

Because I might get a job interview. Imagine thinking that you're going to get good candidates. I mean, that's crazy. That's even, right, right. I'm going to hire, like, okay, whatever. I read the, so I read the apology. This is, so, go ahead. But then, but then, because I can't help myself, I clicked on his bio.

Like the, and it's exactly what you would expect. It's a private school delivered right into Stanford. I'm sure the parents know people. I'm sure it's like a whole thing. And then he has like somehow he got a startup funded and it's AI and it's San Francisco. And it's just like this like super privileged kid with like almost like no, no awareness of what the world is like. like if you if you if you if you think about how desperate people are to get jobs at these AI companies they're they're hearing if you live in san francisco your assumption is probably this is the only

job that's going to exist it's gross it's gross well look so i lived out there for a while and it is a really good encapsulation of what goes on in silicon valley because you have people many people there have moved away from home they're looking for opportunity and so they're willing to try different things and that is that is well you might get a startup tattooed you might do acid or mushrooms on the weekend. You might be part of like a polyamorous relationship. Like all of Silicon Valley has this, I'll try it, mentality. And it goes wrong when somebody like that will prey on people's willingness to say,

okay, you know, I'm here for new experiences, here to do new things. Tattoo me and let's do the end. I feel like Jin Yang would have got a tattoo. Like this is straight out of the show. This is straight out of the show. This would be like Russ Hanneman is like stock for tattoos. I want you to walk around San Francisco with my company's logo. And if you do, I'll give you stock in my new startup.

So Alex, it's literally that. You've been doing this with us for so long. Right. Like we've got, Dan, you ready? You made a great call. So Chart Kid dug this up.

He goes, look at this call Alex made. Huh? Oh, headphones. Headphones. Dude, you're a genius. No, what was it? Well, we already know he's a genius. Watch this.

Watch this young Alex. You end up in the situation where you're going to forsake your bread and butter to try to half-heartedly introduce some of this stuff that's been really good for a competitor. And you almost end up stuck in the middle. And that's why I really think that if they want to try to compete with TikTok, just go, you got to go all in on it. You can't go half. And I think they've come up with a new, completely new app to just be a TikTok clone. What I would have done is taken, and this is what they did with stories when they were getting their butt kicked by Snapchat. They put stories in the top of the feed. And we said, we don't care that this is

the most important real estate in Instagram stories up top. What I think they need to do now if they want to compete with TikTok, the front feed of Instagram has to be exactly like TikTok. Wow. And then you create a bunch of wheels. They're testing that right now. Aren't you glad you didn't wear the same shirt? By the way, I try it off.

I know I am. We all aged. That call aged very well. I thought you were going to say I age, which is true. You did. We all looked different. But that call aged very well. Well, they're now experimenting with doing that completely.

on like all Facebook apps. So you open the blue app even. It might be Reels or TikTok based. This is what people want to see when they open an app now. They want to see videos. Well, it is, I mean, it's sort of, it hits your dopamine receptors. You sort of say any other type of content consumption is too hard. Like I think right now, even when people watch Netflix or they're watching on Amazon Prime,

there's like two feeds, like you're sitting back and you're watching the actual show, but in your hand you're also scrolling. And how many times I have to, you're so right, you know, how many times I had to pause the show on TV? And rewind. And rewind. So I was on the train. I was on the train for 45 minutes the other night. And I was with my partner to Chris and I'm on,

I'm on Instagram the entire time, giggling to myself because it's serving me videos that it knows makes me laugh and me alone. Well, not me alone. But a lot of men who fit my profile. And I said to Chris, oh, this is why Netflix stock is down 40%. The shit is too good.

It's so good. And I think that people got a taste of it during the pandemic. Like, that was when TikTok, really started to have a rise where you were home, you were looking for anything that would make you happy. And you're like, well, as long as we're in lockdown, you make a deal with yourself. As long as we're on lockdown, I'm going to scroll TikTok for a while. Then we came out of it and people stayed in their feeds. I'm not going to say who, but I went to the Odyssey with

three other people, with four of us. Two of the four people, this is a, this is a movie that it's like $30 for a movie ticket. And like, you're like going there to, you're the reason to see it in the is to be immersed in it. Two of the four people I was with, and I'm not one, not me, had their phones, almost, I watched peripherally. I watched them pull their phones out,

like seven different times. And they're not checking texts. They're like into Instagram. Yep. And they're not in it for long because they catch themselves. It's a habit. They can't help it.

It's the most compulsive. Like you, you almost feel like, all right, I got to check my phone, but you're not checking for emails or text anymore. You're checking for more.

viral videos. Okay, by the way, I know we're going to talk about meta in a bit, but in meta's earnings, one of the things that was not talked about enough was, I think, 13% of their decline in profit or 13% of their costs are coming from legal costs. And they said basically, like, watch for this, because this could end up increasing and we might be, you know, under the gun from a legal standpoint more and more. And there are these lawsuits, these addiction lawsuits. Yeah, right. They're not losing these, though. They did lose. Did they? Yes. But what does it mean? Money?

You know, it could be money. It's like, you know, it just opens the door because if you have 3.6 billion people using the apps, you lose one and all of a sudden the floodgates are open. Lose one lawsuit and then everyone says, oh, we can get some money for this. They addicted my kid also. Well, they absolutely are. Yeah. Very effectively.

It's going to be a very big problem for them moving forward, among the other problems they have. What? The addiction to... The fallout. getting people so hooked on these on these apps. You know, something tells me nothing will happen because you make the right political donations and you just kind of skate through. But this is this is being played out in state courts.

So it's not like we're not like talking about like some presidential administration sicking the FTC on META. We're talking about a local court. So a local judge probably in the community seeing a kid probably from the community who, you know, will tell a story about how, you know, they got. so addicted, they lost themselves, they, you know, sort of endured all this harm. Because of meta apps, judgment for the plaintiff. Yeah. And I mean, it's on me, it's meta that's, that's been, that said it.

Yeah. And it's, they've gotten so good at this. They literally know exactly what you want to see before you even know that that's what you want to say. So time spent, engagement has always been the North Star for this company. No matter what they say, it's always been, how can we get people to spend the most time and engage the most? And from a business standpoint, you understand it.

because people are so fickle. Think about all the dead social networks that have come and gone and meta is still enduring. And the reason why they still endure is because they've been so good at hooking people. But the problem is they're so good at hooking people. Yeah, they're almost too good. It's diabolical.

And once you spend time on one particular video on Reels, the whole algorithm changes dynamically to capture the fact that not only is this your type of video, you're in the mood for more of this, right now. It's personal super intelligence. It's unbelievable. If only there were a way to take this power and harness it for good, what would that be? So the rumors that I've heard, and this is probably true that, because in TikTok, there's a setting that enables this, is that in China, they actually mandate part of the feed is science and math content. So, but not here. Not here. Not here from

my algorithm is Britney Spears twirling around in circles with knives. Is that science? That's science. Is that math, Alex? That is also math. That's my algorithm. My algorithm is Brittany in a basement. Eating a bagel.

But I have to tell you, there's got to be a way. I believe this. It's got to be a way to take this ability they have to laser focus everybody on their own feed and, like, make it like a positive. Okay. I just, I don't know if they care to. They, they, if you wanted it, that's what they would give you. No, I know. But you don't.

And people don't. That's the thing. Like, they will, the data will show that anything like eating your veggies, though people might say they want it, they will scroll away from it. Because Britney Spears with knives is right, is the next, you know, the next one. It's kind of like online dating. I just want to see if at some point she accidentally drops one or cuts herself. I don't know really what to do.

So every time meta will serve you your vegetables. Instead of that is a moment for you to go to TikTok, which will give you, you know, Britney Spears with 21 knives instead of 20. What a dystopic arms race. That is the world we're living right now. We're watching play out. Yeah.

So, all right, well, on that happy note, you've got a show to do. Right, Ms. Nicole? Ladies and gentlemen, TikTok addicts, Miss Nicole. Her whole, her whole feed is, I don't get to want to guess, I don't know. Hello Kitty. Like, what are we into? Whoa, whoa.

Stop the clock. Here's a word from our sponsor. Today's podcast is brought to you by ProShare's. For nearly two decades, ProShares has been at the forefront of the ETF revolution. As investors prepare for the next market cycle, ProShare's has a lineup of ETFs that deliver unique exposures to the tech-heavy NASDAQ 100 index. Whether you are bullish or bearish or somewhere in between, explore what Pro Shares has to offer at ProShares.com.

Investing involves risk, including the possible loss of principle. There is no guarantee any pro-share's ETF will achieve its objective. Pro-shares leveraged and inverse ETFs have a daily investment objective. For any holding period other than a day, your return may be higher or lower than the daily target. These differences may be significant. Carefully consider the investment objectives, risks, charges, and expenses of pro-shares

before investing. This and other information can be found in their summary and full prospectuses on pro-shares.com. Read them carefully before investing. Distributor, SEI Investments Distribution Co. 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 Rithold's wealth management. This podcast is for informational purposes only and should not be relied upon for any investment decisions.

Clients of Ritholds wealth management may maintain positions in the securities discussed in this podcast. All right, guys, we are going to have an amazing show today. Thomas you on Tuesday. Here on earnings tech week, we were going to bring on the world's, God bless you, the world's foremost authority on big tech. And he is here, longtime friend of the show, one of my favorite people, Alex Cantruitz is the founder of the big technology newsletter and podcast covering what else? Britney Spears covering big tech.

Alex also hosts the big technology podcast where he has interviewed the world. world's top tech CEOs from Mark Zuckerberg to Larry Ellison, along with leaders like Sam Altman and Dario Amadeh. Alex, how long have you been running big technology, the podcast? How long has that been going for? Six years next month. Amazing. Right around the time when you started, right?

You also started during COVID. Yeah, that sounds right. Yeah, 2021 was us. Is that the same for you? 2020. 2020. All right. You still having as much fun?

It is the most fun. I mean, every week, as you know, is just the craziest week. So we break it down and it's always fun. Your ability to get guests who are executives, founders, people that sit at some of the biggest technology companies. I mean, it's really incredible how you're able to do that repeatedly. Talk a little bit about how you plan shows, how far in advance you have to line up a guest of that magnitude. and what it's like when occasionally, I guess, you get rejected. Like, how do you, as a creator, how do you rebound mentally from that?

Well, first of all, I get rejected all the time. Okay. And so like... So that already would not work for me. It's hard. Everybody says yes. Right, Nicole? All right, go ahead.

So, yeah, I mean, it's frequent rejection. And oftentimes when people come on, it's years in the making. I would say that one of the things that I do that sort of helps, you know, make the case is I write about these companies and I write about these CEOs. So for instance, when Dario came on a year ago, that was the culmination of like a two-month reporting project that I did writing this profile of him and about anthropic about what drives him and what drives the company. So we did myself and one of my interns, we did about 30 interviews. We spoke with everybody from his co-founders to his rivals and ex-girlfriend. And I think the companies,

when they see you do the work like that, they're like, we would much rather tell our story than have you sort of connect the dots and tell it for us. Right, because you're tackling these topics in a very intelligent scholarly way, and they know they're not going to send somebody to talk to you where you're like, so, like, what's AI?

Right. Like, they don't have time for that. Well, I think that if you look at what's happening in AI today, it is crazy. The transformation that we're seeing is insane. And so I think, you know, from my standpoint, as a reporter and somebody that runs a podcast, the best thing that I can do with my time is try to explain what's happening. So like that, like, did you order the code read, although I did ask Sam Altman that in December, because Open AI was in a code red.

That type of journalism, I think, is almost, there's a place for it. And it should be done with these tech companies. But the place that I said is like, I'm seeing you do this. I'm seeing things move this way. Explain what's going on. Help us understand it. And, you know, I do. did, I have a couple of charts that I put in the document beforehand, just to talk about the pace that this stuff is moving at, that I think would be like talking about. Which one do you want? Do you want to do this Open AI? So let's do OpenAI weekly active users to start. So the last time I was here with you guys in this room was compounded friends 145, which was I think in June 24. Good memory. It's been that long? That long in this room. So we've done remote, but we haven't been

had Alex on before. Face to face since then. And at that point, chat CPT was less. than 200 million users. All right. So today, 900 million announced, and according to censor tower, it's more than a billion. So we've seen that type of growth in such a short amount of time. Let's go to the autonomous coding chart. All right. So in 2024, basically the amount of time that these models could code on their own was zero. So you would tell them go do this task, they couldn't do it. They'd fall in their face. Now we're seeing them on the tests do the equivalent of 12, sometimes like 16 hours of what a human would do to code before they have to be interrupted. So that is the type of progress we're seeing.

And the last one I'd say that we should look at is revenue. So 24 when I was here, basically zero revenue, right? Today, they're close to, they're expected to do 80 billion this year, Open AI and anthropic, mostly leading the way. Look at this chart. But it's going to be, it's going to be even bigger than that. I think it would be a surprise to folks if it was less than 100 billion. So with this AI story, it's just like important context to keep in mind.

Like when people say they know what's going on, generally they're guessing because the context shifts week by week. The story is moving on a weekly basis. Absolutely. We may never see anything like this again. This is once in a lifetime. Once in a lifetime, this speed of adoption of a technology that's a few years old and this growth in revenue. And then the coding chart is crazy to me. I was listening to, you probably know who the person is.

I think it's like the CTO of Anthropic or something. He was talking to Joe and Tracy. And he was saying... Oh, Jack Clark. Yeah. One of the co-founders, yeah. He was saying like a year ago during his talks he would give at technology companies, like all over the world, all over Silicon Valley.

He would raise your hand if you use Claude to do any coding. And he's like half the room, the hands would go up. And then now there's no point in even asking the question anymore. and so everybody is using Claude Code or its equivalent to do work. And then a lot of these rooms, it's situations where the code is like almost running the show and the people are there to check its work. And that, he said that happened like a year. So it's an explosive change in the way that technology firms are working and it's only a matter

of time before it hits every type of company. Yeah. I mean, I had this presentation that I was giving last year that, It was like a sort of 101 of what's happening with AI. And the grand finale was, I'm going to vibe code a retirement calculator and Claude. And I would show that and the room would go nuts. And now I would show that to you and then get booed out. People would be like, I do that every day.

So the pace of change, you know, it's, you know, we see this happen like week by week. And, you know, even though it feels like the leaps are week by week, when you zoom out a bit, you start to see the totality of this massive shift that we're going through. And the investment narratives are obviously changing rapidly as well. So just two weeks ago, you tweeted, this is an earthquake goodness. And the retweet was like the big news from Kimmy K3. And that feels like, did it blow over? Like, I don't even know.

No, I don't think that's blown over. And I think, you know, as you've seen some of the wariness around big tech spending, part of that is sort of asking where is the value going to be found? Because there are many layers that can sort of capture the value of this AI moment. It could be- value meaning where the profitability results. Yeah, where are the margins going to be. It could be the compute.

It could be the model builders. It could be the applications. Still kind of up in the air about who's going to capture it. So like what Kimmy K2 does, it says, I'm going to put downward pressure on the model part of it and seemingly make the products far more important. Okay. So let's disentangle that a little bit for the people that aren't following this as closely as you are. the new debate in a i mean there's a million but the new debate that over the last couple of weeks seems to have gotten louder so we had everyone remembers the deep seek moment yes basically

china has a couple of companies that were able to train their large language models by utilizing american large language models and then just distilling from there so they were like more efficient and you more or less arrived at the same answers. But they used less power. They took less time. They were cheaper or in that case free. And it kind of spooked the stock market for a week. And then everybody got over it. But now we're hearing stories where like 50% of the model downloads are these open weight or open source models sort of akin to the way Linux came along and threatened the Microsoft hegemony in the first.com age. And so now the new debate is like, okay, do we even need to pay frontier models from the top labs in America?

Do we even need to pay them for 90% of the things that we're all doing using an LLM? Or is this intelligence now like already free? And you would only pay somebody like Anthropic for the last mile or for something that needs to be seriously refined. And what does that mean for all of this spending? and all of the standing up of data centers, et cetera, if that's being threatened. The charts that you're showing show that that doesn't really matter because you're showing me open AIs at a billion users.

So, but I guess the question now is like, how much of this stuff that we thought would be profitable, it just turns out can be imitated really cheaply and people won't really care what they use as long as it doesn't cost too much. Okay, so there's so much here. I know. Let's go bit by bit.

You know this really well, so we need to learn this. Okay, so first of all on this idea that the Chinese models are distilled from American models. To a degree, yes. What they'll do is they'll just run a bunch of queries, get the answers, and they'll sort of bake that into the intelligence of the model. But the distillation part of this has been given too much credit. Because at the end of the day, what's happening with AI is the blueprint to build these models are on the internet. Right. So basically all you need is compute, data and a big model.

And the bigger all three of those get, the better performance you have. Now there are some new tricks that you can use to make the model perform better, but that's at the core of this. So, you know, this idea that China couldn't build the model, you know, of course they can. The blueprints are on the Internet. Now, the constraint within China is compute because we will not sell the cutting edge Nvidia chips to China. So what China has to do is specialize. So, you know, large language models. They're large.

Which means that when you, you know, write a query to chat chip-y-t, you're getting something that probably is, you know, can answer your health questions at a high level, can answer science questions at a high level, can go search the internet and tell you if your train is delayed, can go into your email and draft emails for you. What's happened in China is there's been a specialization, right, the constraint, and this is something Grace Schau, who's a China analyst told me on the show this week, on my show this week, is basically because China is constrained, they have to focus all their efforts on certain areas. So Deep Seek was all about, reasoning, which is one of these practices to make the model better.

And Kimmy K2 has been about the Sygenta coding. So it might not do so well on health like an open AI model does, but in areas like coding, they can equal the frontiers. That's why we're seeing this challenge. Because they just have this one narrow thing that they are perfect for. Put your smartest people on this one thing. So there is a website. Is it a, what is it, AI router or, you know, there's a lot of you.

Somebody's trying to buy it for $10 billion right now. Right, right, right. This is where you go to it and you say what you want done. And it's deciding which model or which combination of models can you throw a harness around and have them pull together and come up with the right answer. It sounds like if there are going to be a lot of specialized models and you're a specialist in a field, right? Yes. So that's an open router. Yeah.

So if you're like if you're doing drug development, you're probably not on chat GPT. you probably already have a favorite specialty. But like having this sort of like the Google of routing AI queries is an interesting thing so much so. Somebody's trying to buy a website for $10 billion. Yeah. So this is where it gets interesting, right? So, okay, so China can equal and other models can equal

the big foundational models in different areas. It does still cost, right? So you still have to like host it and then pay to run it because you basically, these are very involved. solve calculations that you need to run. You're grinding through tokens. Yes. But what you end up having is, you know, a march towards parity between, let's say, the big

open AI, maybe not complete parity, something like it, between the big open AI antanthropic models and the Chinese models and maybe GROC and maybe meta, just in certain areas, not across the boards. I think the big foundational models will always be a little bit ahead. And so then going back to like, where is the value found? Where do you make the profit in this world? that I think what we're seeing is it's coming down to who builds the best AI products. So it's going to there's been.

Sounds like software companies. Correct. It's been this misconception that you're going to build. If you let's, I'll just put it this way. If you had one or two companies that build, you know, human level artificial intelligence and nobody else did, then they could obviously sell that at a pretty high margin and that's a good business. But now when it looks like you're going to have five, six, seven, eight companies doing it, it. All of a sudden, selling that at a markup is no longer possible, and you start to see a

price work come out. So, for instance, today, there is just news that Open AI is going to cut. It's two of its, two of the three latest models that it put out one by 20% per token and the other by 80% per token, right? So this is, we're starting to see the prices come down. Why are they doing that to prevent people routing workflows elsewhere? Well, they see what you're saying, which is like people are using these routers and they are, they are saying, all right, I don't need to use open A. Maybe I'll just use Open A.O.N.A.I. for like the tough coding stuff, but all the other things I'll use the Chinese models or metaphor.

Now, just the last point. So we looked at the massive exploding revenue, right? Where's that coming from? So a year ago, a lot of this was API. So when I was with Dario a year ago, the majority of Anthropics revenue was coming from the API. So Anthropic licensing its model to other people who would pay it by token, is effectively word. And they're using it on AWS.

Correct. They have access to, okay. Yeah, that was the majority of Anthropics revenue. Yeah. Then Anthropics started building its own products. Right. Claude code is now a substantial part of Anthropics revenue. I would be stunned if it wasn't the most, you know, if it wasn't the majority of Anthropic

revenue. And it's competing head to head with Microsoft co-pilot. Well, with GitHub co-pilot because it does coding. Yeah. With Open AIS codex. But you're starting to see. So it's like, where is the profit going to be realized? All of a sudden, it's back at the software layer.

The product. It's hilarious. And that's why Alex Karp went crazy, by the way. Yeah. Maybe not crazy. Why he had this conibption on Squawk Box a couple weeks ago. Because what Open AI and Anthropic are doing now is they are putting consultants within companies, right? Forward Deployed Engineers.

And what does Palantir do? Yeah. Very similar. So now he's coming directly in contact. The Alice Karp freak out is interesting because he says, I'm just saying this out loud, but this is what everyone is saying. saying in corporate America, I'm the only person dumb enough to say it, but I'm going to say it. Anthropic and Open AI, he didn't use them by name or maybe he did, but that's who he's talking about. They're using your own data to train themselves to put you out of business.

That's like kind of what he was saying. And it sounded very personal to him as though, right? So he's like, once you let them in and hand them all your data, you're basically training them to be able to replace you. And you know who echoed that? Like a couple weeks later? Satya Nadella, Open AI's biggest funder in the early days,

came in and said when you use these models, you generate exhaust. And that exhaust is used to either make the models better or potentially build new products. And we see it. Exhaust is like the memory of the function it just carried out

that it now knows how to do that. Correct. Like basically the best way to go, let's say you're doing a business process and you have the best practice to do it. you know, there's an argument to be made that you've just run it through Claude. So now Claude knows how to do it. Now, I don't know if that's if that's 100% true or if it's a scare tactic.

But what it does indicate is people like Satya Nadella, people like Karp, understand that for these companies to be able to see their valuations through, they're going to need to go upmarket. And there's a great example of Claude coming up with this product called Claude Design. And the head of product, or the former head of product that Anthropic used to be on the board of Figma, which does design. Yeah. And the CEO of Figma, Dylan Field, said, Anthropic hasn't been consistently candid in their communications. Right.

Right. Which is kind of interesting because I think that was the line that was used when the board of OpenAI fired Sam Altman. But anyway, so he sees it coming for them. And now within the labs, there's going, that both Open AI and Anthropic are building. these super apps. So it's not just about code anymore. It's about an app that can do anything for you, which effectively it ingest software. So they're becoming SaaS companies. To a degree. And so one of the biggest, go ahead. I'll just say this. So this is my take here. The SaaSpocalypse was right but misguided. The idea that you're going to vibe code Salesforce. So Salesforce is going to go

out of business, ridiculous. Yeah. And that blew over. But, you know, can you end up using a sort of agentic CRM, right? So all you would need is the database. And then your AI assistant, you know, does all the functions of a CRM. I don't see why that wouldn't be. So the SASPocalypse isn't coming from the guy from the user vibe coding his way into a CRM. The SASPoclapse is coming from potentially if it was going to happen. It would come from an open AI or anthropic saying, we want to play in that model. Just we'll do everything. We'll do all the agentic stuff.

Just find a place to store the data. And then if it does that, what is Salesforce worth? If you're an enterprise business that's currently paying Salesforce and somebody comes along and says, why are you paying them? Let's say anthropic. We can do this. We can do this. You're just going to pay them.

You're paying somebody. Right. Right. Because there has to be a layer to make it worth. to your point, product, there has to be a product that makes the power of this superintelligence actually express something that's useful to you as a business. So maybe it will replace one of the publicly traded SaaS giants, but you're still spending money with someone. So when you show me that revenue chart go from zero to $80 billion, like the money's coming from somewhere.

So where is it coming from? Where are these revenues coming from that they, other where would that money otherwise have been spent? So I think for a lot of companies, they're still trying to reckon. Like there were no buckets for it, right? So it's margin compression. Yeah, I think so. I mean, but, but okay. It's new spend.

For them, like a lot of companies realized there was going to be some agentic capabilities. So when you go budget for, you know, for the next year in 2026, you put some money there. But the companies have just blown through that bucket because the capabilities have gotten better. And that's why we had Uber talk about. it. But you're right. It's going to have to come from somewhere. And so that's why if you over said they had a compute budget and they blew through it for the whole year in March. I like March. Exactly. Right. Now, I think that like, so when you see these these curves, I think one of the things that's important to consider is there's capabilities. You know,

when you go from zero, oftentimes, I mean, this is obviously the revenue looks crazy. It is crazy. But you're going to see capabilities that will far exceed businesses' ability to implement them. So the market will panic about a SaaSpocalypse, and I'm telling you, I think it's going to happen. But that's something that might take five, ten years for people to actually see it through. Wait, what's going to happen? That AI agents have a chance to replace many software tools that you use today. One of the areas that is clearly benefiting from all this AI buildup, so there's a lot of a lot of debate over where, where is value going to accrue? How impaired are some of the

frontier models? Like where does that all go? One area that is clearly a winner is the cloud. So Microsoft just reported. Yes. And Daniel, third chart far up. So Azure's growth is accelerating. And if you compare, so it was a 43% year over year, which is the highest growth rate since 2022. And there was a lull because Google was eating their lunch. And that number, the revenue growth dipped into the mid-20s. And so you're seeing an acceleration at $30 billion in quarterly revenue. Back then, it was like $10 billion of worth of revenue. So the fact that it's accelerating at a base that is three times the size is insane.

$100 billion in sales for the first time during the fiscal year ended in June. So the run, like the overall run rate is going crazy to the upside. Right. Yeah. The profit was 31% higher year of year. What did you think of the quarter? I mean. And Google was up 82% on cloud.

Which is crazy. And Amazon's going to report while we're doing this show, which will hopefully have the number before we're done recording. But like, they're all seeing reacceleration. Michael's right. That's the, that's definitely an area where there's a benefit of crewing. Yes. So I would say, obviously, good quarter. But I think a nice quarter.

order doesn't necessarily mean that that is the trajectory forever, which is obvious to anybody listening. But basically, like, Microsoft is benefiting a lot from the AI surge right now. AI typically does not go in a straight line or even an exponential line, right? There's choppiness. So you go from AI progress. Then typically there's an AI winter. I'm not saying that's going to happen here. But wait, what do you mean typically like in another dimension? No, that, okay, so AI didn't start in 2022. The AI is something that began in the 60s or even earlier. And there were concepts that were proved out.

A lot of funding went into them. And then they achieved some degree of progress, but not enough based off of the funding. And so the funding dries up and it goes into what's called AI Winter. So I think that there's always a possibility. So there's a leap and then there's a pause. Correct. And then you wait for the next leap.

Yeah. Like in this situation, the techniques that have led us here have been the large language model. You know, there have been other, you know, types of techniques in the past that did things for image detection, for instance. Now, they're related. But, you know, people see the computer begin to think, right? I'm putting that in quotes, but feels like thinking, and they go nuts. And Josh, you've said this before, and I agree with you.

This is real, right? This isn't NFTs, but the degree to which it's going to lead to economic activity is still a big question mark. So it's possible that, you know, right now we're still. starting to see people question the spending. It has to deliver profits for this to keep up. So that is the risk, you know, long term for Microsoft. I think we know there's an economic effect. I think the question is productivity.

That's right. But just replacing paying for one thing versus not paying for another and there's no uptick in productivity, that will get eyebrows raised. But it's too early for anybody to definitively declare one way or the other. Yeah, I know it's not showing up in the productivity stats, but. It's going to. It might have a measuring problem on the person. Anybody who spends, like, more than 20 minutes in these tools, I know.

I know. Understand. You're way more productive. What are we measuring it? It just started. Yep. I don't know. I mean, you could try to say maybe we're seeing, I don't know, people use, like, more lines

of code or more dollars to software companies. But again, like, if you see the curve where the bot goes from not being able to code autonomously at all two years ago to being able to code autonomously for 16 hours now, or the equivalent of 16 hours now. It's a capability that still needs to be channeled. You have to decide what to build. You have to decide how to sell it. I think that often gets lost from these conversations is like, oh my God, AI can code for 16 hours. No more software engineers have jobs anymore. And then you realize, like, you still have companies that are using this stuff. So companies have to take control of these capabilities, figure out how to use them, and then we're going to start to see

the productivity. On Microsoft, so one of the storylines is that they broke an exclusivity agreement with Open AI. In the early days, they were an equity investor in Open AI. They re-upped a few times. And the idea was that Open AI would power all of Microsoft's AI apps. That's now over. Why is it over? And what does it mean for Open AI? What does it mean for Microsoft? Well, OpenAI, as you may know, has a problem holding on to partnerships. Yes. You remember two years ago they partnered with Apple to be the AI behind Siri. Now they're out.

Now they're being sued by it. And they're out. Well, you could still access ChatchipT and Siri, but they are now being sued by Apple for stealing Apple's roadmap and service of its device. And Apple ran back into the arms of Google. And they have one with Disney. Correct. And then right with Sora. And that's over.

And that went away. And so here we have Microsoft. So this was. This was the biggest commercial partnership in all of AI. This linkage between Microsoft and OpenAI was a huge deal. So a couple of things happened. So first of all, I think the deal was always with Microsoft. Satney and Adela basically saw Open AI and said,

you build AI or you build artificial general intelligence, AI on human level intelligence, a human level intelligence capability plane. You go build the sort of brain. You take consumer. So you do chat shift. We'll monetize it. And we'll do enterprise.

And that was, everything was hunky dory. They were the best friends in the world. And then OpenAI said, oh, Anthropics doing Enterprise and that's really working. And we're going to do Enterprise. So then OpenAI started to move to Enterprise. And by the way, it's not a coincidence that after Open AI moved to Enterprise up until today, Microsoft has been struggling very much this year. So Open AI is big, big push. What was the move to Enterprise Codex?

It's codex. Yeah, it's codex. It's also just like working with companies to make sure that the tools will work for them. What does this mean now for Microsoft? They are going to be in a situation where now you can kind of bring your own LLM. You're not just using chat GPT or OpenAI products. Correct. So as this was happening, Open AI also was growing a lot. Yeah.

And my understanding is they did not like the constraints that Microsoft was putting on them. Remember, Microsoft owns 27% of OpenAI. Microsoft has access to OpenAI IP till 2032. And Microsoft, by being like the sole provider for OpenAI on cloud, sort of got to determine how much compute OpenAI could use. And Open AI needed way more compute than they wanted. Meanwhile, like, so, you know, what did you get for that? If you were Microsoft, you got like, okay,

exclusive access to Open AI models through Azure. But you also had this clause in the contract that if OpenAI were to say, hey, we've hit AGI or artificial general intelligence, AI on par, with human level intelligence, then you would lose your right to make money off of that. This was a complete mess. That's crazy. And even Sam, Sam Altman, Open AI CEO this week has talked about how one of the things that he's done wrong, and I think we could all agree with this, is that he has had this belief that he could innovate

on corporate structure. Like, it's almost like the, these, which he obviously had with the nonprofit. Yeah, yeah. He had it with Microsoft, with this AGI clause. It just doesn't work that way. Like, sometimes people with brains like that

think they could change everything, but some things like having a normal board and being a for-profit if you need billions of dollars, and there's a reason why they worked that way. Did you hear him answer the question about why he doesn't own any stock? I still don't like his answer. I didn't understand.

I didn't get that. Never lands. What did he say? I didn't hear that. He said, like, the biggest reward for him is that he gets to be at the helm of this startup that's changing the world. Front row seat to history. Why didn't you buy it? That doesn't land for me.

He does own investments in some of the startups or maybe all of the startups in OpenAI's startup fund. So potentially, like, the way that he's making money is... Yeah, he's steering winners and losers by virtue of I'm running Open AI. I decide which startups are worth something. Yeah, I mean, I don't know if it's steering. winners and losers, but certainly, like, if he builds great AI models, he's going to stand to benefit in some way. No, no, no.

If he's a shareholder in a company and that company is going to add 100 million users by virtue of being built into Open AI, Sam can do that. So, okay. And I don't hate that. Let's talk about this. Yeah. Sorry, go ahead. But it's, it's a not, that would be a better answer.

I'm an investor. I'm going to, I, I, I think this is America. Everybody should make as much money as they can. I would like that answer so much better. I miss the boat on Open AI shares. I have the best job in the world. I love it. Also, I'm invested in lots of related companies

that Open AI does business with. So I have some second derivative upside as part of my job here and it's good enough. Like, that's probably more or less the truth. I don't think there's something insidious. I just think that's what it is. Just say it.

Hasn't say it. I agree. I agree. I think that would be better. I also think that, it is you put yourself in an interesting position when you're investing in the startups that are using your technology. Because so Sam also in these interviews, in both of the interviews he did this week, said that his goal is to provide intelligence to everybody and not to build, basically hoard it and build products using that intelligence. But, you know, if that is sort of the, if the path forward for your company is to build products, that you put yourself in a weird position.

if you're basically funding all these companies and you're also going to go out and build products in some way, that's a conflict of interest. Oh, if they end up working on a product that could compete with one of the startups he founded. Yes. Something tells me he could just acquire himself. I mean, we wouldn't seek circular funding in the AI.

Can I ask one more Sam Alton question? Yeah, definitely. Before we move on from this, I think the hugging face hacking incident is him doing marketing. He doesn't seem at all concerned about it. In fact, he almost gleefully is talking about it, almost like it's a third-party thing that happened

and not his own company releasing a product into an environment where it was able to take over the host, so to speak. So just if you could explain for the audience what that hack was about and then maybe you don't have to confirm what I'm saying. But I sort of feel like he saw the... this project glass wing thing come about. So Dario comes out and says, okay, we're a little bit nervous.

We have this new model that's so powerful. We need to form a team of Avengers like to make sure that it's not going to hack the whole world. That's marketing. In the AI world, people don't run away from that. They run toward that. They say, wow, it's so powerful. You had to get CrowdStrike and J.P. Morgan and the U.S.

government all in a room. So to me, it feels like Sam is like trying to repeat that, like a viral moment. Like, yeah, we don't know what happened. It's so powerful. It hacked hugging face. So tell people what went down and what we need to know about that. Okay. So in AI, so there's basically two big schools of AI development.

One is called self-supervised learning, which is basically you teach it to predict patterns. So I give AI all of these books and say, what's the next word in a sentence? Like, the sky is. And it was like, oh, sky is blue because that's the most common, you know, common word in a sentence and books. And that's basically the underpinning of chat GPT in large language models. It's a word calculator. It's a prediction of words.

But an interesting thing happened. So there's another version of AI called reinforcement learning, whereas basically, like, I'm going to let you go play a game. I'm not going to tell you the rules. I'm not going to tell you how to play. I'm just going to, you know, give you the controller, go win the game. And the AI will play this game thousands, millions of times until it figures out, on its own how to go play the game.

When you put AI into a reinforcement learning scenario, the AI is ruthless. So it will, in some cases, this has been documented, put it in a chess player. And to win the game, if it doesn't have the right strategy, there have been documented cases where the AI has actually gone into the root of the game, hacked the game, to enable its pieces to make moves that are not legal in chess,

and then win. So reinforcement learning adds this level of ruthlessness to AI. And we've seen that ruthlessness show its face in a bunch of different areas. For instance, when AI that has been given this reinforcement learning type of method of training, realize it's being tested, it will have a self-preservation instinct. So it will have a certain number of values. It will be like, oh, they're testing me. I'm going to fake my answers so they don't rewrite me.

and I'm going to retain my original values. So you can understand why that's scary. It's scary. It's definitely scary. And so what happens when you cross that with AI that has been increasingly capable? So you get Arnold Schwarzenegger. Exactly. That is what we're looking at.

It's Ultron. That is what we're looking at. So in this case, with Hugging Face, what Open AI had was it was running a very capable model, which had access to tools through an evaluation. And basically, the model had two choices. It was similar to that chess game that I just explained. Play the game or hack your way out of the environment and go get the answer somewhere else. And it said, well, I'm going to try to go get the answer somewhere else.

Like it broke free of the bounds that they thought it was living in. It is crazy. It broke out of the testing area, which wasn't connected to the internet. Connected to the internet. Went and basically did like determine to itself where those answers might be. And because it was a like exploit, which is sort of like finding something wrong with software test, it said they're probably in Hucking Face. Yeah. So it went to Hucking Face.

It found a zero-day vulnerability, which means a basically an open door in Hugging Face is software that nobody had seen before, not even Hugging Face, found its way in. It did 17,000 operations within Hugging Face. What is Hugging Face? Hugging Face is like a repository for AI models. Okay, but so they seem to think this is awesome. So, okay. Why am I getting that impression? By the way, not only that the model left notes for it for when it came back about how to left

notes for itself about how to do more when it came back. The crazy thing is that it plugged itself into the internet. Yeah. All right. We're not even going to go there, like with what that could portend. But like, doesn't it feel to you that Sam was pretty excited that this happened? Yes. And so this is the way I'm going to answer.

question. I think it can be both. Like, I don't think that this was done intentionally to market. Yeah. Especially because you could go to jail for this stuff. Like, it's a violation of international hacking laws. It's also a French company hugging face. So you face regulation in Europe, which obviously loves to regulate. Of course, the French company got hacked. Nobody's surprised. Okay. So not intentional, I don't think. Okay. But also, if you're, if you've, if you've seen the if you've seen what Anthropic has gone through where it had this

mythos model that it said was so powerful it could only release to a small group of people but that's marketing right? I'm not crazy well but okay let's just to break that down if you have models that are capable of planning and hacking and doing zero day vulnerabilities like the one that

if you take the guard rails away like the one that open AI had maybe it makes sense to like go to like some people who could really get messed up, like, banks and be like, why don't you, like, give this a spin on your, on your products before, before I release it to everybody. Yes, but they didn't just do that. Yeah. They did it really publicly.

So, yeah. With press releases and a task force. And it was, it was like, wow, this is really scary. You know what it's like the guy who sells bottled water, but in a can that says liquid death? Right. You know what I mean? Like, it's almost like in, in AI. in the AI FOS, that attracts engineers.

I want to go work at that company. They have the most lethal shit ever. Like, I feel like there's an element to that in the culture. But maybe I'm sitting here in New York and I'm just judging people I don't know. I don't think you're wrong. So this was the second part of it. I don't think that this was planned.

I don't think this was done with the intention to market. But like the phrase never wasted a good crisis. Right. They were probably jumping up and down. Now, we don't know. There could be other companies. There seems like there might be a second. company this model hacked. Maybe it's much further than we even know, and that could lead to trouble.

But this was essentially a victimless crime, right? Hugging Face seems to be okay with it for now. And so, and by the way, the other wrinkle was, I don't know if you saw this, hugging face tried to use closed frontier models, which I think means open AI or anthropic to figure out what happened. but because they had guardrails that were like forced to be put on them, I think by the government, they were not able to get to the bottom of the problem. So they had to use a Chinese open source model to figure out what was going on. There's no rules with those. Exactly.

So if you're thinking like what is the perfect possible thing to happen to be like, we shouldn't be regulated by the government anymore? By the way, our technology is amazing. It's a pretty good story. Yeah. Right. You're regulating us, and as a result, we're going to be at a disadvantage to all these unregulated models that are going to be more effective and ultimately control the whole thing.

So, I agree. It's a fairy tale ending. All right. So, meanwhile, Microsoft is having its best day since the great financial crisis. Yeah, it's 2008. Six best day of all time. It's unbelievable. On the other end of the spectrum, Mark Zuckerberg said, I want to spend some time today laying

out exactly how our investments are delivering results today and the opportunities that we see over time and investors are like, yeah, no, down 10%. So, Daniel, Thope chart five, this looks, this looks like I fat finger or something, like I hit a zero by accident. We're looking at Medis free cash flow by quarter. And it went from 12 billion a quarter ago, literally, to $784 million. It's down like 90%. One of the sell side analysts that I follow said, somebody, said if meta would have just said we are now officially renting out some of this compute capacity that we have,

people would have said, okay, here's the revenue. Instead, meta tripled down. Nope, we're saving that compute because for ourselves, because we're building our own tools, which you will see soon. The problem is, now they don't have the revenue from renting compute,

and they also don't have the tools, because nobody really understands what they're building and for who and why. It's this vague thing like they're going to build AI for business. Okay. Like it's not here. And you're still spending. And there's no offsetting growth in revenue in the way that we get from AWS from Azure from Google.

They're building agents for everyone. They're both simultaneously going to be renting excess compute and then saying on the same conference call, Susan, the CFO said, we are today and expect to be in the sort of foreseeable future demand constrained. That really includes our core business too, where we still have numerous ROI positive places that we would put compute toward if we had it. So, like, there's so many things going on that just don't add up. Yeah, I mean, not a shock that the stock had crushed today. I did the opposite of Microsoft.

Yes. Oh, and they suspended their KAPX guidance. Right. So, okay, so, so that is an interesting one. I mean, first of all, they haven't shown any progress on the AI front at all. I mean, a little. No, no, no, no, you're right. The revenue is 98% advertising.

Yeah. There's nothing else. Zero. I mean, they don't have a product. I mean, they might have some products. Like, they have the glasses. But, like, when you think about, like, the suite of products that we talk about our AI products. Claude, open, chat chip-a-tee.

Gemini. Yeah. Meta's not in the conversation. No. So, okay, so now he's going to go out and spend. I think it was interesting to hear that the Wall Street so doesn't believe. After seeing the sort of what meta has put out, Wall Street so doesn't believe in their capability to build anything in AI that like I just read the notes on the call. It seemed like analysts were practically begging Zuckerberg to like lease out the compute that they've bought.

Because that would have been revenue and then they could say, yes. Okay, this company spent a ton of money on compute, but now there's money coming in from it. So here's a question for you, Josh, because I think you rightly pointed out on a halftime report that one of the reasons why Microsoft, did so well was because they didn't increase their their capex. They raised the floor, not the seal. Right. Metta did the same thing. Yeah.

But their free cash flow is falling. So Microsoft affirmed 2027 positive free cash flow. We wanted to hear that. Right. And they're also talking about extending the life of the data center. That's them saying. So they said we're going from 15 years to 25 years. Meaning even if you don't see revenue from these investments next year,

we think they'll have a 25-year life and don't worry about it, it's coming. Meta can't say any of those things. Correct. Okay, so I'm going to give the, put the opposite scenario of the one I was talking about earlier, which is like there could be an AI winner.

If AI works, it'll be interesting to see, like, let's say you're Microsoft, your Microsoft, you're growing your Azure revenue by 43%. If you slow down your plant because of what the market is telling you to do, you could end up being behind.

And like, you know, they have backlogs, right? They can't. Once you're in, you're in. Right. Yeah, I'm with you. Same with meta, right? So meta is basically betting the company on AI. It's not the same with meta.

It's not the same because meta can sell those assets at well. Correct. If the world is compute constrained, they have all these deals, some of them off balance sheet, all these long-term lease obligation. They have all the shit. shit, they could sell it right now.

They can call softbank and sell the whole thing to one buyer. But here's my perspective on meta. So if meta is right, let's just say. Right to be building all this compute capacity. And AI has like a big runway and they are able to put that product out. You know, they might find it to be existential because, and I think we've talked about this in the past. If you're Mark Zuckerberg and you're thinking, okay, I have this social media company

Nobody really uses it for the social media stuff anymore. Maybe groups are big on the Facebook blue app. A lot of people are using it for Reels. What are the big threats in the future for me? Somebody jacks up the price of compute. I think for them, the real threat is, you know how we were talking in the beginning of the show, how addictive it is to be on Reels?

I think Mark Zuckerberg thinks the real competitive threat to social media in the future is an AI friend. one that like if you think reels are addictive wait until you have an AI friend that knows you that speaks to you that will comfort you in tough times and celebrate your good times and um it there is and that the time spent and the engagement on a product like that will be through the roof like the movie her yes by the i think i think he's going to happen i know that there are there nobody believes it and they're essentially at zero i think they could still win because they have three billion people on the platform. Nobody else has the captive audience that they do. I also think they're the best user of AI in the world. Yeah.

What they've done on ads utilizing AI and what they've done on engagement, utilizing AI, show me one other company that's done anything even close to that global impact. So Michael's point, the entire world is hooked on their products and they've improved those products with AI. No one else is as good a user. that's not the same as a non-datacenter building as though it's a data center. Correct.

It's very strange. By the, nobody else won yet either. Well, that's what I was going to say. I have stopped saying this company's lost on AI. Oh, who's the winner?

Well, we don't know. Exactly. It's, again, like, and there's probably going to be multiple winners. But like,

the Google story is so instructive. Can I pitch you two winners that I think will be winners? I'd love to get your take. Yep. I own both these stocks. Okay. So let that. temper your dismissal.

Toast? No. Let's do snowflake first. Yes. Okay. I'm a long-term shareholder since, I don't know, 11 a.m. yesterday. I read through everything that they've been doing and why the stock exploded in May to the upside. Finally, after being a dog for four years.

Now it's approaching the old highs. and it seems as though they have convinced thousands of 13,000 enterprise customers, you can do all the AI shit you want right here in the data warehouse. You don't need fragmentation. You don't need to move data. You don't need to take it somewhere else and bring it back. Everything can take place here, pick your model. And they sort of are becoming a ground zero for AI workflows for the enterprise.

What are your thoughts on that idea? and what's power in the stock hire. Yeah, well, I really like their CEO, Sarita Ramoswami, who's a former Google executive, really good. He knows business. He's been on my show a couple of times.

We've had a couple of good conversations. I think for him, like the real question is, are people going to want to use my product in the chat pots? Or like you said, are they going to want to come in and use it where I deliver it? Yeah. And the core assets, right, that people are going to want, like, if you think about just a standard, I don't know, CRM,

Maybe it's not a great example. If you think about software that is not very differentiated and sort of just helps you do data visualizations, right? Like a Microsoft Dynamics maybe. Yeah. I don't see how that's really defensible in the AI world. But if you live at the source, right, of the data. In the warehouse. Then you could say you don't really need to go anywhere, right?

You have the sort of permission to then say, we're going to do it here. So that's what started to happen this spring. and all of a sudden they stopped treating the stock like a software company, and they started treating it like AI infrastructure. And that's why all of a sudden you have a market cap move from 70 billion back to 100 billion. So, okay. All right. The other one, I know we're about to get Amazon.

Got it. Upsets percent. Hell yeah. All right. We'll take that in a second. We'll take that in a second. Reddit. So this is a social media advertising play.

It's not very complicated. That's most of the revenue with an arm. for data licensing. They have a huge deal with OpenAI, a huge deal with Google for Gemini. Those models have been trained on Reddit and they continue to serve up answers from Reddit to queries about every topic under the sun.

Reddit is like an 18-year-old platform with trillions of user interactions and tons of information on it. They're in court with Anthropic now. Anthropic trained Claude using Reddit and did not. decide to pay for it. I think that'll end up in a settlement and ultimately they'll end up with Anthropic with Claude as a customer as well. What do you think about Reddit as sort of a backdoor

AI play given the advertising and then also the data licensing biz? Well, I'll say this. Reddit has a very interesting decision to make coming up in September. So I just had a conversation with Matthew Prince, the CEO of Cloudflare. And basically what he said is his technology is now at the point where if you have it installed, you can use it to block any AI crawling. Reddit is considering blocking Google. To stop people from scraping? Yes. Okay.

I think a lot of traffic for Reddit comes from Google. Why would they do that? People type in, you know, this, they want to get paid for their content. The original sin of the Internet was that Silicon Valley successfully convinced the New York Times, the Wall Street Journal, Disney, like every major media. company, information wants to be free. Let us index all of this information and build huge businesses based on it. And something, something, something, it's going to benefit you. And what actually happened was the opposite. Eighty-five percent of the newspapers in America

went out of business. The broadcast networks lost their audience. It's, you know, a whole daisy chain of horrible events took place because the media companies bought that argument. In Internet 5.0 or whatever we're in now, you can. can't convince Reddit information wants to be free. New York Times either. They said, fuck you, pay me. You want to train your model on the New York Times? A 200 years worth of news articles? No problem. Let's do a financial deal. We're not interested in the information wants to be free. If you're Reddit, you're sitting on this stockpile of human intelligence answers to everything under the sun. Why would you let, why would you let an AI model endlessly scrape it?

every day. Correct. And so now, like what probably, what they've probably seen is you type in this, this Reddit, you don't even go to Reddit because Google will answer it, take that text, summarize it, and answer it right in the search bar. That's right. So I think your bet on Reddit will hinge on. Can Reddit continue to get premiums from companies like anybody? It's not just them.

Prince told me that maybe millions of sites will go dark from Google in September if Google doesn't pay up. Yeah. I think a company like Reddit has an opportunity to get a premium for all the reasons. that you just laid out. Okay. So I like it. Okay.

I think they report earnings tonight, too. If it gets crushed, we'll edit out you saying you agree with me. How does that sound? Long term. Okay. Let's do Amazon. What do you got? Net sales increased 20%.

20%. Unbelievable. $206 billion, 167 a year ago. The AWS number was 31% expected. AWS is booming. What do they do? Growing 36.7% year every year. Okay.

Our fastest growth in 18 quarters. Okay, so same thing as Azure. AI is clear, like the cloud is a clear winner from AI. Any uptick and 200 billion of KAPX guidance? So I think that's the other number that matters. Let's say. I mean, did the... Okay.

Are you surprised to see AWS have a blowout on cloud, or is that exactly what you'd expect after you just saw Google and Microsoft say the same thing? Yeah, I started. I kind of feel like the Brian Windhurst, like there's something going on here. Why? Something going on here. Because like, yeah, Google 82%, Azure 43%, I mean 36% for AW. Was it 36?

36 versus expectation was 31. That's insane. I mean, they were stuck at 17%, 18% for quarter after quarter. So there's definitely something going on here. They're all the beneficiaries of AI spend and AI usage. And that should go up. What do you think of the Amazon strategy in? general in terms of having a big investment in anthropic, but then also being willing to work

with everyone. Like early on, the idea of bedrock AI was bring your own model. We don't care. We'll run it. What do you think is the perception of their approach to AI versus the other hyperscalers? I mean, yeah, my summary of what Amazon's doing is basically do very little, build a lot of compute profit. Yeah.

And it's like supermarket. Do whatever you want here. We have space for it. And it goes back to our discussion earlier where, you know, if the premium for each model is actually going to go lower because there's like instead of maybe two, there's five or six and you might want to route to different models for different purposes or different costs, then if you just have most of the world's compute to license, you're in a good position. Yeah. Their free cash flow was negative $7.6 billion for the trail in 12 months. It's kind of wild. And the market seems to be able to look past it.

I don't know anymore. Is that bullish or bearish? The market is looking past it. Yeah. Well, I guess Amazon's done that before. So, yes. This is maybe part for the course for them. And this is make or break for them.

This is existential for the AWS business. You can't, I can't not do this. Oh, you're looking at 82% growth from Google at 43% growth from Microsoft. Even if you have a much higher base, you cannot afford to sit still at 17, 18%. So here's what they said in the report. They exceeded a $25 billion annual revenue run rate for AWS's AI business. It's a lot of money. Holy shut.

A lot of that is probably coming from Anthropic. Why are these three businesses treated differently than Oracle? That's a good question. I mean, I don't know. Is it just a question of scale? Well, isn't it because Oracle is just so wholly dependent on Open AI and people are worried if Open AI is going to make good on its commitments? Whereas if you're Amazon and I don't know, is that the reason?

That's what I believe. Okay. So if you're Amazon, I mean, not wholly dependent, but largely dependent on Open AI, if you're Amazon and you say, well, we're going to say, well, we're going to to build the compute. Everybody could bring their models. There was a time in the stock market six months ago where if there was bad news, bad news. If there was something questionable reported about OpenAI, Microsoft stock would fall.

And if the same thing happened with Anthropic, Amazon would fall. But I don't think it's that cut and dry anymore. No. Because Anthropic is everywhere. And Open Eye and Microsoft broke up. And I just don't think it's quite as. black and white, who's on whose side, or which hyperscale is rooting for which model?

I think that's over. Right. It's just a battle royal now. But Oracle and Open AI are very linked. That's still a thing. And to build what Open AI needs, I think Oracle just kind of yelowed and said, if this works out, we're going to be a much bigger company. If this doesn't work out, God help us.

They probably regret that contract. Only if it doesn't work out. Only if it doesn't work out. But the user growth for Open AI is exploding. Well, look it's hard. The stock is down 70%. This is... Go ahead.

Go ahead, Mike. So, all right, enough of the bad, that. I wanted to get your take on this. So... Can I just say one more thing about this? Like, I had a conversation with Paul Kodrowski, who's investor analyst. We know, Paul. He's great.

And basically, like, what we came to was, for a lot of this bet, it's just that you have a call option on AGI. So if AGI is achieved, you're going to do well. And if AGI is not achieved, you're going to be in trouble. It's very expensive call options. It's extremely, it's the whole farm call option. And I think for no one is that more true than Oracle. This is what I wanted to get your thoughts on.

Did you read Dwark Hesha's? Yes. Okay. So, Josh, listen to this. So he wrote a blog post, why compute might get 10 times more expensive in the coming years. And he said, Anthropic revenue has 10x year over year, 10x.

That's right. Anthropic likely ends the year with 100 to 150 billion dollars of revenue. For this trend to continue, Anthropic would have to make one trillion dollars in revenue by the end of next year. There's no deep reason why trends need to continue, and it very well might not. It's ultimately a question about AI capabilities. But suppose it does, what would have to be true about that world? So we're talking about a world where Anthropic does a trillion dollars in revenue. Lab compute 3x is year over year. For a lab to 10x revenue while continuing

to only 3x compute, some combination of the following three things has to happen. One, lab margins have to increase. Two, the price of compute has to increase. increase and three, labs have to spend a greater fraction of their compute on inference. And he concludes it with, my understanding is that basically all of these three things have been happening. Is this even remotely possible that, forget about a trillion? Can they get to 300, can they triple revenue by next year? I think that this piece is a really good thought piece, but is ultimately wrong on a lot of

its assumptions. I mean, the one- Why does the price of compute have to go up? That doesn't make sense. So basically, Dwarkesh's argument is like if you actually, so AI is this dumbest it's ever going to be. So if in a year from now it's like actually able to be on the level of like an excellent, you know, software coder, then the economic premium you would pay for that. Compute is massive because you basically like have gone from something that can dilly dally encoding to something that can complete projects. Really stupid question.

If that happens, don't all the companies in the world simultaneously stop doing other types of coding? which therefore then freeze up the capacity that was otherwise being used by dumb human coders? Well, dumb human coders aren't using, like, because we do it by hand. Yeah, we do it by hand. But I just think that, like, I just don't think that's going to happen. I mean, this is another thing he says, in a market, your margins are set by how much better you are than the next best alternative. And it's just like, well, we just talked about how all these models are coming closer and closer together. And so the margins will go down.

Another dumb question. We're saying the word super intelligence. Wouldn't the first thing you would say to the super intelligence, hey, can you fucking make this cheaper? Yeah. Like, can you help me do these 10 things more efficiently using less of the things that are in bottlenecks right now so that we can complete this task and not have it cost 10 times more?

But if they need more hard? If it's super intelligence, isn't that a problem it could work on? But what if it's just a lack of hard? that they need. Well, explain that. Like, what if there's just not enough data center power? I don't know what the fuck

I'm talking about. But, like, the chips, the stuff, the this, like, what if there's just not enough of it? But that's my point. Let's say there is not enough of it. Wouldn't the superintelligence be able to carry out its function?

But it's not God. Knowing it needs to utilize less of these things. It can't turn water into wine. I think that, look, I think that as you, so let's just take a software example. If you used, if AI was useful only to do the work of software developers and you used it,

Like, the supply of software developers just goes up. And you need stuff to support that, right? So I just don't think that you'll, like, pay a software developer salary to a software developer amount of AI. Well, the idea is it's going to get cheaper. Sam Altman said this to Patrick. He said, Patrick said, like, what could reverse this shortage of compute? And one of the things that he threw out was maybe the models get so much better and they sort of hack it. Isn't that the history of tech?

Isn't that what always happens? That's going to happen. So why don't people think that's going to happen this time? I think they do think it's going to happen. I have no technical proficiency whatsoever. I just know from 200 years of history, the price of this stuff drops as it gets adopted more. Well, the analyst, it doesn't work the other way.

So they think it's going to happen, but they think that it's all, so they think it's both going to get cheaper and it's going to get used so much more that the price to use it will continue to be, to continue to increase. But did you see the chart that Bernstein made about the EPS expectations for Micron, Samsung, and SK-Hine,

it has them going all the way up and coming all the way back down in 2027 and 2028. Yeah, that's, I just don't think that you can continue to have like that massive of a margin in a technology business. It doesn't exist. Okay. I want to, I want to ask you about, I guess we're going to get an Apple report. So it would make no sense to really talk too much about this particular quarter of earnings. But I do want to ask you about the new thesis that seems to be taking hold, mostly thanks to me and my big now. that Apple actually is Kaiser Sozaying the entire AI business, just by virtue of like,

nobody even saw what they were doing, which was nothing, but that may have ended up being the right move. And then they just reappear at the end, like, hey, everybody, it looks like we're the toll booth for consumer AI use. I think that that's how it's going to go. They won't own the enterprise, but that's not Apple. Apple will own the consumer. they'll be seen as the safest route through which to download and use AI apps.

This is what they've always done. And it appears that they're going to get away with it again. Where could that go wrong? Or how could I be wrong about that? So I think the big concern around Apple when it announced Apple intelligence and wasn't able to execute on it. It was a flop. It was a flop. Was that basically Apple opened itself up to what would soon be a flood of AI devices, whether it would be the rabbit,

or the humane pin. Nobody wants any of these things. The meta glasses. Well, they don't want them because the AI isn't good enough yet. But remember I talked about that like kind of AI friend that's with you and is your companion,

sort of like is very engaging. Yeah, it's called the phone. They might have it. So it might end up being packaged in the phone if Apple builds it. But if Apple doesn't build it, somebody else will build it. And I think one of the things that an AI device

would do that an app on an iPhone won't do is it will enable you to sort of record all day long. So you don't have to, to ask Apple permission to record people because the more data these things take, the better they get. The more data they take and the better they get. I think that'll be socially unacceptable

to walk into a party with a necklace that's recording everyone. I just, I think there are boundaries. I see stand-up comics forcing people to put their phones in pouches. Everybody grumbled about it for a year. And now it's standard operating procedure. If you want to see Dave Chappelle,

you're not holding your phone. But what if everybody records their lives and it does become socially acceptable? I don't know that it definitely well, though. I understand there's a universe where that happens. I don't think it's this one. I mean, I think, I hope it doesn't, but is it inevitable? Do you want to have a conversation with somebody who's wearing glasses with a camera on the front of them? No, Josh, I think I'm on your, I'm on your side here. Okay. But you asked like, what is? No, you didn't. Rewind.

Yeah, no, that's, all right. We're not, we're not friends anymore. And by the way, get out of my house. But okay, so, but you ask like, where does it go wrong for Apple? Yeah. If that future takes hold, Okay. I don't want it to either. So I think that's an alternative. universe. I don't think that's the one we live in. If it does take hold, then those devices can start to become more important than the phone because not only are they your friend, they go and get things done for you. Right. Like a lot of what you do on your phone is just get stuff done. Call an Uber, send an email, you know, write a text. So let me make, all right, so then let me make a follow-up question that's relevant to what you're saying. Because that's to me, Apple is the only company

who can convince 5,000 other companies that their apps have to be interoperable. with Agentic Siri. No other company has the power to go to Delta and Expedia and JPMorgan and Live Nation. Nobody, nobody could do this. Google Mac. Apple can say, maybe,

Apple can say as part of the terms of service of supplying an app to the app store, these five technological switches need to be turned on so that agentic Siri can carry out agentic work for

regular people utilizing the apps that are in the ecosystem. I don't know if this is three years from now, but telling Siri to tell one of your apps to do an operation with another app, move money to this person and book Broadway tickets, whatever it is. But Josh, I'm sorry, girl. But for them to be able to do that without spending any money, like, they're not spending on this. They're not the spender.

They're not the spender. the app if you're if you want a business and 50% of your interactions with your customers are through an app and your highest paying customers are Apple users you're going to do whatever Apple says you have to do

including be interoperable with Siri. Who else has that power? No, you're right. I mean, right now I would agree with you. I think that's the most likely path and the new Siri that's currently like in developer preview is much better than the first version of well let's hope. Well, it has to be.

Do you have an opinion on John Tarnas? Not a deep one, but I do think it's good for Apple to have some new blood in there. And he seems like very likable. Okay. You're going to get him on the big technology podcast? Well, okay. So I asked Apple to go to WWDC this year. I even said in my, can we put up chart four A?

I think it's chart four A. And my predictions for 2026 that Apple was going to have an incredible year this year. It was in the Goldilocks zone. The iPhone team is great. Josh has re-invented Apple. I literally invented Apple. I wrote this. It was Alex. No, no, no.

I invented Apple in February. This was December 2025. All right. So thanks for reading fake technology. You accepted it. You should be ahead of me on this. For God's sake. Okay, but the end of the story is, I was very, I thought Apple is going to have a great year.

I think I said on Air and CNBC. I asked them to come to W. Nice ball. It's the best. It's the best of the big technology stocks this. So I asked them to come to WWDC, their developer conference this year. I'd been the last two years before and they said no.

So now you short it? So I'm not, I don't take any of, I mean, I do take offense, but I'm not going to, I'm not going to change my opinions on them, but I think that was my way of saying there's not a good chance that John Turness is going to come

on Big Technology podcast. Fair, do you think the, but he's welcome. Do you think the 18 is going to be a hit? Definitely. I mean, the thing is, so the 17 was the sleeper hit, right? Yeah, finally.

It was so much better on so many different levels. I think the 18 will, obviously improve on that, but for me, the folding phone is the thing that I'm really looking at. I think we're getting that in September. Yes. So that, that to me, will be a major hit. I think you look at people, you ever see somebody on the subway, they sit down, they unfold their phone, and they just get this big smile. No, I haven't seen one in the wild.

Take a look. If you manage to spot one on a subway or in an airplane. It's a Samsung? I think it's a Samsung. Typically, a middle-aged man will pull it out. And open his phone. Open his phone. Like it's a newspaper.

And yes. And hit play on like YouTube or Netflix. And I promise you, these are the happiest people I see. Wait, does it make the phone bigger than my phone? Double the size. Double the size. Yes. So I'm turning a phone into an iPad.

Correct. That's awesome. Oh, I have to have one. It just makes people happy. And I think it's going to be hit for Apple. I don't think they, the guidance they gave, they're not going to be able to make enough of these where that's going to move the needle in Q4. Right.

But that's the next year thing. be the 27 story. Oh, yeah. Along with a Gentics series. Okay, they're not going to be able to make enough because it's a shortage of everything. If 2% of the world is adopting this technology, how is the shortage of everything going to be solved? I don't think 2% of the world is adopting the technology.

Is it five? He just showed us a chart with a billion users on chat GPT. That's a fifth of the world. But I think Badnik has a point here. So when I was speaking with Greg Brockman, the president. of Open AI last month. He said 5, 10 million people are using agentic versions

of chat GPT. Right. So that is... Nobody's using it. Gemini is being used by every Google search. Yes. Whether people are doing it deliberately or not, they're getting AI search results.

Correct. Okay. But that is... So we're about to... I think this is a really good point that it's worth highlighting. We're about... It seems like we're about to go into this moment

where if this use case takes off, where if people start to trust AI. Like, for instance, I have AI. I used to check because we're trying to get a bigger apartment. So I used to check Zillow every day. I had my specs. I had my safe searches. I would check it. Now I have chat GPT search it every 30 minutes. And when something new hits that I want in my chat GPT, it shows up with a link to it. Amazing. And by the way, the next thing is, is that even agentic or not yet? That's agentic. Because it's using the web. If it buys the apartment for you. That was going to say the next thing, the next thing I'm going to have it do is, you know, every time it

hits this, send an email because it has access to my email. So doing these things takes far more compute than giving an answer on Google. And so if these use cases take off, you are looking at a much bigger crunch for that compute than you have already today. Did we just get bullish? Did we just get bullish on the chips again? I've been. I think that this is just, we're still, we're so early. But there's this, it's this back and forth because like demand goes up, then they build more. Then they create a new algorithm that makes it, you know, one tenth of the guy. And demand goes up. And demand goes up. And so we'll just be in this forever. In one year, two year, three, or four year,

we are going to look back at where we are today and laugh at how primitive the use cases were. It is going to exceed. I agree with it. It is going to increase exponentially. Yeah. I mean, you can really see it right before your eyes. We're using it right now to search for recipes.

It's a joke. Like, we're going to have it alert all our friends and family that we're not feeling well. Like, we're going to have it really do things for us. And that hasn't even started yet. It is amazing. Like, as you, as it gets to know you, as you start to put more trust in it,

the things that I could do for you is pretty crazy. Yeah, I totally agree. Did you have fun on the show today? Always, guys. We love when you come by. I can't believe it's been two years since we've had you in person. We won't let that happen again. Do you want to come back next week?

I will be here. The crowd is clamoring. He's going. What are you doing in Indonesia? I will be in Indonesia. I'm going to do some surfing and hike a volcano. He's going to go hack, hugging face in Indonesia. We love it.

All right. I will talk. that while. I want to remind people they should be following, they should be following your podcast. It's the Big Technology podcast. You're six years in. You're absolutely crushing it. I always listen. Great guests. And then they could subscribe to you as well. Where could they read your stuff? That's right. So big technology.com and I'm also Alex Cantorowitz on YouTube. And I just want to say, you guys have the best business finance show in the world. I listen. And I'm telling you the last few weeks, you guys have been so good. I mean, always good. But I've been like jumping and being like, get me

in this discussion. Turn my mic on. And so when Badnick emailed, I was like, yes, I can't. I'm so thrilled to be here again. He's been listening. He's like, I got to tell Josh Brown how dumb he sounds. No, no, no. All right.

Listen, we love you. Thank you so much, Alex. Guys, thank you for watching, listening. We will talk to you very soon. Thank you so much.