The AI Revolution Call: Is Software Truly Dead??

All right, well done, Craig. Yeah, that's our, uh, the AI Revolution intro call. Now I would like to hear it say the AI Revolution, and, uh, maybe what you ought to do, just lay your voice over the top of it, another track. But the AI Revolution call with— and say it should be spoken out loud, but it's coming along. All right, better than no intro video and better than a 40-second— you know, I— as much as I even liked the song, uh, even sampled it for a rap song with Moses before, uh, Greatest American Hero theme song. Takes up too long in the beginning of the show. I don't like a long intro, you know.

Yeah, I don't think anybody does, really.

You mean Believe It or Not? Um, that one?

Yeah, of course.

Yeah, that would get the whole, uh, whatever canceled. You won't be able to monetize it.

No, you can't.

No, that's all you hear from these people is like, oh no, I can't do it.

Yeah, I say just sample away. Give it away, give it away now. Oops, I just sampled the Red Hot Chili Peppers. Uh, it is the Revolution Call. That was a— we might have had a short intro video, but that was a long and winding pointless banter of an intro to actually, uh, the talking part. I'm Cody Willard, Craig DeLong, Bryce Money— Bryce Money Smith. Uh, they— we all work at 10,000 Days Capital Management. Uh, and John Dio is a, uh, lifelong AI, career-long AI expert. Yeah, 30 years or more. He's really old, can I tell you. Um, yeah, so, uh, Bryce had an interesting thought today, John. If we, uh, we might start a little bit markets-oriented, which I know sometimes you enjoy and/or ask good questions and/or comments with.

I hope to dominate the software industry. That's what I— market.

Yeah.

Exactly. Bryce mentioned this afternoon that the market on the one hand is like software is absolutely doomed, and on the other hand, it's like, but AI isn't actually creating anything or changing anything. Anywhere. Right? Is that the phrase, Bryce? Is that how you phrased it? What'd you say?

Well, yeah, it's the, the market seems to think the AI is so powerful that it can replace all SaaS. And simultaneously, it's not generating a high enough return to justify the investment that the AI companies are making. Right? So Amazon spends $200 billion on CapEx. They're like, nope, that's stupid. Why would you ever do that? And on the other hand, they're like, you know, AI is going to kill the entire multi-trillion-dollar software industry. But I thought it wasn't generating— if AI can kill the entire software industry, Amazon is quite reasonable in investing $200 billion in CapEx, I would think. They're the ones replacing the software industry.

But both of those things cannot be true at the same time. Well, like, that was—

stuff about this is not— you ever talk about, oh, is it a bubble and it's going to pop? It's like, well, this is the opposite of a popping bubble. This is like popping somebody else.

Right.

Yeah, this is popping the bubble of people saying that AI is in a bubble.

It's— I do think that software as a sector has long been elevated in its valuation. And there's a lot of like egregious stock-based compensation out there where, you know, GAAP earnings are like minus $5 per share and non-GAAP earnings are $15 per share. You're like, like this is becoming a little bit absurd, right? You can't have those kind of discrepancies between GAAP and non-GAAP financials. But I think the SaaSpocalypse, as it's been called, has more to do with the long history of software being overvalued and playing games with numbers and less to do with cloud code.

Yeah, it's like in some— it might be both because it doesn't have to be mutually exclusive. Like cloud code and the threat to software future made the market actually much more judicious about, you know, the numbers and how it gets reported. And, you know, valuations were not cheap in software pretty much, you know, other than at the bottom of the great financial crisis, the bottom of the dot-com, the bottom of COVID There are moments in time where it might have gotten— moments in time where it might have been relatively cheap, at least historically cheap. And now it's, you know, software sort of like been brought down to earth and it's fairly or somewhat undervalued from a traditional market perspective, looking at like growth rates and P/E.

Cody, you know, I, I, um, I was thinking like not everything can be overvalued, and we have a new area to be overvalued. So this one had to give that up, and now you've got like, um, Salesforce goes from what, 30 PE to 15 or something. It's like, that's, that's reasonable. 15 is reasonable. Um, so it's—

well, and that's where you might argue it's even undervalued because the market, the S&P 500 trades at 23, 24 times earnings, so it's a low market multiple. Um, but it's certainly not like, like, you know, 5 times earnings like GM it trades at or something like that. It still has a little bit of growth priced in. Um, you know what is interesting also, and addendum to this whole topic, is we— I think we've even hit on this topic before— that the value chain, the value chain of semiconductors and the ex— like, the, the complexity of designing and making, especially making semiconductors, is making— and the demand for more power and more semiconductors of all the sorts that go into running AI data centers is so through the roof that semiconductors are— if they historically traded at lower valuations than software as a service stocks. And that is clearly flipped, and probably rightly so. I think we've predicted that in the last couple, 3 years, even here on this AI call, certainly in our writing and in our investments, because it's so much harder to replicate what NVIDIA has built than to replicate what CRM has built. It'd be very difficult to actually replicate what and then maintain what CRM is built to, but certainly is more possible for like a startup to build their own Salesforce software suite internally than it would be for a startup to be like, hey, let's make our own chips for our computers, for our cloud business customers, you know. So, um, anyway, it is We've certainly seen a rotation. And then not just— I don't mean stock rotation, but we've seen a rotation in actually where the value, the margins are. Another example then would be NVIDIA's margins, historically 35-45% gross margins, and they peaked, what, about 18 months ago, Bryce, at 85% gross margins. And now they're back down around 80% gross margins. Am I getting those numbers right?

I don't know that they got to 85, but they're about 75 right now.

Yeah, I think they got over 80, but I could be exaggerating.

I'm—

some want to do, um, but you know, so, and then, you know, Microsoft's margins have gone down from 99% when I studied that business the first time in the late '90s to what, probably 60, 65% these days. Um, so anyway, you know, software— CRM's margins are still outsized, and Arm Holdings is still outsized. Synopsys still has outsized earnings. So it's not like, uh, gross margin— so it's not all gross margins and all software have gone away. But as Craig wrote in his morning note, it is very important to remember that all software is not a monolith. Thoughts, comments, arguments?

Yeah, I don't know. I mean, I just think it is being treated as a monolith, right? Um, for the most part. There are some exceptions in Palantir and things, but even those have been hit pretty hard with the sell-off. My— I think just from a sector perspective, like You have like the software industry as a whole is one of the most technologically advanced of all industries, right? And that's a simplification, but it's pretty true. And AI is software. And I think the software guys are going to figure out how to use a new software tool to probably benefit their software business. And maybe they can't figure that out. Maybe all of them won't figure that out. But as a whole, I think the software industry will figure out this— figure this out largely. I think the business models probably will be different. But like the markets, like this entire sector is screwed. And for the most part, maybe without exception, anytime the market decides an entire sector or industry or asset class is screwed, dead, going, never coming back from the grave. It never works out that way. Crypto has returned from the grave multiple times. The financials in 2008— JPMorgan's like a trillion-dollar business now, almost. Like, these things—

yellow price though, the exceptions to your rule are Yellow Pages, the horse and buggy.

Like, those industries, those are specific things, not sectors, an entire sector.

Fair enough. And, and, you know, we did download Claude bot. What's its— is that what it's called right now? What's it called?

Open Claw.

Open Claw.

Open Claw to my Mac last Tuesday. We had our most recent AI call was on last Monday. The next day, Bryce and Craig were here at my offices. We downloaded OpenClaw to my computer.

No, no, you have, you have Clawed Cowork.

Yeah, do not put OpenClaw.

You got— all right, OpenClaw is the one that has like, it, it, it could, it could like completely like destroy every file in your system and things and give away all of your accounts. Yeah, this one is The one that I downloaded— what did I download, Bryce?

Claude Cowork.

Claude Cowork, which Open Claude used to be called ClaudeBot, but I downloaded a ClaudeBot named Claude Coworker. And don't download Open Claude because that Claude— that thing that used to be called ClaudeBot could destroy your computer. This is—

now you know why they had a trademark problem.

Yes, I understand, but it is the most confusing thing for someone who's not like actually doing this. I'm already confused.

You're confused because it was a trademark Let me finish, John.

This is taking way too long to simply get to the point of the story that it was mind-blowing to see just how holistically it did take over my operating system and computer on my Mac. And we just told it, hey, go make a video game using Gemini. What was it? An RPG, role-playing video game using Gemini. It went and did it. And then we were like, why did we tell it to use Gemini? You can actually just tell it to make the video game and have it run on my computer. So we did that and we told it to make the game and then run, actually build the game so we can play it on my computer. I probably open it up. I haven't looked at it since last Tuesday, but you know, like that stuff you do on AI, but the way that we were interfacing with it. And I guess what I'm getting at is you could see how quickly we are getting to a point where you could certainly tell an AI agent on your computer Hey, I run a newsletter business. I think you should send out this letter to every new customer that soon as they sign up, send this one out 3 days later, ask them how they're doing 3 months into it, make sure they don't have any questions. And like HubSpot and some other CR, like their, their customer resource management, like light even software programs. You can see that AI is going to get to a point where it's going to be able to do so much that those things are currently doing for us. But you can't get— you can even make a CRM thing with it, but to maintain it and get to the complexity of like Salesforce.com, Salesforce itself, It's not gonna happen anytime soon. Maybe in 3 or 4 or 5 years, Google Gemini is so amazing on a holistic level, it truly could do all of it from the ground up on just some prompts. That's what the market's been worried about. Bryce counters that with, look, the software guys at CRM are not sitting around twiddling thumbs. They're gonna go figure out how to actually integrate it and make it do it inside of their program and make sure that there's still so much value add from the Salesforce CRM itself and the magic sauce therein and how it integrates with everything in your company or whatever, there's still ways to do that, to grow, to modify, to become even bigger and more powerful for some of the big CRM— let's call it SaaS companies. We do not have a position in CRM anymore, by the way. We did sell it a couple, 3 weeks ago. Traded it out for Snowflake specifically because we think Snowflake will help Salesforce win. And even if Salesforce doesn't win, Snowflake will help whoever does win win.

So Snowflake did not take a hit the way these other things did because it was treated as, um, more like support. Okay, I don't know if it went down a little. It wasn't like the others. It was reported that it's being treated like Databricks where it helps AI happen.

Yeah, but John, it got crushed. Yeah, that's why we got to sneak in and buy it. We'd rather buy CRM, uh, Snowflake when it got crushed than hold on to CRM while it was getting—

it got crushed as part of the software thing or before it?

I literally bought them the day, like last Thursday, with all of software.

But I mean, it's not trading at 13 times earnings, like—

no, but it, it— I mean, it was, you know, down at that moment, it was down 30% year to date.

In any case, it's recommended to be in it. Because it's supportive of AI as opposed to dependent upon it.

That's where we are. Yeah, that's what our thought process was. Threatened by it, it's actually an enabler and a beneficiary.

I mean, the other part is, John, like, let's say ClaudeBot or ClaudeCode, Cowork, whatever you want to call it, was AGI-ish. Okay, intelligent. If you have it on your smartphone or PC or Mac, and you have no apps and no software on this thing, what good does that coworker do for you? Is it going to rebuild a browser? Okay, I see I'm on a Windows operating system with no apps. First thing, I should probably build a browser, then I'm going to build an email client. And then I want to be like, it's, it's so— why? There's no reason for that cloud coworker to do all of those things. Oh, I see customers. I should probably build a CRM. It's like, probably what it's going to do is go sign up for Gmail, sign up for Salesforce. Well, yes, but that would be the reasonable thing for the co-work thing to do in mind.

Yes, but you won't, um, the— all those things become supporting actors as opposed to the main star.

And they are supporting— Chrome is a supporting actor for me as an employee. Well, yeah. OARC becomes the employee, not me.

Oh, Clearwise, all these apps that you name that you used to visit and spend time in that judge themselves by monthly active users and how much time you spend in them and so forth become not used by you but used by your agent. And so you originally, yes, you go get a license for your agent for it perhaps in the short term, but then it starts to be, well, I, you know, I don't really use its UI or these things about it. I, I let my agent do that for me. So people aren't going to be as excited about, you know, the AI that happens in Google Docs or whatever different thing. It's like my AI is the central point and I happen to have these other things around it.

Yeah. And you, instead of, um, head, you know, number of human subscribers to these things. It'll be how many agents are paid subscribers, how many API customers are paying for access to CRM data, etc.

Yeah, I think that kind of brings us back to that framework that we've discussed and worked on and how some are more resilient than others and positioned for agentic commerce versus the ones that are just UI wrappers for humans, you know. So again, you know, they're not all the same, but I do think there's some nuance to be had.

Can— I think the most obvious example, like, there's several probably, like Duolingo is clearly threatened by ever smarter AI interfaces.

The—

like, you could literally, you know, like Duolingo's differentiator in some ways is like, hey, you can play games to learn. Well, You could literally just tell Claude Cowork, go make me a game that teaches my daughter to practice on her whatever language she's wanting to learn. Another one is obviously Adobe and just, you know, the being able to prompt pictures and then prompt improved pictures and prompt fixing your actual pictures. Like all of the things we used to pay for, the Adobe subscription is been completely like disrupted. Now that being said, we're Flip It guys. We should always invert. We should always come up with the other side. We should empathize with the people who are buying Adobe stock right now. We should empathize with the person who's actually signing up for it. Surely there are still people going and signing up new customers at Adobe right now. What are they doing? What value proposition are they seeing? Can any of the three of you guys explain why I should sign up for Adobe or buy Adobe Stock?

Well, I mean, it is— you still do need something to open and edit like PDFs. Things like that. There are alternatives, there have been alternatives for a long time. Gemini doesn't do that. Somehow. I even told Gemini before, build me a PDF editing app, and it doesn't work.

And here's something I don't know about Adobe, but I know. So I've been playing around with Lingodeer, which is like Duolingo. but it has a lot of East Asian languages. It kind of specializes in that. And I asked Gemini to build me a, a language app just to see how well I could make it. And I got like a few prompts in, and I was like, dude, like, this needs to get so much better for me to actually build my own app when I could just pay $15 a month. And if this is something I really want to use, I'll just pay it just to make my life easier. I feel like I'm doing something. But I don't have all the time in the world to build apps. I know there's a lot of people that feel, feel that way. But once the AI gets powerful enough to just build you something amazing like that with like an interactive interface, I mean, I truly think, you know, then it'll be disrupted really. But for now, I can, I can see why some people would still just rather just pay.

Craig, can't you— you're a video game guy. Can't you picture a first first-person shooter game where what you're shooting has to be the actual words that are Spanish words that you're learning, and you have to be quick enough not to shoot the words that aren't Spanish words but are Latin or something, Italian, and you're practicing reading as you're playing the first-person shooter game. Like, that is not that far away for you to prompt Gemini. What's the Gemini one that I pay $250 a month for? That's Premium Ultra, whatever.

Ultra, yeah.

Yeah, you get on Gemini Ultra and you get to play with Genie, Gemini Genie, and you make video games. And you could totally, within, you know, even right now, you could get to a point where you could make the video game I just described. It would take you work But what your point is, in another year, in another 2 years, it won't even take you any work. The AI will be so good at doing it that you'll just be like, hey, make me a first-player shooter game where I have to shoot Spanish words and not other languages.

Yeah. My, that's my point is that the, the technology is going to get there right now in the short term. I can, I can still see it having a, a decent value proposition for most people, but in the future, I mean, the sky's the limit. For what you can imagine.

Can we just also say, while we're on the topic of stocks and these trends and things, it is— it was a surprise to no one, certainly not at our firm, that Duolingo was threatened by ChatGPT and the ilk. But what is so stupid about this job, about Wall Street, about markets, about stocks, is that stock went— it spent the first 2.5 years after ChatGPT moment going up a double, a triple, a quadruple even, and went as like most wildly overvalued place it could have gone in the face of obvious disruption, obvious revolutionary trends undercutting its entire business model. And it didn't matter because For whatever reason, the momentum was there. Short sellers kept shorting it and getting squeezed out and having to cover. It becomes a self-fulfilling cycle. The numbers themselves were not collapsing. People obviously didn't immediately go cancel their Duolingo classrooms, didn't cancel their Duolingo. But you go look at that stock now and it's down 75% in 365 days. So if you think it was easy, just go short Duolingo because ChatGPT came out— no, we actually learned, you know, as we always do, like, it's not easy. We lost money trying to short it, and I think we missed the entire move down.

Yeah, yeah, it was at 5:40 at the time.

You have a real job. Uh, Doug says, I don't think anyone is buying Adobe Marketing. They focusing on it? They always— they're always trying to upsell, cross-sell, take over your computer. You log in, it's more to do. Are you sure you don't want to pay us some extra money for this other bullshit feature? Oh, it's the worst. Schools have firewalls? Yes, they have firewalls, but lots of schools are using Duolingo. To teach secondary languages. I think that might be what you're referencing, Emma B. Um, if I'm missing your point, please elaborate and we will address it.

I just think the markets are always discounting the future and probabilities and things, right? That's really what stock prices represent. That's why they get so wildly off at times, from overhyped to undervalued, you know. And there is a lot of uncertainty. The— why we have these debates, the reason we can have these debates is because we don't know for sure what the future of the software industry looks like, right? 5 years ago, it was pretty easy to kind of extrapolate out what, what software looked like for the next 20 years. And now we have to reconcile the fact that AI exists and it can program things and it can make shitty language translation apps that aren't quite as good as Duolingo, but maybe they might be someday, right? And I think that's the game as a trader and an investor is like, where is— what is certain now that might become uncertain, and what is uncertain now that might become certain, you know? It's sort of a game of predicting that in the long run, whatever happens in reality kind of is the ultimate test, you know, and if Duolingo were to grow 20% per year for the next 25 years, stock's gonna be incredible value right here, you know, but—

And he says, yes, I think we were hitting on the point. It's like, yeah, the schools have firewalls and And I guess also maybe what she's mentioned might be getting at is, you know, it'd be difficult for students to just access a Gemini game that you randomly built through a firewall, etc. Absolutely all true. But I guess what I'm picturing is, you know, 5, 7, 9 years into the future, this stuff gets smoothed over and there will be just so many different options for even not necessarily even learning languages at school, but just You know, Duolingo has millions and millions of just random people like my daughter who's homeschooled that are students use— trying to use— or not just students, but people trying to learn a secondary language. And, you know, I personally would enjoy building a game like for learning a language that I am learning. So I would find— I find the building of these games and apps like Writing a song, it taps into some artistic gene somewhere inside of me that I'm like, oh, that's fun, let's work on that, let's make it better. If I had more time and more compute, I would probably do it a lot more aggressively. But right now, that's, by the way, another thing we just got to say again. If you use any of these AI apps, you still see there's— they can be sluggish doing very simple stuff. And every time it's sluggish, you know, we need a 10x, 100x more compute power.

That's so true. I just today I was using the Gemini Autobrowse feature. And I was like, alright, let's, let's put it to the test. So I was having to update our employee handbook. And I had to do it with code. And I had AI, I said, can you go ahead and insert this code for me? It was a— I don't know how to pronounce it, John. There's like L-A, like latex or something, T-E-X.

LaTeX.

LaTeX. Yeah. So I was using that for the PDF.

And it goes in there.

And I had like, it tried to do it, it erased everything. And I had like 38 errors, all kinds. I said, alright, stop, stop.

Now went back.

And I just asked it, I said, where do you want me to put this? And it told me exactly where to put it and what to do. So the knowledge is there. But it actually going into the browser and doing it was, you know, it was a little rough.

LaTeX is really for making, you know, documents that have like mathematical equations and stuff like that, you know.

Well, it's— that's what it told me to use. So I used it and it came out great. But it was just funny trying to get Gemini to go in there and actually do— use the autobrowse feature. That's like an agent to go in there and input it itself.

It had a hard time. Is about 30 years old. And it's interesting that these AIs, they don't care. It's like whatever does the job, if the stuff is there, not everything had to become modernized. It just has to use the stuff.

Yeah.

Which is the whole point of AI making everything easier. Again, I think a couple points we should make here is that when people are worried about everyone needing to learn a new skill set, the only skill set that's really required is prompting well. Speaking accurately, like if you're a good communicator, you're going to be like— AI enables you to be a good accountant, attorney at law, like whatever it is, whatever your profession is. If you communicate well, AI is going to make you better. That's the whole point. You shouldn't have to train yourself how to use AI. The promise of AI is that you don't have to be trained to use it. It's just so easy. It's got the intelligence and the interface for you. And then the other thing I think that was interesting, apparently I didn't really pay attention to the Super Bowl ads. Frankly, I hardly paid attention to the Super Bowl after about the second quarter because, boy, that got pretty boring. Um, but there was a lot of AI ads, right? A lot of AI application ads. And I think one of the things that we're just hitting on with like Snowflake and why we're not in CRM and There are probably, you know, hundreds of thousands of AI application companies, startups out there right now, many of them raising millions, some raising hundreds of millions, and a handful raising billions. And that's— there are going to be, there will be some billion-dollar AI application companies. There are going to be $100 billion AI application companies, just like there's a Netflix out there. There are Airbnb, other, Carvana. There are $100 billion smartphone applications, but the value stack truly created the trillion-dollar businesses on the platforms that these applications run on. And that, for those smartphones, was Android owned by Google and iPhone, iOS owned by Apple. And that's been— that was like almost— it had more to do with having a number of developers and customers that could all bring together a critical mass of both customers and developers. So it made those incredible applications. In AI, there are only— there might be hundreds of thousands of application companies out there right now, but there are only maybe 5 or 8 companies on the planet that have hundreds of billions of dollars to spend creating the AI infrastructure that all the applications run on. And that is another reason why we've stuck with the hyperscalers— the Googles, the Amazons, the Metas— because those are the only guys that can actually build this AI. It doesn't matter how the government— France, isn't there this whole meme thing going on right now that France, it's like, take this, America, we're going to spend $2.6 billion on AI this year, and they make the chart look bigger than the $800 billion that's being spent in the US on AI. It's very fun. I've seen it twice on my Twitter. Maybe I was bored this weekend and spent too much looking at it, but it's out there. Like, literally, even like no other country can spend— not even the United States government could spend $800 billion on CapEx making AI data centers in one year. Bryce, what'd you say today? Like, I think we might completely— not we, the economy might completely blow people's minds for the next year. Because if getting us to this point— like, you remember we've talked before about how The ChatGPT moment brought on this CapEx cycle, and that CapEx cycle actually kept the broader economy from going into what looked like it was a for sure recession as the COVID bubble had turned into a COVID crash in the markets and the economy had all contracted and it looked doomed. And then this spending cycle went nuts, and along with it there was animal spirits brought into the market so people would chase stocks and That becomes a self-fulfilling reflexive thing. Hey, more money flowing into the thing. Hey, more money to spend on the thing. It all builds up. That was amazing. Like, but it's not slowing. Like, we're going— like, Amazon doubling the spend this year off of last year's doubling. And if that last year's doubling of getting it to $100 billion of spend was that impactful. Well, now it's $200 billion from one company that's going to impact the market, the economy, the cycle. This is incredible stuff. We need to just step back and sort of appreciate the magnitude of it.

Yeah, I don't think anybody appreciates the magnitude of it. Like, I don't know if even like Amazon people can get their heads wrapped around $200 billion in spending. In a year, it looks like.

Like, we can't— we, like, we, we're trying to get our heads wrapped around it, and we all— and we know, like, we strive to actually understand exponential growth, right, Bryce? Who's the guy who said that nobody understands— human failing is we can't understand exponential growth? Like, we can't, no matter how hard we try, even we are not going to fully understand and grasp the magnitude of this incredible $800 billion one-year spend that's going to grow again next year and the next year too. Clearly, these companies, they can't stop it. They have to do it. What, Bryce?

Um, well, I was gonna say something different. You said they can't stop.

They can't stop. The return on the investment is too attractive.

Oh yeah, for sure.

They're like, hey, Would you like to spend $200 billion and make it back in 3 years? Yes, I would like that opportunity.

Right. I thought you were going to say that they had to because of this prisoner's dilemma thing, and that's been out there forever. And I don't agree with that. I never was like, you're telling me because Meta made this decision, now Amazon also makes this decision. And then now they're betting hundreds of billions of dollars because Meta bet hundreds of billions of dollars. It's a silly game. That—

come on, it's a return on investment. It's a clear return on investment. Like, they're like, hey, holy crap, we actually had 30% return on what we spent this year. Do you think we should ramp it up next year and do that same thing and get bigger billions dollars return? Yes, we should. It's a— like, it's— you'd be a terrible business opportunist if you weren't doing this.

But the The market's not used to it, right? I mean, you hear that people are— that part of the tanking comes from people going, oh my God, look what they're doing. It's not a question whether they're doing the right thing. They're not used to it. So they're acting like they're not doing the right thing.

Again, with Bryce's point earlier, like, that's what provides the opportunity. The market is not agreeing with us. And if we turn out to be right, we own a lot, you know, big positions for many years, frankly, in Google, Meta. Tesla and others, but there's bigger than they've ever been.

And but why did, why did the market get excited when Meta reported $125 billion in capex? Is it just how they framed it because it was such a direct correlation to their ROI?

I think the, yeah, I think the pundits misinterpreted the market's excitement on Meta's quarter. I don't think The market just was like, holy crap, look how much return on investment. Look at the acceleration of this business. Look at the acceleration of the margins. Look at how incredible this whole fundamental business is. And then the market was like, oh crap, they're spending a bu— oh my God, but it's so good. They're spending too much, but it's so good. And then Amazon, you know, was sort of a different thing. It was like, yeah, the numbers were all right. It wasn't accelerating except Oh, Azure Cloud was accelerating, but they're going to spend $200 billion. And you know, it just— Craig, moment in time, it, it, it— trying to explain— I've said this a million times on my writings— that trying to explain any one-day action in any single stock is an impossible task. You can always come up with reasons why it did the stuff, but nobody really knows. The other, I mean, look, Amazon is cheap, but it's not as cheap as Meta is. Amazon is also very cheap though. And as you know, we bought more Meta in the last week. We bought more Amazon after its earnings report last week. We bought more, uh, Arm after hours when it was down after its report. We thought it was a great report and turned out now it's up 25% since that report. Report. None of it makes sense. None of it. Short-term never makes sense. Long-term earnings make stocks go up.

It's hard to believe Meta was at $580 a couple, a few months ago.

It was at $6— we literally bought some in the hedge fund at like $603 last Wednesday night. Yeah, right, Bryce? Or was it Thursday night? Whatever night it was, we bought some after hours at like $603. Am I crazy?

I don't remember.

I think we did.

It's possible.

Word is they're bouncing back with their AI model is going to be really good. Next round on that for them.

Avocado, right?

Yeah.

What do you know about that model?

Just that it's doing really well, even on its pre-training. So that's before it's gone through any stages of refinement. It's just like straight out of the gate, um, which is a, you know, it's a good sign for— and they're— it's multimodal, I believe.

Yeah, I just had an incredible deep thought on this. So Meta and its infrastructure, AI data centers, that's the garden. Avocado is the vegetable that they're growing. They personally use some of it to make pico de gallo— uh, I mean, guacamole. And they're like, yeah, look, we actually throw some other mixed stuff in here and use— make an application with this avocado that's delicious. And then even other people are like, hey, you know what we can do? We can make some amazing shrimp tacos and put some of that guacamole from the avocado from the garden in the shrimp tacos. There's a billion-dollar business somewhere along the way making— not literal, but digital.

Just a sec. B Money, do you have any idea what he's talking about?

Yeah, I follow him.

Sorry, John, my boys. Although that might have been like The Sopranos. Tony goes home and tells his wife, hey, you got to hear this joke I told at the Oscars today. Everybody thought it was so funny and everybody cracks up and everybody's cracking up. And she's like, okay, tell it to me. He tells it and she's like, that wasn't funny at all. He's like, well, everybody was laughing. And she's like, you pay them to laugh at your jokes.

Isn't that interesting?

Yeah, I don't laugh at your jokes most of the time.

Actually, that is true. Bryce, actually, Bryce is uncomfortable. It makes me uncomfortable about the lack of laughing he does at my jokes.

But Meta does— if this part of the analogy— but Meta does have a garden, has a market for its AI even before it tries to go sell anything anywhere. It has its whole installed base of you know, a couple billion.

So, um, yeah, they can distribute their AI and get a billion people— what, it's like 2.2 billion people per month across Insta, Facebook, WhatsApp— that can interface and learn, use to help train all of Meta's AI. And then they also are using it a lot of that compute power to just make those algorithms ever better. And my gosh, they are good. I don't really do this, but my wife will sit there and like— so this weekend she's scrolling her Instagram and I'm like, let's watch some of this. She's like, all right, let's go to the dog puppy stuff because that's— she knows I like the golden retriever puppy stuff. And so we're scrolling through, it's hilarious, and then there's this ad for, uh, Oh, a basketball that tracks like you're dribbling and you're shooting, and it's an AI basketball thing. And we just hesitated because I was like, oh, that looks sort of cool. Anyway, let's keep going, puppers. And then a few minutes later, they hit me with that same ad but slightly different variation. And you can tell even the ads are AI-generated, but they're good. Like, it makes me want to buy the basketball thing. And You just know the algorithms themselves are feeding you more relevant ads, and they're figuring out how to make you want to buy it ever more. You can see it in real time.

Yeah.

Yeah, I was actually talking to Gemini last night about just planning some upcoming vacations. And I turned on the Gemini or the Google Personal Intelligence. And I was talking about, you know, going to Seattle. And it was like, hey, I know you've been really into like Filipino cuisine recently, he said. And it just some random because I was just talking to it about that the other day. And it said, while you're in Seattle, you should try these spots for Filipino food. And I was like, dude, that's so cool. And I asked it, I was like, did you look in like use personal intelligence to feed that to me? And it was like, You caught me. Yes. You know, I went through your stuff and I know you like that. And so just between the amount that Google and Meta know about me, I can only imagine how much more money I'm going to end up spending at some of these businesses because I've been—

It's funny, Craig. I remember having the epiphany like in 2006 or something and learning about Facebook and just social media business models and thinking it through. And I was like, You realize I wrote about it. It's like, you guys realize, like, if you're pregnant and you post about it, then the Meta algorithms will know that you're pregnant and it'll actually tailor its ads to you. Are you serious? And it really is like, it's both horrifying on a privacy level, but it's also just absolutely helpful on a shopping level.

Absolutely.

Doug says your devices are eavesdropping on you. I don't think the devices are. But I guess the device itself has to be if someone is. John, you believe that?

Yes.

iPhone is—

happens all— it happens where we— Marcy and I turn to each other, we're like, how could it have done that? How could it have just put that in our, like, YouTube feed? We have—

You don't think it's conspiracy? You actually think— you think, like, the devices are listening to you, sharing the information with Meta, and then Meta's delivering you an ad based on something that your device sitting across the room heard you talking to your wife about?

No, that's silly. Google. But yes, and I think it's actually been proven. So we got to go— I got to look it up or something for you. But it's there, the devices.

Bryce, are you buying this? Do you think?

I've never— I've seen it happen. And the ads are remarkably good. But I don't— let's just think about it in compute terms. Before, because this conspiracy is old. This is way before AI came around, right? So you're telling me billion smartphones doing real-time voice analytics in 24 hours a day, always listening, right? Processing all that data, feeding it into machine learning algorithms, and then spitting out ads on the platform. No, no, no.

First of all, how hard is it to listen to voice? My phone does it every time I go to, you know, to talk in a chat. So it's, it's certainly got the capability.

I don't know, my phone doesn't understand. I, I asked Siri for the sports or the weather and it never gets that right. I mean, it, it can't even tell me what the sports, and somehow I'm supposed to leave it, it's finding WD-40 ads for me. Here we go.

And I saw a great documentary on this, this Chinese kind of— what do you call it— cyber hack software stuff that got into about like 200 different utilities, whether it's this wallpaper or whatever. And they would go and put the stuff out and let it go for a couple years without anything in it. And then in the update stage, when it's updating the software, they'd sneak something in there and all of a sudden it's keylogging.

And I would—

and millions and millions.

Do I believe there are nefarious actors in foreign governments who spend lots of time and money trying to listen to our conversations and probably sometimes successfully figure out how to do it on random cases? Yes, 100%. Do I believe that the United States agencies that would want to listen in on our— yes, they, they can if they want to. But are they recording what I'm saying right now, even on this stream, much less on my phone over there? There's not enough compute. There's not enough memory. We already have a shortage of memory. Now you're telling me we need memory so that every conversation I'm having, even randomly to my daughter about golf, is going to get recorded so they can sell me better ads? No, John, it's not happening.

I know, foil hat.

So I have to— I know my mom is screaming at the stream right now saying, yes, they do listen, because every time I talk to her, she has like another anecdote about this. But I feel like if they did listen, I should be getting all kinds of Tesla ads in my phone and I don't get them. And we talk about Tesla vehicles like all the time.

What it is, it's selective memory. And what happens is the algorithms aren't good. They're going to— if you Google something, you ask Gemini something, you actually interface with the internet, you're on Meta, you search the topic. It's going to remember it. It's going in your algorithm and occasionally it's going to feed it back to you.

And if you—

that one time that you happen to have been talking about, for example, WD-40 and remembered that you needed some WD-40 prior and somehow the algorithm kicked it in at that moment, you're going to selectively be like, holy crap, it's listening because it knew. But the other 200 times that day that you were discussing things not WD-40 and you did not get an ad for them?

That's—

it was just good algorithm and it happened to be coincidental at the time you were talking about.

Yeah, it is true.

Here's the bigger picture in my opinion. The reason the algorithms are good is because— it's not because they show you things you need, it's because they convince you to buy things that you don't need. They're like, here's this thing you've never heard of like a basketball AI tracking app, and we think it would— you would love it, right? And you're like, yes, I would love that. I never have heard about basketball AI tracking, never had this conversation with my wife. Yes, I would love it. I will have two, right? Like, it's not like, hey, here's some Clorox bleach. Like, that's what the old television ads were, right?

Right. No, exactly. Another one that they were showing me up last— this weekend on my wife's feed was like a smart mouse, and it, it like has— you can change the settings, and you can actually— has a screen on the mouse, right, for interfacing. And I thought I was like— I, I, again, I actually watched the ad. So then about every, you know, 5 minutes, it was feeding me another version of that ad, and I was— it was really selling me hard. I, if I, I, I was this close to clicking through and buying it. I did not.

That's a good point though, Bryce, because I feel like the stuff that I get is usually stuff I definitely don't need. But I've bought quite a few things from—

yeah, that's— I mean, that's the long tail of advertising is what Facebook does. They're not doing like General Motors and GEICO Insurance. Like, you're not just seeing tons of fucking ads about— excuse my language— tons of ads about like the same stuff you see when you're watching Fox Business or whatever, you know, like, it's just, that's a poor usage of ad dollars is for Coca-Cola to be spending, running expensive ads on Meta when it could be like much more targeted to the people that actually, like, the people are gonna buy Coke by Coke. But they're trying to, it's, it's the, it's the people selling some niche product on Shopify that want to use Meta to get you to their shop.

Well, you don't, number one, you don't, you know, you don't have a budget. Most of those ads that you're seeing don't have budget for mass marketing, print, national print magazine, slash commercials on TV, slash Roku. But you can certainly, you know, any— I've made ads in years past for random things. I was— businesses I've been involved with, and you just, you know, even before AI, you can make a great ad on the platform itself and then tailor it. Back in the day, you could even tailor it based on race, age, religion, et cetera. Over time they've tweaked that, but that's like everything. There's like dog whistling. You can figure out and find your targeted audience triangulation anyway.

You talk about, you know, AI to make ads, but what about ads in AI? And which is what, uh, you know, Anthropic's commercials were about in the Super Bowl, which were kind of funny. Uh, you saw those?

I didn't see them. I saw them when they got posted on Twitter and everyone was Yeah, yeah. So Altman was all fired up.

Yeah, he got all pissed about it. But it was very mean.

You're not gonna do it like that.

Right.

But, you know, and, you know, you could say, well, Anthropy, you got anything else to talk about? Maybe something about your product? But, but it was a good— it was funny. So it is.

And, you know, it really actually was about their product. We don't We will not do this. Our competitors are. And we just want to remind you, we won't. It was about their product. But yeah, I don't know. It's ads in AI.

I don't—

I mean, Google's going to be doing it. I mean, of course, like, it's just what— so what?

Got to get your money somewhere. So now Anthropic's not because they don't need to because their money— they know their money is going to come from a different place. So they're doing pretty well, I would say.

Yeah. Well, I'm using the $8 a month Go option for ChatGPT. So I should be getting some of those ads whenever they roll out, or I don't even know if they're alive yet. But when— if I get some goofy ones like the commercial show, I'll definitely share them with y'all.

Yeah, no, I know this. I am agreeing. If you—

that's very different than what we were talking about.

What's that?

This is very different than what we were talking—

that's not spying.

That's like, absolutely, anytime you interface with Alexa, Gemini, Siri, anytime you interface with the internet, yeah, they're going to track you. They're going to know what your interfacing interest was, your interest in interfacing was. So 100%, I agree with all of that. I'm just simply saying, when your phone is sitting on the kitchen counter and you're talking to your wife about WD-40, it did not listen to that.

I'm, I'm not gonna tell Alexa.

My conviction level is high.

Like, if you're using Alexa to buy a bicycle or trying to fix your bike, and then later it's like Amazon's like, hey, here's WD-40. Of course, that's what the freaking thing is designed to do.

Doug, I mean, I agree with you. I think we must be missing some sort of premise on this discussion. Because yes, Alexa, Siri, your phones are listening on device for the prompt.

Correct.

Well, they are always listening to you. And if they think you said Alexa, it turns on and sends it into the internet where you are immediately tracked. But I do not think the Alexa device sits there processing, thinking, sending what you're discussing. I mean, it's literally against the terms of services. And we're talking collectively between Apple, Google, Meta, and Amazon, $10 trillion of market cap that is at risk if it were ever actually proven Android is not following its terms of services other than when United States government agencies ask them not to.

John, you're muted.

Are you doing that to be funny?

No, but usually I can't talk when I'm not muted either. So the— just kidding. The— it did happen though that Google got a large lawsuit against them for having their incognito mode not be incognito? It's supposed to be all hidden data and Google was collecting the data, right? That happened to the tune of billions?

No, I don't think billions, we would have heard of it.

You did hear of it?

Yeah, I've heard some, like, they showed that like they were maybe still getting some cookies on you or something. I can't remember what the— and you're done.

Your premise is back in play.

Right. Definitely. Like Google, like Google, I don't know how much they're monetizing the data, but they, they had more data than they were supposed to have on incognito users.

5 to 8 billion estimated.

Yeah, but it hasn't been awarded or something, right?

They agreed to a settlement.

Yeah, they were sued for $5 billion.

The Senate agreed to a settlement, and they didn't say what the amount.

I'll go with $50 million.

Estimated between $5 and $8 billion.

I'm reading Google's own page about Google's, you know, I'm not buying any of it.

You know, back to if we're talking about ads and AI, I found it interesting that on Facebook and Instagram, I get a lot of ads for Claude. Like, I get ads for Claude, like to use Claude, Opus, and all that stuff. I see it all the time, which I find interesting. You know, I never see any OpenAI ads or, you know, Gemini ads really, but, you know, I feel like Anthropic does kind of go out of its way to get more consumer users.

Well, I get those ads too, Craig, but we are power users of AI and we are enterprise users of AI and Claude knows both of those things and they're going to hit us with Claude ads. I don't think my dad and mom are getting Claude Code ads. Yeah. On Facebook.

Yeah, they're advertising to you specifically because you are obviously AI man over here.

And immersed, I guess, in those I use. Yeah, yeah.

No, that's right.

You saw someone on Twitter again this weekend.

I—

look, Ameris had an eye infection, so I spent like 2 days laying around the house with Ameris, hanging out. And, um, you know, we actually had fun, Super Bowl steaks and frites. But nonetheless, I spend a lot of time apparently on Twitter this weekend because it's now about the 4th, 5th time I've referencing it. I, I I saw someone complaining on Twitter that they were getting too much porn in their feed, and I was like, I don't think you should post that because I'm not getting porn in my feed. I, I bet if I were clicking or searching for porn, I would be getting porn in my feed. And if you're going to complain to everybody— and they like— it was like complaining. It was like, can't believe X has all this porn on here, it's always in my feed. And you're like, oh God, dude, that's you. That's, that's your—

well, my point is, isn't that— I mean, I understand why it targets me, but I'm saying I don't see ads for like OpenAI or anything like that. Like, it's just interesting that Anthropic's paying for those ads.

Like, I see ads not targeting me on OpenAI. I see ads on TV. Yeah, going back to the Coca-Cola discussion, I like ChatGPT's Coca-Cola and maybe Anthropic is more long tail.

Yeah, I just thought it was interesting the different modes they're taking for—

it would make sense though, right?

That maybe—

yeah, maybe Anthropic knows that they're not going to compete against OpenAI in the general consumer, right? But they definitely want to capture a Power AI user like Craig. Did you know you can't spell create without AI?

I do know that. Yeah, there you go. Check that out.

I just still think it's surprising that everyone says that models are commoditized. They're commoditized, everything's commoditized. And like Anthropic and OpenAI, xAI, and Google are all taking very different paths, but they're all commoditized somehow.

Well, and again, let's assume that there are only— that there are 7 commoditized LLM AI things over the next 5 to 10 years. I'll own each of them. Like, that is still an oligopoly. That is not exactly like a commoditized heavy commodity, everyone can catch up. No, there's only 5 companies that can ever create it. And so just own them, even if it is a little bit commodity-oriented.

Yeah, I think the idea of commoditization is like a little overblown. I'm bearish on commoditization as a trend.

You're a commodity.

Somehow, like, the internet, you would think would be like the ultimate commodifying-ish tool that— and it's dominated by like a monopoly. The biggest monopoly in the world we've ever seen dominates the most like open source networked freelancing thing that's ever existed is owned by a monopoly.

Which monopoly? Google.

Google is the internet. Google The whole open internet search, everything.

Whole dream died.

The whole thing.

Yeah, we democratized nothing.

Yeah, Trade Desk wants to run ads, you go through Google. You want to— like, there's Google, and then like if you're on an app, you use Meta products. The App Store is amazing democratization tool. Meta has 90%, like 90% of the revenue that goes through the App Store probably goes to Meta advertising or something. Like, this is very egalitarian and very monopolistic at the same time. How is that possible?

We live in the world of irony. Yeah, most ironic outcome is usually the most likely. Is that the Elon Muskism?

Yeah, yeah.

All right, last words. We've been on here for an hour.

I'll go. So, um, we weren't talking about it, but the, um, I mean, we said it last time a little, but we could— we talked about the, you know, the SpaceX merger, but the notion that it's, um, power in space is the real, uh, reason is what he said, you know, in an interview. And he said you can't do it on Earth. There's just no way to get the hundreds of gigawatts to come online. It's like, he's like, what are you talking about? We're gonna have 1,000 new power stations. Where's this gonna come from? So, um, I, I didn't hear in that whole description— I want to kind of understand the, um, not that I know what, what it means, but I would study it a little— the, um, the distance from Earth, like which orbit it's in, because I assume it's got to be like farther than the Starlink and be out there somewhere where it's more um, geostationary as opposed to geosynchronous and just always sun-facing. And like, like, how do they stick this giant thing out there?

But yeah, John, it is worth studying. I've looked at it somewhat, and it, it's, it, it's always mind-blowing how even orbit, the concept of orbit works. You're constantly— it is gravity that's constantly pulling you towards the planet, but it's never fast enough to actually make you fall to the planet, like all of that. And so they end up figuring out even just orbital planes that get the 24/7 sunlight even in low Earth orbit. And so I don't know, we should— yes, you should study it and then talk to us about it because it'll be interesting.

Yeah, so that was pretty cool to understand that that was the reasoning, sort of. It's not like—

well, it's got to be, right? I mean, this is where we talked about how China's got this huge lead on building solar on Earth. And I told you guys last week, I did some research. It'd take 24 years of them doing as much as they did last year in this record year to catch the United States energy generation on, on a per capita basis. It's like 8 years to even just catch total, uh, production. But if the— if the— in the end, the most economically and energy efficient solution of AI is what's going going to be the biggest and best and most dominant, then you've got to figure out how to do that. And the US slash SpaceX can do that, and no one else can. China cannot create a low Earth orbit of general—

anyway, that one's mine.

So, uh, I just want to comment on that because I did listen to Elon talk about that, and I've been skeptical of this XAI-SpaceX merger and everything. And Elon might be wrong on his timing, but he actually might be like, the timing could happen even faster. Because my main concern with AI right now is that we can't build as fast as the hyperscalers say they can build it. Yeah, there's not— we're hitting the limits of construction, labor, concrete, steel, getting things, transformers, all that stuff to the right place. Like, you know, my family's already hit the limit, right? I mean, yeah, it's like, like, we're like, we already can't build what we said we were going to build last year. And now we're going to— now it's doubled, the budget has doubled for this year. And what Elon kept saying on that podcast, they're like, yeah, but it's this, that, and this and that, like, you can just do this and that. And he's like, try and do it. Go try to build a gigawatt data center, one, just try to build a 1 gigawatt data center, you choose the place, try to go do that. And come back to me when you finish. He's like, He's like, it was like a Herculean effort to pull it off in Tennessee. And they were like firing up temporary generators. The Tennessee, like the local Memphis people protested. And so the government shut it down. So they had to build a power plant in Mississippi and then build power lines from Mississippi to Tennessee. He's like, that was for 1 gigawatt. And all of the, like, everyone's like, we're gonna do 5 gigawatts here, 10 gigawatts there. And where I think Elon might be right is just like, even if space is like not that cost efficient at the beginning, he can actually do it. There's nothing stopping him from putting up those— like, if he can figure out the physics and engineering, like, all of this, like, the constraints on Earth, most of the constraints don't apply. He just needs solar panels and he needs the chips and the spaceships to send them up.

Yeah, you need the minimus costs to get to orbit. Which is what—

you can scale to gigawatts while Amazon and Facebook are sitting around waiting on permits from the California regulators. Like, he, he could actually win on timing. Yeah. And I was like, wow. Let's—

they said he'd have to do how many launches in a year? Uh, 10,000.

Yeah. And that was to hit— I don't remember how much Was that a terawatt or something?

It was a huge number. Yeah, yeah, to do 1 gigawatt, it's not anything like— I mean, it's 1/1000th that.

So even if it's a ton of launches, it's like, so there are a ton of airplanes that fly every day.

So, right, that was his point, was like, yeah, that would actually be small by airplane standards. And he's like, I could probably do it with like 30 or 40 Starships launching one every 3 hours or something like You know, I don't know, Elon can speak so like far out and crazy sounding. And then whenever he explains it, it's like, it's like really rational. And he's like, I just solve for the hardest constraint. And right now, if I order a generator, it's like 5 years out. If I order a transformer, it's 3 years out. And I'm in a race to build AI before anyone else. Maybe I can do it faster in space. That's pretty much his thinking.

And then even though why he wants to build a terafab for semiconductors inside of Tesla is sort of this, like, what's the biggest constraint for him building more and more technology? And Taiwan Semiconductor is not going to scale to the scale that Elon is going to— that space that Tesla is going to need if they're going to build millions of Optimus robots over the next 5 to 10 years. Craig, last word.

Um, well, I've been using Claude in Excel and it has been mind-blowing, to say the least. Been pretty fun, the stuff it can build and help me out with. Um, I need to get pretty granular with it, but I plan on doing some more modeling and utilizing it soon. So I'll report back with that. So maybe the ads are working. Because I'm probably going— I used Copilot, give me a $9.99 a month special. But I feel like my money will probably be spent better just going with Claude.

Why mess with the Microsoft product? Like, why even— like, Craig, why start going to— cancel that. I don't even Don't even— I seriously, I don't want you to mess with the Copilot. It's horrible.

Yeah, you know, it's kind of that thing where you have to know, you know, your enemy. I was like, what are they doing?

Prove how right I am, how bad it is.

I will say, on one comment on that, but I had two people this weekend were like asking me about AI. And like, what do you think about Copilot? I have Copilot now. And if you're in business, or if you're in government, like Microsoft just pushes that shit out onto you. And like what we were saying earlier about how things have to be so easy, like the barrier for people like consumers and businesses to adopt technology is like really high. Like if it's hard to use at all, people won't use it. If they have to go download a new app, it's got to be damn good reason for them to go download a new app. Like If you're not in the tech space, you just don't do that. And like, my dad— I've tried to get my dad and our construction company to use AI and use Gemini and build like some custom apps and stuff. They won't do it. But they're using Copilot because they can click a button in Outlook and it like makes them emails, makes emails for them. That's Microsoft's advantage. That's why they've been able to do this for 30 frickin years, because they are so entrenched.

Counterpoint. Crypto exists. Billion people fuck with this crypto thing. I still— I've made my own crypto. None of it makes any sense. I'm always constantly amazed where we're like, you'll have to cut the token code and then you're going to go paste that in your wallet, but you have to approve inside this wallet. This code, like, and somehow— so I'm just saying, Bryce, somehow crypto exists. But your point, I think, really is for a— like, there's not a billion people in crypto, right? It's— no, maybe tens of millions have actually ever traded crypto out of the 7 billion people on this planet. And for a billion people to use your product, it does have to be as good as an iPhone or as good as Meta. Like you got it to get to that scale. It has to be easier to— but the whole promise of AI is that it does that, right? Like, that is the point, that your father will just simply quit even clicking the button and he'll just be like, hey, do it. The thought— he's like, I'm about to build this thing, do that for me.

And good enough risk is a thing, right? Like, just something being good enough to where people don't use the other stuff, you know.

So I Yeah, it's just going to get better in technology. It always gets better. Here's my deep thought for the day. I put my hand on this while we were doing our last words. And I use this, I use this, this archaic 100-year-old calculator. Technically, I was the first calculator, 19— 60. John, you're on mute again, but call it 65, 70-year-old technology.

72, the Bomar brain.

1972, same year I was born. And somehow, despite all of the AI that I'm immersed in, all the PC— I got my monitor over there, I got, I got all of this technology all the time surrounding me, and somehow it's more efficient occasionally for me to pull this thing out. And there's got to— I, I've got to think about that. There's an— now there's some something there. There's, there's, there's importance on that thought that we've got to get the— why that— why does this thing still— why am I still using this? I have a smartphone, I have a computer. It's the interface. It's as archaic as it is, it's still good.

Yeah, and those are harder to set up than anything on your desk.

It's a lot easier to download this app and make a phone call than it is to boot this phone that Craig spent hours upon hours trying to make work.

And what about this? Yeah, I mean, why do we need this?

Yeah, everybody thought the Kindle was gonna kill books and that didn't happen.

Hold on, I'm making a phone call.

Okay.

No, no, Audible killed books.

Yeah, uh, Carozozo692. Yeah. Oh, Mrs. Walden's not there right now.

Tell her.

Party line.

No.

All right. Yeah, yeah, it's a party line. Yeah, tell Miss Walden to call me when she gets back on the line. Thanks, guys.

Was that your last word, Cody?

That was my last word.

What's— what is your—

why—

what would you say, like, if you had to answer right now? Like, why calculator, paper, handheld phone.

Part of it is— most of it, I think, is legacy. I think it— I think we look— I grew up using this, and even through college, you know, I did not use a computer for calculations till I got, you know, like 1995 or something. I started my first Excel spreadsheet. So it's partly generational. It's partly that this is actually a great invention, and it was designed great, very well, and very ergonomically, and very efficient. So for those reasons, it still exists. But we should—

I—

there's, you know, just from a revolution investment perspective, it is something to think about how certain items live 70 years of useful life and others don't.

Staying power.

Staying power.

It reminds me of when I was in school, I was the only one using notebooks in my class. All the other younger kids, they were using tablets and laptops, and I just can't take notes that way.

Here's the first smartwatch, watch that I was ever able to find. It's an Android. It's got the telephone operating system inside of it. I probably bought this 12 years ago. It was so poorly designed and such a poor interface that it sits in this drawer and never gets touched. But this is still used on a weekly basis.

That's a good point. Yeah, some things are tried and true, huh?

There you go, another very good standard, another good revolutionary idea right there. Tape and tape dispenser, both brilliant. There's actually something about that in the analogies too, right? The Dispensing of the tape was actually almost as important as the invention. The invention of the dispenser is almost as important as the actual tape. That way Spotify— Spotify— Spotify, the music is the tape and Spotify is the dispenser, right?

I like that.

See, they, they like my idea.

You gotta send me your, your 50, uh, you have to send me your songs too, by the way.

Oh, for publication purposes? Yeah, yeah, I've got to do that. I want, I want you to do it when you're here on my computer so I can see you do it, and then I can do it in the future. That's why I didn't send it to you. All righty, plan on coming up here next week.

Okay, dude.

All right, Johnny, Greg, B Money, B Money, were your last words about SpaceX and XAI, or did you want to get something else out?

No need for me to pontificate any further. This—

no, you've done excellent pontification there. So you're the pontiff.

I like it.

They're listening. They're listening. All right, guys.

Okay.

See you later.

Thanks, John. Thanks, everybody.

Oh, well, we got to do our video.

Are you doing it?

Yeah, you just hit end stream.

Set outro video, probably the last one.

Oh, if you just hit end stream, it'll do it.