The AI Revolution Call

Yo, what's up? It's the AI Revolution. I'm Cody Willard. Bryce Smith. John D. Oliveira. Bryce and I run a hedge fund. We invest in technology mostly we're Revolution investors. Specifically Value Revolution investors. According to my article that I wrote today. Go to trade with Cody.com and or email supportrain with Cody.com and we'll send you the article for free. Why not? Johnny D been working in AI for 30 years. Yeah, I said that, right? Check the math. It works. All right, topics for today. Oh, there you go. Yeah. Support train with Cody.com. yeah, send it to that if I'll send you the article. Yeah. And then we won't spam you a bunch. Maybe occasionally. Bryce. It's been like a year since we sent an email out. I was thinking about that this weekend we have like a list of tens of thousands of people who have like requested to get an occasional email from us about investing in stuff and we never send it. It's totally.

I have a list here. We're bringing out a new employee this week and I've started.

Is that on the list?

List of things. I'm adding it. I had thought about, but I'm going to put that on his list.

Put it on the list.

Yeah.

All right. Anyway, this is the AI revolution call and I don't know if you heard, but AI is like happening. Like it's. It's actually funny. Bryce and I were around the world. We got back, I don't know, 10 days ago at this point, but we spent 19 days going around the world and it was one of those things. After talking to, what do you think? 45, 50 different institutional investors, family offices, high net worth individuals and. Or random really not many random meetings. It was mostly the first three things I said there. And I think most people recognize that the AI revolution is happening. I think most people are trying to figure out how to invest in the AI revolution right now, if they're not already. It might be because they don't believe it's a good idea to invest at these valuations or something can. Even if they believe that it could be happening. They worried about a bubble in AI, which A fair discussion, we'll have that here in a bit. But everybody is sort of like, I get it, they. Not many people are thinking about the robotics revolution yet and few if any people are excited about the space revolution, which I think is the third leg of the triumvirate revolutions that we're living through right now. But. Or not even living through yet. I mean really, the robotics Revolution is like the first if the AI revolution's 10, 15 of the way through. I realized in Abu Dhabi, Dubai and Hong Kong that not everybody got my baseball references. So I'm not. Even though John D. Oliveira's brother in law is one of the greatest baseball players, he's a Hall of Fame. Let's just say Alex Bregman will be a Hall of Famer.

It's possible.

I would say it's very likely. Yeah. So that's his brother in law.

Nephew. He said.

Yeah, whatever.

It's my nephew. It's my sister's boy.

Yeah. Oh, it's actually. Oh, it's his nephew. He's literally blood. Yeah, I, I mess this up all the time. I, I've got generations off on this because I'm still Picturing us at 30, John. We're 50 something. So at any rate, not everybody gets my baseball analogy. So I'll speak in percentage terms. If the AI revolution is maybe 10 into the maybe 15% started here, the robotics revolution is like less than 1% and the space revolution is less than 1/10 of 1%. Just barely an idea in someone's mind at this point. But all three of those revolutions are going to be worth tens of trillions of dollars in economic value and prosperity enhancement and what they create for the world and eventually the solar system and maybe even eventually the galaxies. So the AI revolution is here, front and center right now. It's run through a few headlines, meaningful things, and then we're actually going to talk a little bit about is it a bubble? Bryce and I have some insights that we've been building on a model, actually trying to put some numbers instead of just in the air pontificating. I used to get paid to do. Bryce, your first topic, I think the.

News of the day is the AMD OpenAI deal. And so that should be the starting point. And I'll just give you the basics of the deal. So AMD and OpenAI announced form of strategic partnership where OpenAI is going to build like 6 gigawatts of capacity, which is a huge amount of electric. We measure data centers now and the amount of electricity that they use, 6 gigawatts is like three nuclear reactors or something worth of electricity. And they're going to buy AMD chips to, to do that. And in exchange AMD is going to give a OpenAI warrants to purchase the stock for one penny a share. So basically giving them the stock over the next five years, as long as OpenAI buys the chips that they say they will buy and AMD's stock hits certain targets. So I think at the end, AMD stock has to be like $600 per share. Right now it's 250 or something like that. After today's move, I don't remember where it closed at. But that's, that's the gist of the deal. And it's one more deal that in the line of sort of circular revenue moving around between the chip companies like Nvidia and AMD and the Labs OpenAI.

And so say that again, it's hey, AMD. No, AMD says, hey, OpenAI, if you buy our chips, then you can buy our stock.

Well, we will give you our stock.

Yeah, it's basically give.

So it's, it's somewhat. You can see who's got the leverage in the industry based on even the. In the last three weeks, OpenAI has done a deal with both AMD and Nvidia. When they did the deal with Nvidia two or three weeks ago, Nvidia got equity in Open AI and gave cash to Open AI to help them build buy Nvidia chips. In this deal, AMD gives, not cash, but they actually give equity of their company to open AI. And then open AI can eventually sell that equity, $100 million at a pop, in order to buy AMD chips with the proceeds. And so OpenAI needs Nvidia more than Nvidia needs open AI. But AMD needs OpenAI more than in OpenAI needs AMD. 1.

Not to complicate it, but how does. What is Nvidia getting in terms of a piece of Open AI if Open AI is not fungible? You know how it's.

John, they're, they're not wor. Nvidia is not worried about trying to sell whatever they get in equity from Open AI anytime soon. The assumption is for everyone that over the next three, five, seven years, that Open AI figures out how to pay off Elon Musk for his equity stake. They figure out how to negotiate with Microsoft for their heretofore equity stake. And they also deliver the equity stakes that they've promised to Nvidia and other investors along the way and that there'll be a publicly traded vehicle at some point in the next, say, two to five years, and then at that point it would be worth trillions of dollars would be the expectation. And at that point, if Nvidia so chose, they could start selling down their equity stake, which would probably be worth hundreds of billions of dollars. If all that works out the way that everybody hopes it does. Except for the Bears and Shorts, of course they don't.

And then how does Stargate figure in, in terms of, you know, they're off building these data centers? Is it those data centers have Nvidia chips or they have both or what?

Both.

They would have both. Initially they'll have Nvidia chips and Oracle is a major partner in building those, financing them and helping run them and using Oracle software inside of them. But yes, I mean the idea would be that initially it's probably Nvidia chips almost exclusively. As they get ramped up already right now they're turning on capacity in Abilene, Texas. OpenAI is/. Oracle and others are turning on some of that Stargate stuff right now. And eventually it's even this AMD deal. Those chips that they're talking about in this deal are not supposed to be delivered even for another year. It's second half 2026 before any of those chips are even delivered in the best case scenario. So you know, some of this stuff is pie in the sky. I mean look, we're, we're revolution investors and we're big believers in the AI revolution. But let's be clear, there is some bubbliciousness in some of this stuff and there is some circular dealings in some of this stuff. It's very different from Nortel and Lucent in the year 2000, 2002. 2 actually, I think that was 2003 when it all came to light. But from 2000 to 2003 Nortel and Lucent were secretly vendor financing their biggest customers and Cisco to a lesser extent. Cisco is a lot more transparent about it. But Nortel and Lucent were hiding that and it was fraudulent. And this is not fraudulent. You can dislike it if you want. It is a bit circular to have your suppliers investing in your their biggest customers who then use that money to buy more of the chips from the suppliers and all of that. But it's not fraudulent. They're out in the open. We all get to make our big boy big boy and big girl choices big they them choices on whatever you choose for to do with the knowledge that it is a bit circular, but it is going to be measured in trillions of dollars in coming years and eventually trillions of dollars.

I forgot how, how much into the telecoms you used to be. Yeah, yeah, I had a, a box of your textbooks in my garage for like I don't know how many years back in the day.

That's fascinating. I forgot that that had all those.

Sorry, I lost track.

Well, it does it does underscore that I do immerse myself in any revolutionary technologies that I'm studying, including 25 years ago when the telecom revolution was present and booming.

So, Cody, for the people on the call, what is your view on AMD after this deal? Your view on Nvidia after this deal? Your view on open AI after the still?

All right, Nvidia number one. Look, I've been in Nvidia since 2016 and it's been, it's frankly been the best investment, including Bitcoin in 2013 and Apple in 2002, at least on an annualized basis. And I still believe in it very much. It's one, you know, it's not our biggest. It's probably a top five position in both of our funds, but certainly not our biggest positions, not our, you know, top three or four even. But obviously I'm still a believer in it. I think you still just, you know, you let you want to invest in the most dominant platforms and the most revolutionary technologies and there's no better example of that right now than Nvidia. I wouldn't necessarily like scream into it and buy every call option I could right now, but I also can think of worse ideas, frankly.

Before you go further. Well, while you're on a video there and you said, was it 2017, you said 2016, 2016. What was it you saw at that time that made you go, I already.

Know this is, I didn't know I made a bet, an investment and have ridden it since, but it was in large part there. Move into turning GPUs into platforms for autonomy, car autonomy, robotics and AI. And then certainly Jensen Huang's brilliance, one of our biggest tenants is always betting on brilliance. And I did recognize Jensen Wong was a genius back then. And so those would be the real factors that got me into it. And I, you know, started small, but over the last eight, nine years now it's like a 220X or something. And it's, you know, I, so I, I hold it like I continue, it's my biggest position, my personal account actually at this point just because again the gains have been so big and I've been trimming over and over. Never should have because it's at all time highs. But I, I trimmed when it was probably up 30, I turned when it doubled, I trimmed when it was up 5x. I turned when it was up trimmed when it was 15x20x25x100x150x200x220x. And I still own a bunch and now it's my biggest position. Even with all those trims amd, I like it less. It's not the dominant platform. They certainly have done something well to get in to the Open AI plat into open AI's own platform of software and LLMs and all the magic that goes into making Open AI Open AI. And that's a good thing. I, I was just speaking to one of my good friends who bought AMD at 96 months ago and still owns it. And his quote, I'll quote him that today's move wasn't big enough. Didn't underscore how big this actually is. I'm not chasing it, I'm not in it. I actually, I don't even have FOMO about this one. I'd much rather own Nvidia. But if AMD does actually sell tens of billions of chips to Open AI, it's likely they'll also be selling tens of billions of dollars worth of chips to other hyperscalers, even Core Weave eventually, who knows. So that there would be upside from here. If that's true, I think there's better ways to play the AI revolution than amd.

Yeah, there's about ways to play it is like if you wanted to be in OpenAI but you can't, what would you be in that is kind of rides things similarly.

It's a great question and I could. At the end of last week's call we talked about the reason why we're in EchoStar Sats and it's because they when the deal closes in the next year and a half or so, they will have $10 billion of SpaceX equity as part of their company and the entire market cap of EchoStar right now is like $22 billion. So SpaceX is 40 or 50% of the actual value of the stock right now. OpenAI as a percentage of Nvidia is meaningless. It doesn't move the needle for the, you know, Nvidia's market cap's $4.5 trillion and so there's, you know, they own, I don't know, 5% of open AI. You think price.

Yeah, I think they own $5 billion worth of it or something.

So you know, 5 billion out of 4.5 trillion is a 50 move in the stock. So there are other ways to play Open AI in the public markets, including a closed end fund called DXYZ. But it's trad at what's it probably now 8 or 10 times book value. I haven't looked at the stock in a long time, Bryce.

I I think it's only about five times book value, but only five times.

So instead of paying 500 billion for your open AI in DXYZ, you're paying 2.5 trillion. So I, I don't think that sounds like a good bet either.

I think core weave is the best way to.

Yeah, core weaves like as open AI does. Core weave will probably do also, but there's cor. Weave comes with its own risks. It needs access to capital it ever, you know, for many years to come at reasonable prices. And that's one of the reasons, you know, why the hyperscalers like Google, Amazon and Microsoft, who heretofore have been funding their AI revolution, investments with cash flow and cash on the balance sheet and very little debt versus core weave needing to borrow and even do IPOs and secondaries to get access to more capital. There's a difference in the risk, but there's also a difference in the potential upside, obviously. So anyway. And finally. Or did I hit all three? I don't want. I. I'm not a believer in open AI. I think Sam Altman is a scumbag. I want nothing to do with him. I would never invest in anything he ever touches. I don't like him at all. So I think OpenAI is a great platform and there's no doubting, no arguing that Sam Altman has changed the world with OpenAI. It is a different world today than it was three years ago and in large part because of what OpenAI started and then it's ability to drive this thing forward with those animal spirits and the access to capital and the magic that happened after the first chat GPT moment and that has forced all of these companies to spend now hundreds of billions of dollars each. And that's made the economy better and it's making AI better and it's a flywheel of prosperity. That's happened in large part because of Sam Altman, but that doesn't mean I would want to ever be on his team.

Yeah, that's really interesting because he's so. It is so central. They are really central to a lot of what's happening that not the only players, you know, there are others that, that I think matter as well. So central and yet maybe just not even trustworthy in a way where you'd want to directly be involved. Right?

Yeah, that's right. I mean, that's it. You don't have to. It's. My mom told me when I was 20 and I was hanging out with John D. Oliveira and she was like, you know, Cody, you know what your number one superpower is. And I was like, what she said, surrounding yourself with good people. And to this day I, you know, in investments, I, I invest. You can argue if Elon Musk is good, if you're in a partisan viewpoint or not, but you know what he's done building businesses and being pretty straight up with his investors, and what, what he's delivered over the years is nothing less than amazing. Also changing the world. I mean, what a different world it would be today if not for Tesla creating the battery revolution and eventually the robotics revolution with humanoid optimus robots. And what a world it would be without SpaceX getting people to the space station. People would still be paying billions of dollars to Russian, to the. Russia, the country to send human beings to space.

I don't have enough history knowledge to say, but who would you like if you had to think of like, it's kind of like Edison or is it like.

It's like nothing else.

What's up?

Elon Musk is an enigma.

Yeah, I was gonna say like he's a combination of a couple of them, Edison and Rockefeller or something. It's like multiple industries, but being an inventor as well.

And that's right. Yeah. And, and, and not just. What's amazing with him is even just like the Starlink stuff where it. Other companies had tried before to become a constellation for broadband Internet and they needed their own rocket ships to be able to send that stuff up at diminish de minimis price compared to the revenue it would generate. And only Elon got it there. But once he gets it there, he's like, the way he's handled even his Spectrum rights over the years has been brilliant and not. He didn't spend a dime on Spectrum until he made this deal with echostar. And somehow he was able to deliver broadband Internet to my house for the last three or four years without having purchased spectrum rights, which otherwise are worth tens of billions of dollars typically. And so he just, he is a great businessman. I mean, I'm, I'm no Elon, I'm no blind Elon fanboy, but there is no doubting the revolutions he's driven. Even open AI. It wouldn't exist were it not for Elon's money and brain.

Talk about like thinking big and thinking long term. There's, there's nothing like it to be able to go, you know, I'm going to do this thing and it won't happen for 15 or 20 years, but, you know, with the rockets or with the cars, it's like and then it actually does happen and it's like I feel cons, I, I worry that he, he's so. Such an enigma. He's so, so one of a kind. Because what we could do in our economy if only we had some more, right?

I mean it is, it would be amazing if there were a hundred Elon Musks in the world. And look, to be fair, that's one of those things I think about a lot where it's like, you know, Mozart only existed because his father was a musician and he was in the right time period to get funding from princes to write incredible music that could be played in the public and then passed down through the deck centuries. And you know, like there are probably people that could have and are like Elon in their brilliance but you know, if they're sitting there in Central Africa struggling to avoid violence with their time instead of channeling it into creating inventions for the world, we would never know. So hopefully as the world gets ever safer and we bring hundreds of millions of more people constantly into upward mobility, middle class and wealthy places, you know, there will be more Elon Musk's in the future.

I have an example of that that proves the point. I think there's a guy named John von Neumann that a lot of people haven't heard of who was basically on a level with Einstein in terms of brilliant. You know, he's involved in everything, game theory, inventing the aspects of the computer architecture. He was involved with the bomb as well, the Manhattan Project, but all these different things. Mathematician, genius. And three of his, I think it's three of his classmates from like elementary school won the Nobel Prize. And so this is, this is from, they're from Hungary and there was this one area where just socioeconomically it was middle class, it was a ton of kind of emphasis on education but also the right kind of freedom of thought. And these people came out of it, it's like they're not related to each other. So you know what, what happened to enable this to come out of just this place. So yeah, it argues that this could happen with them.

Yeah, the old, you know, their nature is a large part of it and nurture is part of it too. You just can't get away from that paradigm.

Yeah.

Bryce and I have been working on a model. Bryce, did you have one more talk before I get to the model?

I was just going to talk about OpenAI's Dev Day that they did earlier today.

All you, Bryce.

Sure. So they did a big developers day, invited a bunch of software developers out to California and rolled out some new features that you're going to start seeing if you're a ChatGPT user. But probably the most notable of the features was the integration of apps into the directly into the Chat GBT interface. And so they partnered with about, I don't know, they launched with maybe 40 apps that are already available on Chad GBT, things like consumer apps, Spotify, Zillow, things like that. And also, you know, specific things for developers that they might use. Canva, figma. I'm trying to think a few other ones. But anyways, so it's, it's, there's been this big debate in the software industry about what's going to end up happening to the app companies because it's so easy.

Software as a service companies, sure.

And the. You took me off a track. So the software companies have been trading down, not trading very well because everyone's worried that the AI is just going to replace all software. And today I think just kind of showed the market and the world that there's still a lot of value add in the application layer above the LLM layer. And OpenAI and the other AI companies aren't going to try to replace every single application from scratch. And there's actually a lot of value to just integrating with all of the existing apps that are out there and the existing workflows and existing features that OpenAI doesn't need to go reinvent the wheel again. And so it's pretty cool. You can, you know, you can basically control these apps through the Chat GPT interface. So if you're like starting a business, they gave an example and he is, you know, something silly like a dog walking business. And it, you know, so Chat GBT makes your business plan okay. Then it integrates with figma. You can make all your designs, your logos, brochures, your pitch deck and then you can search. You're like, okay, chat, what should we, what would be a good business city to expand into? And it's like, oh, you should expand to Pittsburgh. It's like, okay, show me some real estate in Pittsburgh for an office house. So then it opens. Zillow automatically finds the house. Find me ones with three bedrooms that are near a park, dog park. Okay, here's this one, right? And so you can just see that you basically stay within the ChatGPT ecosystem and you can be pulling in dozens of different apps that might be relevant to what you're working on. So I think it's a pretty exciting development in software and it's Going to be like the App Store.

So any.

They published developer guidelines and if your application complies with those developer guidelines, then you'll be approved to go on the ChatGPT app store. And they released a software development kit, an SDK so that if you have a existing app you just integrate that SDK and you should be able to get on the ChatGPT app store. So it's going to be another. Their goal here obviously is to become an ecosystem like Apple or Android and have, you know, millions of apps that are built and integral to their platform and then can be distributed to 800 million chat GPT users.

And then you just talk to it. You're like, hey GPT, go through my Salesforce app and tell me the three people I need to talk to today and send me an email and send them an email with what they need to know. And it happens.

Yeah, I question whether how much talking is going to be long term the interface, because there's still a lot more bandwidth to seeing stuff out and maybe that's where the apps do retain some value is that you talk about. There are particular workflows that go with particular kind of work and there's thought that's gone into what to. To show me. So I'm not sure exactly how that plays out. But like it. It is clear that open AI has concluded that money is in eyeballs. I mean it can't just go feed other people their, their kind of knowledge, you know, in. Through other people's apps. It's like they want the, you know.

They want to be the interface. Yeah.

They want the actual individual eyeballs. I don't get it. The interface. Ears versus eyes. I get it. The idea of that it's the users, the, the mass of use, the billion users is directly to them, not people in. They do the in between thing as well. But they want that direct to them thing happening. They want to be in the.

That's a great point, John. It's interesting from the other perspective that just, you know, I, I felt like it changed. I, I forever have been saying that no matter how smart I am, I'm not smart enough to know all the different applications and services people are going to come up with on the AI platforms. But this made me realize I'm also not smart enough to realize all the different ways that AI and AI agents will be integrated with the world we currently already live in. Yeah. And so the existing services from Zillow to Figma to CRM suddenly have the ability to expand their total addressable market in a much simpler way, a much more direct way. They don't need you to ever download CRM, but you sign up for CRM through OpenAI, and whether you're looking at it visually and. Or talking to it and listening to it, you'll be able to do both. And you'll be like, hey, pull up my CRM dashboard. Oh, yeah, look at all that stuff over there. Yeah, we figure out a system that would address that bottleneck in my cr, in my handling of this customer that we're trying to, you know, the customer flow that we're dealing with here. And it just, it becomes the interface. But it also, like what Bryce is saying, it doesn't supplant the existing. It's not that you're going to talk to CRM and, hey, make me a software program AI app that does what CRM does, rather just use CRM and let me integrate with it through OpenAI and the AI Intelligent Layer on top of it.

It could be they become the new App Store, the new marketplace, the new browser, the new operating system, the place you do things which, like, you're saying the. The stuff that gets pulled, since it's like a marketplace, the stuff that gets pulled into it, other people made, but because it's so smoothly integrated, I don't. The user even doesn't think about going off and doing whatever they're like, don't worry about it. I'm just paying my, you know, $150 a month now to OpenAI, because I'm not paying others. And OpenAI is determining, like Netflix, you know, which show is getting which piece of that money based on who's watching what it's like, who's using what app, you know, let it happen smoothly kind of thing. That would be huge.

Yeah. And it's. You underscore the fact that Bryce has been saying for a year that it might be Microsoft that's in the most trouble, because the operating system level becomes ever less important if you can interface with OpenAI on a meta Glasses on a smartphone and pull up your CRM dashboard instead of doing it on a Windows PC, which you almost were required to until now. And if you're right, it replaces several layers from the operating system to the app stores and etc.

I never liked the apps stores from a standpoint of how disconnected the data was. You know, I was a data guy. So it's like every time you use one of these is its own little island holding onto its own little data. And, you know, you get a hundred of Them, they're great in terms of like variety and handling all different, very specific, you know, tasks and things you might want. But then they also separate. But now they can sort of come back together in that respect.

Yeah, that's interesting John, because a lot of people are, I think the driving force behind the strategy of both Google and Apple in their app stores to keep that data at least somewhat segregated was privacy. But in this day and age as privacy just erodes constantly anyway, people are just like, ah, whatever, share your, share the CRM data with the Zillow data. I don't care, just I want my say, I want my answers quickly and easily as possible.

The Gen Z programmers I work with say privacy is for old people. So one of the things that clearly going it goes away more and more over time.

Yeah, another interesting thing that I saw that is sort of novel but that they're going to start suggesting apps. So as apps get built onto the platform, you know, someday there'll be thousands of apps on chat gbt and if you're working in there and talking to it and something comes up and says hey, you might want to integrate or use this app that's out there and obviously you know, there can be paid promotions but then you know there's going to be organic cases where it just suggests an app that should be used. And so there is actually some unique distribution possibilities there that if you have a novel app, people can find you because they're going to be doing things in chat GBT and chat GPD might suggest your app if you're the best for that use case. I know if you, if you ever look like I don't spend a lot of time in the Google Play Store or the App Store when I had an iPhone and you know, they're suggesting apps on there and they're trying to advertise on there, but you're just usually when you go to the App Store you know the app you're looking for. Whereas in chat now it's going to be suggesting apps for you to use that maybe you didn't even know existed.

Even with GPTs before the apps happen. But like you guys went and made a GPT, you know, you're able to go and you know, have them come to you and find that was like, you know, making an app. But it's just keep getting better. Yeah. Hey, before we go to the main event, the other thing that happened was OpenAI came out with Sora 2 and it is about as far ahead of Sora 1 as you might expect. With a year and a half having passed, it's amazing. There's still some things it's not got, but it's got lip synced actual voice and sound behind things and.

Really good.

It's, it's like. And so now they're going in. The question is whether they even go kind of impinge on the, the tick tock territory. You know, people getting on open AI as the place that they go and say I'm going to whip together this, this thing and instead of it being me, like dancing around for five minutes, whatever, it's, I'm going to make a, a video that's of way higher quality than I, than I could as an individual just by saying what it is I want. And they, at least so far they were very, they threw out the whole concern about copyright. It was like, you know, make it about whatever, you know, SpongeBob and you know, anything today.

John, they, they, they now the copyright holders have to opt in in order to like.

Yeah, I didn't expect that to stick around but that, that was like kind of the, the first step. But anyway, I told my brother who's a Hollywood producer, I told him, you know, a year and a half ago, Hollywood's gone in 10 years. And he said that's good that I'll be retired.

Like it's another just indication of how much this stuff can advance and how quickly it can do it and how we're still though at the dial up Internet phase of the AI revolution because you still can't, you know, eventually you're, it's going to get to a point where you can prompt it to make you a 30 minute episode with music, voice actors, special effects, everything you want and it'll do that. But just now we're still at these like how long, what's the longest you can make a Sora clip?

15 seconds maybe. It, yeah, it's. Other ones do like, and it's just.

Compute like the, the ability to make a 30 minute movie exists, but it would just cost so much on the compute side. And that's why there's, you know, as the AI revolution continues to innovate and follow Moore's law and other laws of ever more efficiency, that stuff will come.

Well, Pixar used to be, you know, they'd have those, try to remember Silicon Graphics workstations, very high end computers that they would leave like running overnight to do a couple of frames.

Right. You know, to render two frames.

Yeah. And then it'd be months later they'd go and piece it Together and they'd have this amazing, you know, Toy Story, but.

And it would literally cost them 100, $150 billion in time, power and compute to get those movies made a little.

Lower money wise, but no, come on.

Those movies are $100 billion, dude.

150 million, not 150 billion.

Oh, yeah, sorry, yeah. In order of magnitude less. 100 million. 150 million.

Yeah, yeah. So it's, but it's, yeah, it's, it's. The stuff is advancing very rapidly. But as you say, we're still at the beginning. Anyway, go ahead with the, the model.

All right, here's where we're at. Existential. Whether you know it or not, if you invest in stocks, you're long the AI revolution and you need the AI revolution to work out from here seamlessly, rather seamlessly, if you don't want another downturn, maybe even a crash at some point, the Bears, the Shorts, the AI haters will tell you that there's no way that the Cap X that is being spent, the money that is being spent building data centers that run the AI stuff, they say there's no way there's going to be enough revenue to justify the hundreds of billions of dollars already being spent building this stuff, much less the trillion dollars plus expected over the next five years. And that's the question. That is the question. If the Bears and Shorts are right, there's a bubble, the market needs its ass kicked. There's going to be some brutal downturns here. Economy is going to get ugly, markets are going to get ugly. Access to capital for these AI companies is going to get difficult and more expensive. Nvidia is going lower. AMD is going lower. Avago's going lower, Google's going lower. If you own the S P500, you're in the stuff. If you're a dividend investor, maybe you've got a little bit of okayness, but even that, the markets will come down, PEs will come down, things will contract, and that's how this stuff plays out. If it doesn't work out right, and that's a possibility. On the other hand, instead of just shooting from the hip and pontificating into thin air with no semblance of triangulating some reality, you don't have any answers. It's anybody's guess. So what Bryce has spent today doing, he came up with this framework that he should look at the last five years spend on CapEx, building the AI data centers, etc. We'll do a rolling total for that into the Future as he builds his model, we want to know what is the impact on GDP and productivity? We're not talking about the impact on the capex and spend on gdp, but how much are the services and applications that are being built on AI impacting productivity in corporate America, in the economy and retail world, et cetera. And then we want to know what is the actual revenue being generated as a percentage of that five year rolling total? Bryce, did I say it all properly?

Right. And so we're just trying to see if there's some kind of a, if the returns make sense, if there is enough return to justify the spend and whether it's going to justify continued growth in capex spending. Because a lot of people are just looking at, well, OpenAI is going to generate $12 billion in revenue this year. That's not enough to justify trillions of dollars investment. They're going to lose $150 billion over the next five years. But it's not just OpenAI, it's a much larger ecosystem of AI companies. The cloud companies themselves are generating revenue independent of the AI companies. And if the returns or you know, exceed the cost of capital by sufficient margin, then this train rolls on and nobody is putting pen to paper on that. Basically. I think that the arguments basically boil down to people think AI is productive and helps people actually do their jobs better, more efficiently, or it doesn't. It's this joke, it's a, it's a fad, it's only good for making stupid videos, you know, and it will never justify trillions of dollars in annual capital expenditures. So that's the, that's kind of the framework. The other thing is just that this, the amount of spending is so big by any metric you want to look at historically, even compared to recent, you know, infrastructure build outs like the fiber build out in the 2000s and 90s, these numbers are huge. I mean, $400 billion this year by seven companies is unheard of.

And that's how much they're spending on making AI.

Yeah, that's how much they're spending. And the bears love to say, well look, this is an inordinate amount of money, can never justify this. And one of my points is like, look, the, the scale of it is actually a testament to the usefulness of the product. This isn't like a sideshow, this isn't something like Google decided to do one day and they're like, hey, we're going to try this AI thing. Every tech company on the planet is working on this. The biggest tech companies are investing maybe 80 to 100% of their cash flow this year in capital expenditure, something they've never done before. They're all in. And so if it doesn't work out for them, their stocks are getting cut in half, down 70% maybe in some cases. And so these guys are betting it all. And I think that's actually a bullish thing that the smartest people in the world are investing their own money and all of their own money on this new technology. It's not just some sideshow. It's, it's, it's the real deal.

Importantly, other than open AI and core weave, I mean, look, there's certainly other people's money being spent. There is IPO money, there's venture capital money, there's other corporations money like Nvidia giving money to open AI to spend. There is other people's money being spent building the infrastructure. We don't want to pretend otherwise, but the vast majority of the spend, probably 70 to 80% of the spend and building this infrastructure is from the six or seven most sophisticated, most valuable, most profitable, most technologically advanced companies this planet has ever known. And that says something. That doesn't mean they'll be right. It all could be wrong if it, if agents don't get, getting better, if you don't end up being able to make a 30 minute movie off of one prompt, if the coding doesn't just eventually solve itself, if design software doesn't keep exponentially advancing, there's risks. That's the bet. I mean, that's your, that's our job. I mean that's what we do for a living. And so we immerse ourselves in it, we learn it, we study it, we put numbers into it, we build models, we build algorithms, we build analysis and homework and we do more and more and then we do more. And then just like my dog Jim, the golden retriever, when Lori asked me how many pets does Jim actually want? And my answer was more. How much work does Bryce and I do, Bryce and I need to do to be as good as investors in the AI robotics and space revolutions as possible. The answer is that's all we can do more.

You were talking about these. All the companies are going all in on AI, but then an AI in order to do what? So they do all whatever companies have been doing. But I was saying somebody talking about how, you know, a few years back the investment was in every stupid thing like, oh, another delivery company or whatever thing it might be. And now there's a lot more focus on energy other Kinds of infrastructure, things that, you know, are needed to make all the stuff work. So what about that aspect where, yes, they're all in on AI. Are they making the right kind of decisions about what they're using it to do?

Probably. I mean, the marketplace will sort of triangulate that out on its own, and the services and applications that grow will require ever more compute and get ever more customers, and their own individual respective flywheels and ecosystems grow upon themselves. The idea is that people save time. Workers, employees, white collar secretaries, and eventually in the robotics revolution, blue collar manual labor, etc. That people save time, they are more productive, they do more in an hour than they used to do, and then they have extra time to do more stuff more productively. And that feeds upon itself. And the gross margins at the S P 500 continue to expand, just like they have been based on the Internet revolution and the smartphone revolution, the app revolution, and all the stuff that they. Now you can send you a text message that your flight's delayed so you don't go to the airport at the wrong time. All of the stuff feeds on itself. And if you get more productive, you're willing to spend more money to get yet more productive. And then your competitors who are not investing in AI are not getting more productive. And then they start falling behind. And so then they have to go, oh, crap, we've got a prisoner's dilemma. We either lose because our guy, our competition, is getting ever more productive in handling their customer service, their supply chain, bettering their businesses in every way, retaining customers, you name it, whatever. AI does the magic inside there. People learn more, answer questions faster. Everything that happens, it all happens. And then the guy who's not doing is like, oh, I better start spending too. And then they have to start spending and they ramp up spending faster, trying to catch up with the guys that are already doing it. And eventually it. You know, that's sort of like the creative destruction thing. The most productive companies using the AI, the best beat their competitors out. And then the gross margins of the S P500 are higher than they used to be, and the net margins are higher than they used to be, and the growth rate is higher than it used to be. And that's what's happened for the last 200 years in this country. It's made the bull market over centuries happen.

Also, you know how somebody goes in and spends $20 on lunch in some city and then that money goes out as a kind of payroll for the folks that work at that place. But it also goes to paying on the supplies on that restaurant and then those suppliers who got the money from that can use it to buy more. It's like the whole thing about how a dollar is worth way more than a dollar if it's moving around.

That's the magic of velocity, of money. Yes, the best. I think you actually gave me this example once when we were kids, John, in college. You said there's a scenario where like one guy owns the dollar, a bank, and the bank owes another guy, a corporation, a dollar, a dollar, and the corporation owns their owner a dollar and the owner gives a dollar to the guy at the beginning of the chain and he gives the dollar to him and the corporate bank gives it to the corporation, the corporation gives it back to the guy who gave it to the first guy. And now there's no more debt in the economy.

It is legal, legal check kiting, it is.

But that's the magic of the US system above all. And that is the magic of any economy, at least how it does work. And you know, you can go to even the poorest country in the world and there's probably at least one high rise that somewhere the magic of their economy has created some wealth. And what's beautiful about the US is we have, you know, a higher percentage of people in the middle class and upper wealth than we, than most other places do. And obviously that percentage has declined, but so has the actual prosperity level of even the bottom people at that disparity. So like, I'm not going to pretend it's a perfect world or that the US is in, you know, perfect either. I mean, clearly we've got our own issues and disparity and, and poverty and sadness and kids who don't have access to health care and food that they need and everything else. But you improve any economy, any society, is to create more prosperity and more economic activity. And that's hopefully what the AI revolution is all about.

Bryce, your model, is it more macro for being able to tell, like, are things actual? Is it paying off that the capital is happening the way it is? Is there the correlation there, so to speak, or is it also micro that you can go in a particular, you know, a particular firm and what they're doing and go now because we found out the fact that that model works, it makes sense for them to do this investment or like, how do you see the model getting used?

Sure. So it's mostly a macro model. We're looking at employment as a whole in the United States and estimating how much of a productivity impact each sector of the economy is seeing from AI. And I've read a lot of studies, there's some, some data out there on that. You know, the hours saved per employee, the adoption rates, things like that. So basically went sector by sector, try to estimate what, you know, the degree to which AI is actually helping the productivity in that sector. And then, you know, extrapolating that to the whole economy and comparing that to the level of spend by these top seven or eight companies, which make up, you know, 8 or 9, 90% of the spend, probably. So that kind of gives us the macro framework to understand whether at the system level, returns are generated, are adequate to justify continued investment. But then from there, the use case for us as investors is to find the companies that are doing this and the ones that are benefiting the most. And we own two of the major hyperscalers, Google and Amazon. We also have bets on Core, Weave and Meta and others in AI universe. So if we're confident that the overall system makes sense, that we should, as an economy, keep spending this level of money on investing this capital in AI infrastructure, then it gives us the confidence, we have a bullish macro framework that gives us the confidence to make individual bets on the companies that are driving the revolution.

Let's put some numbers to it, Bryce. Right now, in your model, we're looking at $130 billion of GDP productivity. Let's not call it GDP, let's call it productivity enhancement within GDP. And if GDP is 30, 25, 2025, and GDP this year is 35 trillion.

Yeah, it's right around it's 30, 30 trillion.

So you're talking less than 50 basis points, less than one half of 1% of GDP impact on productivity. So this is a very reasonable, I think, assumption to make. I think AI is certainly doing at least a little bit of impact. Less than 1%, less than 1/2 of 1% on GDP. I passes the smell test. And the spend over the last five years from the eight or not 80 or 90% of the spend has been what price?

And around 1.2, $1.1 trillion over the last $1.1 trillion. That includes the 2025.

And what did you get for revenue over the last year? What this year? 2025, Revenue on AI agents, LLMs and percentage of cloud that is being generated as revenue because of additional demands from.

The AI revolution, somewhere from between 50 and 70 billion dollars.

So let's call it 60 billion. So 60 billion on 1.1 trillion is 5%.

So no, 60 billion on 1.1 trillion.

Yeah, yeah, that's about 5%.

Okay. Okay.

Yeah, yeah. So right now at this very moment of dial up Internet version of the AI revolution, you can't really justify the CapEx yet on those numbers. But they start working quickly over the next two, three, four years as the revenues are going to compound quicker than the spend compounds. And what's your model show for 2029 Bryce? Total spend trailing five years.

The total spend in 2029 will be, will be up to 3 trillion at.

That point and total revenue will be.

2029. I think the AI revenues will be $614 billion. So it's about a 20% return on investment.

That's probably the threshold at which it needs to be. That's the minimum. And so there you can, if this model is right, there's some concern. That's not. Those numbers don't work as well as bullish.

Fantastic. If you can invest, it's not so bad return. That's a fantastic rate if you get.

From 5 to 20 in that short number of years.

All right, that's the point, I guess the point being that you need to see the, the growth as a percentage of the spend. If you go from 5% to 20%, that's a trend line that people can get excited about. 20% itself is the baseline for a five year depreciation schedule on the infrastructure. And so five years, it's 20% per year of depreciation. And so that's the baseline for this to start working out. But I don't disagree. If you get that Trend going from 5 to over 20% over the next four or five years, the spend will continue because most likely in seven or eight, 10 years you're somewhere at a spot where it's generating a lot more cash than you are spending. And then it actually is all justified on a major big scale that makes it all work out for everybody. Is that what we're getting at, Bryce?

Yeah, I mean, two things. I do think that the rate of return is going to get up to closer to 30% within 10 years. But even at the 20% return ROI, even though that's just a, you know, five year, what's less than if I'd be. If you roll a 70, it. What's that? That's a three and a half year payback on a compounding basis. And I like that you're. I don't think that the asset base is going to be worth zero at the end of the five years.

Two points you just made that, that are fantastic. You're right. That the 20% compounds while the depreciation doesn't. And so it actually is a nice little math thing you did right there. Good job. And then you're right. Even right now when you look, there's still a one hundreds from five, six years ago that are doing inference for chat, GPT inquiries and whatnot. And so on that basis, there's still value, even though it might only be 5% of the actual revenue being generated by those platforms. If they're still generating 5% of your revenue from products that you bought five or six years ago that are running at 100% capacity, by the way, that means you're actually generating pure free cash flow on an accounting basis in that scenario.

Yeah. So right now we, we think that there's, you know, if we, we think this model, we've tried to use some conservative estimates, we've tried to use realistic estimates, we've tried to support it with data from surveys that we've found and triangulated. And it, to me it looks pretty clear that there's room for this Capex to keep going and to keep spending on AI, to keep driving higher and higher productivity over the next five, 10 years. The train rolls on.

That's, and, and, and look, this is a snapshot of what we've spent today doing this. Bryce told me this idea six hours ago and I was like, yes, you know what? Like it's, the more I kept thinking about it. I called him back a couple hours later and said, drop everything else you're doing because this really is probably the single most important call question out of everything we're doing in our investing lives right now.

Right.

Can the revenues justify the spend on capex? The bulls are blindly saying yes. The bears are blindly saying no. Let's try to triangulate some reality around. Let's build some models. Let's put some time and energy into doing it. We'll update you guys next week after we think this stuff through. We're going to deep dive on it ever more. More. Just like Jim the Golden Retriever will do. More and then more.

Yeah, I built the underlying economic model back in April, the depth of the market when we were just trying to understand what the impact, the total impact AI to GDP could be from AI. Just because, you know, at that time everybody said we're going to go into recession. Trump's tariffs were going to mess up the whole supply chain. And you know, the bank's average recession odds for 2025 was like 50%. And so we built this, I built this model and I just, like, just AI's impact alone is probably 50 bips on the productivity side, another percent and a half on the CapEx side. So you're looking at, you know, 2% GDP impact from AI, just from those two things. No, I don't think we're going into recession. And so that kind of helped us, you know, stay bullish throughout the crash.

In April, the depths of the market and the panic. And you're right, Bryce, you've been building this framework for probably even longer than that. But you built that model first around that time. And it was a. I told you this today. It was. You were way ahead of everyone on understanding the impact that GDP was actually going to have in 2025. I don't think anybody was there until. And now even today you can read articles and they're like, 2025, the impact of GDP on GDP from AI was.

Yeah, right, 2%.

But I'm being. Nicely done, Bryce. So that's where we're at. That's the story. I guess you can go to tradingwithcody.com if you want to get that article from Bryce. We're not sending that one out for free, but you know, we write a lot. You can read that stuff. Train with Cody.com. reach out to us if you just actually want to get an occasional email, which will be more frequently than it used to be once we bring on our new analyst partner guy, Craig, in 72 hours.

All right.

Yeah, yeah. Very exciting. Onward and upward. I guess that's the final words for me. What do you got, John Dio?

I. I don't have any really relevant final words. Yeah, I told you guys, you were on your trip.

That's it. No, no relevant words.

I went to Boston and hadn't been there in a very long time. Saw the house I lived in 30 years ago, so got to show that to my daughter and. But it was. It's a city I really like.

It's a.

It's a really, a really cool city. And you just feel like every city along the east coast has its own character. They say that in Boston it's what you know, in D.C. it's who you know, and in New York it's know how.

So when. How long did you live in Boston? What year?

Three years.

What years were that? Early 90s.

I. In the 90s? Yeah, some. Yeah. Somewhere in the 90s. Yeah. After college.

Bryce Eristopher, what do you got? Last word.

Tomorrow, Tesla's supposed to do some kind of product unveiling and product Unveilings at Tesla always disappoint. For some reason, Marcus always hates it. And so just a heads up, if you're long. Tesla.

Did you see the new hands on the Optimus?

Yeah, yeah, but that's not. That's. That was a week or two ago. This is. Yeah, they, they did something over the weekend. A seven second clip of a wheel spinning. I. Bryce, I totally, totally disappoint the world. And here's what it is. It's H Vac systems for home and business.

Okay?

And it's the. Everyone's like, wait, we thought it was like a new Optimus Fly flying robot or it was like the partnership with Archer. Have you seen that nonsense?

Yeah, yeah.

Everybody's like, it's a robo taxi that flies and they're partnered with Archer.

I heard it was.

Look, I'll bet you anything Archer was up 20 today. And that's the stock you don't want to own tomorrow. Tesla might be down 5 or 7%.

You're right. 18, 17.9%.

Yeah, give it all back. And more. If it's not Tesla partnership tomorrow.

No, it's a fan inside their iPhone.

H Vac. It might be the most boring product that Ted Tesla's ever announced. It could be a hundred billion dollar business in five or ten years, man. Like if they announce an H vac that's 30 more efficient than any other thing on the market. Like everything with Tesla, it's their market to lose because the most energy energy efficient products win over time.

What if they just made it look big and it's actually this very tiny fan inside something electronic.

It's.

Hold on, inside their iPhone. Compat, you know, competitor.

Yeah. So somebody this weekend was like, did you see the new Tesla phone? I was like, what? And they're like, yeah, Elon dropped a phone. I'm like, no, he didn't. I saw it on Facebook.

No, that's this Nigerian video that's always. What is that?

It's an Optimus robot. Nice model version. What if, what if they actually. What if they're releasing like a miniature action figure?

Optimus robot that can walk around?

That would be awesome. I would buy that. I paid 800 for a walker for a miniature Optimus robot.

Well, it can do kung fu, so.

Yeah, that would be bad ass.

All right, good one, Brian.

It's going to be H Vac.

I like H Vac.

All right. Peace and love. Thank you, Johnny boy. Appreciate you as always, Bryce. Thank you everybody watching. Thank you. Peace and love. We'll see you next week.

Later.