The AI Revolution Call: Tim Cook departs! Robot Marathon? And Amazon's $50bb side gig!

That's why the oil prices are so high.

Is that what that was spewing out of his chest, was oil? I thought it was like some sort of like—

it was Iranian oil.

All right.

All right. Well, welcome to the AI Revolution call. I'm Cody Willard. That's Craig DeLan, who makes our weekly intro video, and, uh, it— I don't know if it improves every time, but it's still fantastic every time. It's hard to top the Michael Jordan dunk cyborg video thing.

Yeah, yeah, that was really great.

Yeah, that was our fair— our first, first and best. And Bryce Smith also works for us at 10,000 Days. John Dio, AI aficionado.

And it was full width, Craig, right? That time you got it to go full width. Yeah, I don't know why sometimes it does, sometimes it doesn't.

But I didn't check to see if it went full width. But yeah, I guess it did. All right. Thanks. Great job, Craig. I always enjoy the intro weekly video. And I haven't not chuckled yet. My mom would mention that was a double negative. John, two wrongs make a double wrong.

Double wrong.

Ladies and gentlemen, AI revolution called. Bryce, do you have something on your mind?

This is just pressing breaking. I had not seen this till just now, but Apple has named a new CEO.

What?

Oh, wow. Is it the guy that we don't want? John Ternus, who was the head of hardware, head of the company now. I'm just glad it's not their AI guy. I think that guy's an idiot. Sorry, he's listening. He's—

no, that's actually interesting. You know, it's a long time coming. Steve Jobs handed off the reins, passed away. It's been 14 years.

Just showing this guy's face. I've never seen him before.

Oh, I've seen him.

Yes.

I wouldn't be able to tell you his name.

And he's young enough.

You know, it's sort of exciting that we're long— I bet Apple's up. After hours. Not much, but I bet half a, half a percent, 1%.

Looks like down about 80, 90 bips right now. $3.

Tim Cook was a great op— I'm not going to say anything revolutionary, revelatory, or revolutionary here. Um, Tim Cook was a great operator. He didn't mess up what Steve Jobs built. But from the fact that my Apple Watch can't take a phone call even though I pay for LTE service on it to the fact that it needs charging. And the Apple Home devices sold a whopping 12. The, the products that, that Tim Cook rolled out were pointless, but he didn't mess up the App Store and didn't mess up the iPhone. So, you know, there's a little added risk, but frankly, I would take a little added risk to have someone who actually is not the scariest, most anal retentive CEO in the top 50 tech companies on the planet, which I think Tim Cook was. Any arguments here? This was not supposed to be the cup, Bryce. What are you throwing this out?

This is huge news.

This just happens. It's, it's just happening.

So John, your thoughts?

Well, I think that's interesting that he's the, like, the hardware guy. Uh, and which says that it's really where they've been really successful was like their, the whole M series of chips, you know, the M1 to the M5. And if he's behind what they did there, uh, that's really good for them because that's where they're doing really well.

I don't know, I would think technically he was in like head of iPhone and Apple Silicon team, I think, is headed by somebody else. That's right. It's a silicon team.

He's not a silicon guy, he's just the hardware guy.

Yeah, the hardware guy, not the silicon guy. Okay, it's a point well made that if anything, the one thing I, I couldn't— I, I would imagine frankly that, uh, Steve Jobs probably got them started on the M1 chip before Tim Cook was there. Craig, would you search that on Gemini while we're talking here?

Well, it wasn't called that at the time, but Right, but that he— Apple Silicon.

Steve Jobs got them working on their own custom silicon and away from Intel, and probably not Tim Apple as well.

They went there and back, right? They were, they, they were on Motorola chips, and then they went to Intel, Intel, and then they came back to their own.

Then they went to their— I wouldn't call that back to because they weren't—

I thought there was something in between Motorola and Intel, which was their own, but I guess not. It was straight with Motorola to Intel and then to their own.

What do you got, Craig? Did Tim Cook or Steve Jobs start the process, the semiconductor business?

It's almost done.

Come on, Gemini.

Just for background, the head of Silicon is, um, Johnny Shroji, and he will now be the head of hardware, taking the job of the guy that's now the CEO, John So he does step up like that.

Okay.

Yeah.

The brainchild, it was Steve Jobs' brainchild, but Tim Cook incubated it.

Okay. So thank you very much. Tim Cook did nothing in 15 years running the company except not mess it up.

Executing.

That's how well—

Executing is pretty hard.

It is. And Elon talks about it all the time. The idea is almost worthless. It's the execution around the idea that actually creates the value. So you do absolutely. I'm being somewhat facetious about my commentary on Tim Cook. Not entirely though, because it has been like Apple has ceded being the technological leader, um, over the last 15 years. And on the other hand, profits are $100 billion plus, or close to it. I mean, yeah, right, Bryce? 30—

sorry, I wasn't listening to that part.

What are Apple profits?

Apple's profits?

$100 billion. Yeah, that's about right. Earnings?

Yeah, I think so.

A 30 PE on $100 billion gets you $3 trillion. Interesting though.

I mean, this is, this is pretty cool.

Fact of the matter is it's actually great for Apple shareholders. I would be a buyer if Apple sells off on this news much at all. One 80 bps or whatever it's down after hours isn't meaningful. But if it were down 5% or more— we bought a little Apple recently, um, lower prices in the whoosh down in the market, plus Apple got taken down too. It's up nicely since we did some purchasing, but I would— I'd be willing to buy more 5% lower. Call it— what's it, probably $271 right now?

Yeah, Apple's profits are $117 billion. So you're very close. Yeah, the stock price is $270 and change.

So yeah, call it $260. I'm a buyer. Again, I've owned Apple since March 2003. Cost basis is something like something.

This is pretty cool.

So when Steve Jobs was still around, he had a secret team working to use— around 2010, Apple had a small secret team successfully porting the Mac's kernel to ARM architecture, and they were keeping their options open because they didn't have faith in Intel, uh, to keep the vision— to keep the vision going for them. And then Tim Cook, after he passed away and took over around 2015, um, he was like, all right, now's the time, we have to pull the trigger. And so that was when it —so again, some credit to Tim Cook.

He did pull it off after he finally decided to follow in Steve's vision. I don't know, profits are better, the computer is better because it's not on the crappy Intel x86 chip and it's on Apple ARM CPU. What's been my favorite stock for the last 6 months, Bryce and Craig? ARM Holdings. What were you about to say, Bryce? I was— I—

you may have said this, I don't— when I wasn't listening, but in terms of market cap creation, Tim Cook has certainly created more market cap than Steve Jobs did. And in creating— the amount of market cap that was added under Tim Cook's tenure—

now do percentage. Well, certainly Apple, certainly Steve Jobs percentage-wise created Ironically, millions and millions of millionaires. And you know what Tim Cook did too. I mean, just percentage-wise though, yet had a much better return under an actual visionary.

I did the math one time though, and I concluded that of all human beings, Warren Buffett probably saw the greatest increase in his personal net worth from the rise of Apple than anybody else. Because, because Steve Jobs sold, he sold after he got fired, he sold everything. And he got more stock when they acquired NeXT Computer, I think was that.

And they also acquired, what's the movie company, Pixar.

But that was Disney. Sorry. Yeah, my bad. Yeah. So he sold, Tim Cook didn't have a lot of like early equity or anything like that. Mote— um, Wozniak sold pretty much. So like, there, what— there isn't even— Steve Jobs's daughter is worth a few billion dollars, but she— most of that's in Disney stock, not Apple. Um, so it is, it's just funny. I mean, it's sort of sad, I guess, but you know, the original founding people did not see huge net worth gains because of what they did.

And even, uh, the original investors.

Yeah, that— I can't remember that guy's— one of the originals. Remember it either.

Yeah, side anecdote, uh, Lincoln's best friend lives in Brooklyn, New York, and she goes to school with, uh, Steve Jobs's granddaughter. Oh, cool. Um, you know what, I was actually going to start the call today with the fact that I was just outside, look, just getting some fresh air, um, looking around and breathing the nice— it's a beautiful day here in Carrizozo, New Mexico, and there's bunch of cranes, smaller cranes moving railroad ties. Birds redoing, you know, remodeling the railroad tracks, for lack— I know there's a better term for that. Anyway, I was just thinking from a first principles aspect, I don't think getting cranes out there to move railroad ties to move oil and shit on 150-year-old technology railroads seems like the future. And what I was going to tell you guys is that, and it's related to Tim Cook here, you know, that it feels like Apple's gotten away from first principles of technology, first principles of information rules. You guys haven't read this book. I've got 4 copies on my desk because I hand them out a lot. I give them to people all the time. Information Rules, Hal Varian, Carl Shapiro became the head strategist of Google, not in a vacuum. It's in large part because of the principles in this book. But the Information Rules, Elon's approach to first principles, all of this is very germane in a world of AI and part of the promise of AI, probably the biggest promise right now is improving margins on corporate S&P 500, Russell 2000, millions of smaller businesses around the world. And it takes them a leap forward towards a first principles approach. First principles approach being the most economically and energy efficient method of doing anything, including laying railroad tracks, which is actually simply a competitor to other distribution methods. And so it really is interesting that we are about to apply first principles and utter efficiency principles to all aspects of the $100 trillion global economy. Thoughts?

Yeah, I just think, like, with AI, everyone is sort of like, it's a— same with robotics. But these, these things aren't like incremental advances. They're sort of step change, step function changes in the growth rate, you know. And we like Cody Willard built an economic model that shows that we'll double the world's wealth if we grow for 3% per year for the next 20 years or 25 years. And like, it's really good. We think it's going to be greater than that, because the whole curve is going to shift up, you know, the entire productivity curve shifts up. And like, we think of AI as these chatbots right now. But when you think about it as sort of like a new operating system for computing, and a new capability for consumers to do things and get work done and shop and everything else, like That's when you're like, look, the way—

can I just mention, lay railroad tracks. Like, everything gets— AI gets applied to everything. That's my point.

Yes, keep going. And I was just gonna say, you can say, like, you know, the markets have rallied, the war fears are dissipating, although they're still here. But like, I know the next cycle is going to be, hey, valuations are expensive, this is tech optimism. But like, if you do believe this AI stuff. And you, if you use AI, you're probably a believer. And it's like, this is really changing everything. Like, I don't know if you can be too optimistic in the long term, right? Short term, yes, you can overpay. Yes, you might pick a loser instead of one of the winners. But there's going to be some incredible wealth creation come out of the fact that the entire economy is being sort of reorganized to be more productive. John, hop in there.

I think about, um, you know, what you look out your window or walk around and see, and how I'd like to be a, you know, a fly on the wall backwards through the decisions that landed those cranes in that place doing what they're doing, and how, uh, they, um, they were probably looking at some, uh, folks who aren't, you know, plugged into it yet, aren't making a lot of use of making the best decision. Not that they wouldn't lay track anyway, but just that, you know, where has this stuff made its way around yet? And if it does make its way around, how does that affect, you know, what you're, what you're seeing there? Obviously, you know, rail is tremendous for its like low cost per mile once the stuff is, is running, but it's probably got a slower, um, get things up into going state.

It's slower to get things up, but it's also a slow, a much longer depreciation curve. I mean, this stuff lasts for decades. Yeah. So it's like data center, which lasts for 5— depreciated over 5 to 7 years. It lasts longer. Yeah.

So like, what's the— how does this stuff you're describing that Bryce was just going through, you know, just hypothesize how that takes effect and how it makes its way into that project?

That's, John, that's exactly my point. So I was just thinking about whatever the railroad company is that's out there up upgrading. That's the word I might have been looking for earlier. Remodeling their railroad tracks, probably publicly traded. Even if it's not, it will 10 years from now have better margins because it will be more efficient, more economically efficient, more energy efficient in laying railroad tracks forevermore. And that's what I think the market is trying. That's why it's looking past the war. That is why AI is the only show that matters. And war, geopolitics, and every— oil prices, inflation, you can run through a litany of bad things in the economy and in the political world, but the market just keeps looking past it and hitting new all-time highs even yet again here with this amazing move forward, uh, move up over the last 20 days. Um, but beyond that, even over the last 3 years since the ChatGPT moment, and over the last 15, 20 years since the great financial crisis bottoms or something, that the market's seen that profitability, gross margins, net margins, cash flows, are likely to hit all-time highs in coming years because of AI. There will be— the thing about capitalism and the promise of capitalism is that other than technological platforms that become monopolies, there will always be competitors that will use the AI technologies to become more economically and energy efficient and thereby beat their competitors. And so if you aren't reacting you miss out. And that trend here is magnified across the entire global economy in a way that probably nothing since electricity has the power to impact every single industry, every single thing we do. And what Bryce was talking about and what I've been talking about so far is just sort of like the margin and the productivity and everything. But what happens is when you have higher margins and higher profitability, you also turn it into more growth and more economic growth, more company with top-line growth. All of that stuff feeds into this. And what I always come back to, my favorite story right now is my wife asks me, what is Jim the golden retriever? How many pets are enough? More, more, John. More margins, more profits, more business, more flow, more stuff going over the railroad tracks, making them depreciate faster, but better technology making them last longer. More, more, more, more, more. Bryce, you were about to say something.

I had a great idea. I was going to tell with you privately, but I think it should be shared publicly because such a good idea. Do it. That, um, on the Jim Moore thing, we— I bought Cody Willard for Christmas an NVIDIA Jetson AI robot chip that you like, a developer's kit. And I think you should— it can run cameras and do all kinds of— I think you should build a robotic arm with the Jetson AI that Jim can walk under and it senses he's there and pets him. Unlimited pets for the dog.

Knows when it's a human that's petting him.

I don't know if that's true, John. You make that robot look like me, you put my skin on it, you put my voice in it. Uncanny Valley. I'm sorry, but Jim would be very happy to have a few extra Codys running around.

I promise you somebody will make an AI dog, AI-powered dog petter.

Why have the dog?

Why have the dog? Where's your dog, Johnny?

Where's the boy? Uh, he's upstairs.

He was— oh, come on, he's the star of the AI Revolution call every week.

No, folks are gonna have to wait till next week for Luke.

You can't replace human creativity, and you can't replace dog cuteness, right, Craig?

If you think he's okay with this, that's right, just get yourself a robot dog and see how okay you are with that. It just does not work.

You know, we have a picture of, uh, Bryce and I in Hong Kong petting a robot dog on the sidewalk. Did the dog appreciate it? No, there's a cute girl and her boyfriend. And so it actually made a great picture.

Um, that made me think of a small piece of news that I thought was interesting. I shared this article with you, Cody, and Craig about the Hong Kong or the Chinese Olympic half marathon for robots. They started it last year. And it was sort of like the joke of the robot industry. They're like, look how stupid these things are. They all fall down. They can't run very fast, the humans are operating them. And the winner last year of the robots ran the half marathon in like 2 hours and 50 minutes. And this year, the winner, yeah, won the race and beat the human record for a half marathon. It was done in like 50 minutes. So that's a pretty stark improvement. But not only that, 4 different robots ran the half marathon in less than an hour, which is quite a feat.

And some of them were by themselves. Some of them were remote control, whatever. But some of them are just totally automated.

According to Google, I asked this question, the winning robot was the true winning robot. I don't know which one is the true— the robot that ran it in the shortest amount of time was human-powered, but then human-controlled. But the second-place robot was fully autonomous. And by the rules of the marathon, if it was human teleoperated, it was it got a 20%. Or if it was autonomously operated, got 20% haircut to its time. Yeah. So by the definition of the rules, the autonomously operated robot won the race. But the same company owned all 3 robots.

That second one that won the race, did it also beat all the humans?

Oh, yeah, I think so.

Yeah, I think. Yeah. Yeah. You know, what's interesting about that is we can debate the gimmickiness of it, like, what's the point, right? On the other hand, the improvement, the exponential increase of capabilities of those robots were something to behold. And 365 days worth of improvement on the software and hardware in China, mostly. I mean, that almost exclusively were Chinese robots. Bryce sort of made a joke in his email to Craig and I about that, where— to Craig and me about that, where he referenced that Tesla Optimus needs to get hurry, like hurry up. And yes, it does. On the other hand, I think the holistic approach of the Optimus robot and how they're approaching it on, again, a first principles basis of understanding physics, gravity, the world, doorways, stairs, and not just being able to run a half marathon as quickly as possible is the whole point. We don't need robots running half marathons for human beings. We need robots like making shit and building things and helping me carry a Maris.

But is it still best? Even like when you go in the all-around? Because I see other robots doing those kinds of things. Maybe most of the Chinese ones that are doing those other things are human-controlled.

Yeah, John, I'm— until proven otherwise, I fully expect American robots to be orders of magnitude more capable than Chinese robots because of software.

Well, the Boston Dynamics one is showing off pretty well. Um, yeah, let's be showing— I know you don't love the, the CEO, but, um, Figure seems to be doing fairly well with what they're creating—

with what they're showing. Let's not call it creating.

Okay, what they're showing is great reveals. What can I compare? I mean, what I have to go by is what I can see that's shown, and I'm getting— I don't believe you're not getting enough, uh kind of show and tells from Optimus here. I want to see— I'm fine with that.

I'd rather Optimus just keep mostly under wraps and then come out one day and be orders of magnitude more capable than the closest competitor. Do you think—

one of the things that struck me, I watched like the NBC coverage of that thing and they interviewed a dad who had taken his children to watch the deal because he's like, yeah, robots are freaking awesome. And my kids love this. And I wanted to bring them to show them. And to me, there is like a little bit of a cultural difference between the Chinese and Americans, because they seem to be like a little bit more open to technology. And in America, anyone I talked to about robots is like, the jobs, the jobs are gonna take our jobs.

It's a really interesting point. I want to clarify, you're saying the Chinese are like optimistic and excited about this stuff. And you're saying they interviewed a Chinese father, not an American father. Yes. And yeah, it is interesting. I have— I've seen visceral backlash against AI, specifically robots, less so because it's, you know, not here front and center on the news all day, every day, whether you're an AI investor or just watching your local news, there's probably something about AI threat to the economy and jobs every day. And yet I've— it's interesting, I feel a visceral backlash against AI amongst the average Joe in America right now. And I certainly don't think that's true in China.

No, it doesn't. It doesn't seem that way from the outsider's perspective, at least.

Like in China, it promises better economy. In the US, we're like, please don't mess up the world. We like our status quo.

We were like that whenever the Model T came out and everybody was betting on their horse can get somewhere faster than the original Model T and stuff. And I mean, it kind of reminds me of that a little bit.

So maybe the full self-driving is the only example I see the opposite where I don't hear that much negative. I hear more positive around that. AI. You know what's interesting, John?

I just yesterday was at a birthday party for one of Lincoln's little friends, and one of the grandmothers there, Reed Trading with Cody, and been following me for many years and bought a Tesla like 5 years ago. And they were like, yeah, we never use Full Self-Driving even though we paid for it. And I'm like, wait, what? Well, yeah, they live in Chicago. And they zero, zero percent of the time they've never used Full Self-Driving.

Well, this is a good time to start because 14.3 is, you know, is really hits different.

Good luck convincing the people who haven't even tried 14.2 or 13.6 or 12.9 or 11.3 to try the 14.3 because they're not. But that's sort of my point is it is interesting that there is just this like random backlash against using AI in this Crate, Bryce, how much more productive are you guys today than you would be 3 years ago without AI in your lives? 10x, 50x? I mean, it's not even close. It's at least an order of magnitude.

Yeah, I would agree with that. I mean, it's— I don't even know how to— I'm doing things that I would not have been able to. That's the thing.

It's a tough measurement because you're doing things you wouldn't have done. Yeah, yeah.

Today we were spending some time with our partners in the UK working on a model, and just in real time, as we're all on a live stream together modeling a company, we're using AI to ask questions and do research for us on things we don't know the answer to while we're building the model. And like, you just practically couldn't do that. Like, if for a company that you didn't know that well, that you didn't know all the inputs and the various things prices of their goods. Like, it just, it would be impractical to have 3 people sitting around trying to build the model in real time. And now we can because we can get so much of that information given to us by the AI.

Literally, Bryce, I mean, you probably saved dozens of man hours today by utilizing AI and your own expertise. I mean, this stuff doesn't happen in a vacuum, but you use your expertise and our existing model and get— refine it. And you get the minutia and you get the details and you get the nuances put into the model because you're working with AI in real time as you're building it.

Yeah, I mean, like, we were looking— we're like, what's the average selling price of an industrial robot and what percentage of that industrial robot is of the bill of materials goes to like motors, actuators, and electric drives. Like, that would take a fair amount of research to triangulate all of that, you know, because that's not like a publicly disclosed thing from the company. And you know, the AI just does all that research and triangulates it for you and gives you a range. And you're like, okay, the range is $7,000 to $10,000. Okay, well, let's plug in $7,500 to be conservative.

It's amazing how most of the time when we do that, we're in the right universe. Converse anyway, right? But it is so like the— having the confirmation, having that actual validation of your numbers and what you're modeling and your approach and the margins and the costs and prices of it all, it's— it, it just changes everything. And we make the joke all the time about like, sorry Craig, sorry Bryce, I guess we don't need any analysts anymore. And that's so ridiculous. Yeah, we need the help. We need the manpower. We need the brainpower. We need the person inputting things. And maybe in 5 or 10 years, AI will be part of that even. But I'm still going to need the creativity, the human brain. I'm still going to need the thought process, the understanding, the experiences. Right, Craig? You're not nodding, Craig. I agree.

I agree. I'm listening.

You know the, um, thing where Craig was saying, uh, well, I wouldn't even have done that thing. Similarly, uh, there's things you do but you would have done differently. So Bryce is on this call with London and they're, you know, talking through some things, and if there are things which would have been, well, next week when we talk we'll have this or that and we can continue on such and such, it's like No, it's not next week when we talk. It's right now. I did it alongside. And so it becomes part of what we do now. So it just changes even what's happening.

Yeah, exactly. You know what's interesting?

I decided, you know what, I need to put a link to Information Rules up on the chat. So I just Googled Information Rules. That's all I typed in. And of course, the first thing is Amazon book. Amazon was founded as a bookseller. And to this day, with all of the last 30 years of advancing, building, making the internet a shopping mecca, frictionless, empowering the end user. If you planted those seeds in that place, if you built that platform 30 years ago, you're still the first thing that shows up 30 years later on a Google, on an internet search. And not even to confuse the fact that I'm Googling with the broader concept of searching the internet. And that's what I'm going to put in there. And I'm like, you know what? Basically everybody who's ever going to see this call or listen to this and click on that link has bought stuff from Amazon before. That's incredible. And that's— there's something to be said why we're so bullish on the guys that are actually building the, the compute power, the AI data centers, and the LLMs themselves, because it does mean 30 years later you're at least in pole position to be the dominant platform. Here's the link for, uh, Information Rules if you're curious. Amazon, Amazon.com.

I'll put it in the YouTube comments as well.

Good looking out.

I didn't know. Yeah. No, I don't know. I mean, I think one of the recurring things with the market and people that are bearish is the valuations of things are so much higher than they used to be. And They want single-digit price-to-earnings ratios or low double digits and things. And to Cody's points about margins, like, a company that earns 40% operating margin is worth a higher multiple than a company that earns a 20% operating margin. There's more defensibility of those earnings. It's less cyclical, you know, it's, um, all these things. It's more reliable. And so, like, I just, I think it's a little bit irrational to look at like operating margins across the S&P 500 just go up, have gone up for the last 30 years. They're not anywhere close to where they were. And I don't think you get to buy companies outside of like a true liquidity crisis or something at those kind of historic, historical multiples that everyone's sitting around hoping show up again, you know. And I'd much— I think it's much better to pay 20 times earnings for a company earning a 40% operating margin than 10 times, 10% growing top line, 20 to 30%, and not 3 to 5% like the lower margin historical defensive staple or commodities company or something.

I mean, that's, you know, and relative to where I've paid for these Revolution Investing companies historically, it's as low as it's ever been. We've had a huge pop in the last 2 weeks when we were perfectly posi— almost perfectly positioned for it. Shouldn't have had any shorts on. Um, but you know, I mean, the— in large part it was because I'd seen pitches there that I had never seen in my lifetime before. The valuations that I'm buying Zscaler and Amazon— Amazon cheaper than Microsoft, cheaper than, than Walmart 2 weeks ago, now probably about the same multiple because it's had a nice bounce and Walmart hasn't. Um, yeah, I say no.

I think Walmart's still on like 40 times earnings or something, and Amazon's 25, 30. There you go.

So they both sell chips. Yeah, listen, you know, Amazon— we should just mention that Amazon CEO last week did give an interview, maybe CNBC or somewhere else, but, um, talking about that if they were actually selling their Trainium chips to the market, there'd be a $50 billion run rate, right?

No.

Are you kidding?

I'm gonna look and see how big that would be among semiconductor stocks.

I bet that's top 6 in the world.

Um, it would put them 8th. Yeah, right behind Intel, in front of Qualcomm.

I pictured in front of Intel, right behind Intel, but I thought Intel might be 5th.

Cody, what are AWS revenues as compared to Amazon retail?

Oh, Bryce, correct me on this stuff, but I'm thinking 15-20% of total sales, but of margin, it's like 95% of the company's margins. AWS. Yes, AWS is because it's such a higher margin than You know, it's, it's got to be bigger than you think.

It's $128-129 billion of revenue. And the whole company generated $700— like $600 or $700 billion. Yeah, I don't know where my numbers are at. $750 billion. And the profits are See margins. $37 billion in operating profit on AWS out of— yeah, out of 80.

So it's more than— it's about half the profits and 15 to 20% of sales.

They've gotten the regular retail business to be quite profitable. They have. With— they like third-party selling has grown tremendously. And they charge a lot of fees for distribution, all kinds of services and things. I mean, they get like a 50% margin on third-party.

I'm still bitter about the fact like 20 years ago, they introduced Prime. And they were like, it's $7 a month. And you get next-day delivery even to rural New Mexico. Greatest bargain ever. Now I pay $25 a month for Prime, and it's one week. It's like 4 days, work business days, to get things delivered to rural New Mexico.

Oh, it's supposed to be 2 around regular places, right?

Yeah, I mean, lots of— most of the stuff I order in Las Cruces, it's next day. Yeah, welcome to—

is Las Cruces ain't exactly metropolis, but it's certainly not—

but they built like 2 giant Amazon distribution warehouses in El Paso, which is 30 minutes away. That would do it.

Yeah, 2.5 hours from where I am.

Right. And the other— Prime, Prime generates, by my estimation, about $50 billion of revenue and subscription revenue.

Like, you know, you talk about their, their processors and if they were involved in that regard. So what are they in that's relevant? They, they have huge, like, robotic automation in their warehouses. They have, um, tons of, uh, cloud, you know, which means that they have data centers. Um, and well, AWS, we were talking about it, but they have even the ones that, you know, they run for the rest— running the rest of the business.

The whole reason they have the AWS business is because they built all of the data centers to run their business. And most of the time, except in rush hours, rush days, it would be idle. Yeah. Nice. Head of cloud at Amazon told Jeff, like, 20 years ago, he's like, you know, we got to do is like rent out the excess compute. Amazing. Yeah. Like, yeah.

And yeah, Amazon is probably the largest robot company in the world. Yeah, they have over a million robots in active deployment across their warehouses and things.

Not to mention drones and delivery, right?

Oh, yeah. Geez. Yeah, they're, they're just, they're in a lot. They don't come up in the conversation as much. But because they're more on the side of using it themselves, except, you know, the cloud, which is for everybody. You know, John, it's funny.

Amazon comes up a lot for us. It is front and center on our brains. Um, from cloud to robots to just internet app dominance. And it doesn't come up in this conversation as often. I stand corrected. You're absolutely correct on that. Yeah, Microsoft a lot more than Amazon. Yeah, even though we think Microsoft certainly not as well positioned. Spoke—

okay, not speaking of not as well positioned, We talked about a little last time, but Anthropic and OpenAI. OpenAI seems to be really hurting. I don't think they're going away anytime soon, but they're definitely in this kind of in-between phase where they're trying to figure themselves out.

It feels like they're on their heels playing defense, and Anthropic, despite the— you wouldn't have guessed this a couple, 3 months ago when they had the big throwdown with the Department of War and getting banned from government contracts, that 6, 8 weeks later, like, the, the vibe, the trend, the story, the narrative is all Anthropic and, oh, OpenAI, what's going on with them? And I agree with you, OpenAI, I've said all along, I think it'll survive. It'll be around from decades. I think it'll be better than AOL, but I do think it's— Anthropic looks more like the Google, and, uh, OpenAI is Yahoo.

So with installed users, OpenAI still dominates. Are you like sort of market, market share?

I would think if you went back to like 2003, Yahoo still had bigger market share in search than Google. Okay, it was a better product.

By the way, Anthropic had a meeting at the White House last week. So there's like some kind of rapprochement going on.

Throw out a nice French word, rapprochement. Rapprochement.

Is it just a— I mean, it's kind of— it's made its way into the English vocabulary at this point.

Craig probably knows it because he's Cajun.

No, the Cajuns and the French don't call each other I actually don't know that one, John.

Enlighten us.

No, just, just—

but apparently the NSA is using Anthropic despite what Trump and all of them said. So they said, screw it, we're just going to use the best. John taught me the word gestalt in college.

So rapprochement is the establishment or resumption of harmonious cordial relations between groups or nations, often after a period of tension.

So I think it's the word for that.

So very good word. Well used.

Would it work in crossplay on New York Times?

Yes, it counts as English.

I think I was gonna say bootlicking was my interpretation of what that— but who's—

which one's licking the boot?

I mean, it seems like the White House is licking the boot. Anthropic is licking the boot. Let's be real here.

Really? Yeah, yeah, I think so. Yeah, I think my— I'm a little bit skeptical of Anthropic at the moment. Not because I don't think they're doing well. But like, I think they are just playing— their incentive right now is to sell their stock at an IPO in the next 6 months.

Yes. And the Mythos thing is marketing.

Yeah, those things, marketing, the White House thing, like, they're like the DOD, DOW thing plays perfectly into their like, we're so important that the government is like fighting us, but the Defense Department wants us. And like, it's such great marketing. It's also publicity, right?

Also, they get to come along and they're like, we're the only ethical AI company. Well, we wouldn't let the Department of War do some shit that the other AI companies wouldn't let them do.

Yeah, that's a whole— yeah, we've had some of that conversation. But if you look from a standpoint of though, the frequency of release of useful products. Again, this week they released, um, Claude Design takes on Figma and Opus 4.7. What's that? Yeah, Opus 4.7. But it's like Figma, uh, like, um, Claude Design, people are saying about it like this thing is another leap forward and it takes over another piece of kind of the knowledge worker.

Bryce, what was the Autodesk knockoff they released?

Well, I think it was just OPUS 4.7 integrated into some kind of a wrapper orchestration layer that could do AutoCAD. Yeah, computer automated design. Yeah. So you could just talk—

you were saying, Craig, sorry. Yeah. Yeah, no, no, no.

I was just, I was just agreeing with John, because I mean, you can't keep up with it. Like there's still design that I would like to use and try. And that seems, you know, everybody's talking about Figma, but that was a big—

Great products coming out.

Hey, can I tell you guys about the new crypto application that's so excit— Wait, sorry. Shut up. There's no exciting application. I love Bryce. About a year and a half ago or 2 years ago, he turned one day, he was here at the office and he was like, you know, Cody, we're always desperate for some crypto revolution that is actually like meaningful and it's used. Meanwhile, every day AI is rolling out something that blows our minds. And you fast forward 18 months, 2 years later, and it's still 100% true. It is why AI market caps, valuations, related businesses, private industry stuff are worth hundreds of billions of dollars individually, trillions of dollars collectively. And most crypto is in a winter.

What did I tell you about that Bryce guy? Smart, smart, nice guy.

John, if you were a venture capitalist, would you invest in AI startups? Like, not outside of the labs, outside of XAI, Anthropic, and OpenAI. Would you invest in Replit or Harvey, or I don't know, some of these big rap—

rappers Maybe like, uh, not a Replit, but maybe a Harvey, uh, with— if, if it seems like it's doing a good job of dominating its vertical, um, you know, where you— and it's a vertical that, you know, the other is like, oh, you know, Opus does this and that with law better than it did, but Anthropic is not about to come out with a law product, then I'm like, cool. Go, you know, I'm for it, you know.

John, I think what Bryce's point is that it seems like the AI improvements are getting so darn good within months or a year or two, you're going to have native app LLMs doing better than Harvey can with its— I don't know, did I take words out of your mouth or is that where you were going with that?

Well, I think it's an open question. I don't know that we know the answer firmly.

There's a wrapper and then there's a wrapper of a wrapper. And the first layer wrapper is what we are calling harnesses. And it's like, you know, Anthropic has the best harness. It's beyond their model, the stuff they've got going on immediately around their model. In Claude Code, in Cowork, and so forth, in just the way their chat works, you know, uh, their projects. It's the whole thing. It's just, that's their— what, what they're doing, like, around theirs. And it's just happened to work out this way that the ones who put out the model and the API for that model did themselves go put out the first level wrapper because they do want the, um, immediate kind of install base of direct users. I want the billion users to me, says OpenAI, says Anthropic, and they're getting that. Now you go out one layer from that to businesses and the workflows that have to actually get thought about for, well, we're going to be taking our case to court and doing this and that with it. And, you know, Is it really going to be the, you know, the OpenAI or Anthropic themselves, or is it going to be somebody else who's got that one extra step of, I know what you're up to in your— hey, lawyers, you know, in your work— that I'm going to go do it for you. I'm going to have this be the, the Salesforce of, of law, you know, so that a thing will still exist. That's where I wonder on that piece of it.

Which I think is literally how Bryce framed it.

Yeah, I think— oh, I already said that. One of the ideas I just had was like, if you compare AI to electricity, you wouldn't want to invest in like, the utility companies that resell electricity, typically. I mean, that's not that great of an investment. You know, there's so many of them. They don't really have an edge. I don't know. But like, the companies that made the appliances, like General Electric, did very well over time. And maybe in AI, like, almost there, like the hardware layer is going to be a great place to invest. And that's been different than how it's been for a long time. And robotics might be a great spot to have a durable edge for a long time, because you integrated AI and hardware is hard. It's like, I don't know. I mean, I just, maybe, maybe there's some analogy there. I don't, I just, I don't, I don't think there's a clear answer on what, how this all shakes out yet.

I have a proposal. I think we should change the term software to easyware. I'm actually being serious. Like, AI makes software easyware, right? And robotics remains hardware.

I think we should call it macro hard. Oh my gosh.

Last words, Craig.

Um, just to kind of go off what we're just talking about, I think Fiscal AI is a tool that we use. It's one of the best examples of what a wrapper should and can be. It uses multiple models, smaller models for just LLM summaries, and then larger models for actually putting cool information on your screen as lightning fast as possible, lets you search, and it's absolutely incredible. And that's what a— that's what a tool should be that's powered by AI. And I don't think a lot of the other legacy software companies are harnessing AI in that way, but Fiscal AI is natively AI, and they just done it so well that I, I think that's what I want to see more going into the next phase of of the AI revolution, and those are the software, the easywares that are going to make it in the, uh, in the corporate world.

I think I, I applaud that answer. Very fantastic. Yeah, fiscal.ai, if you guys haven't looked at it, if you're in this business of running money or even investing your own money, let's check out fiscal.ai. It's replaced Yahoo Finance, Bloomberg, A lot of tools for us. Really good point, Craig. Very good harness/wrapper. Bryce.

I'm gonna pass. I can't— I don't have a punchy— I'll have one in a second. Let me think for a second. He said enough.

No, there's never enough Bryce. We always need more Bryce. More. More. How much price is enough? More.

John, what? Um, I have one left over from, uh, the John Ternus, uh, conversation, um, where— because I was asking, uh, Claude about him, the new CEO at Apple. And actually, I think he has more to do with the, the new chips than we were giving him credit for. Uh, yes, he's not the designer of them, but he was a key champion for, uh, them to, um, kind of go, go back in-house and to make— I said, it said that the, uh, bringing technologies in-house was one of the most profound shifts in Apple's thinking over the previous 20 years, and he was a key person in that.

Really good insight, actually. And I think my— I need to do some work on this guy because if he's as brilliant as the guy who probably made Apple's M chips happen, that is something exciting for their AI future because that really is what Apple, the stock, the story, the company is going to be all about for the next 5, 10 years is how are they going to harness Apple, uh, AI. Harness— I used it twice.

Especially with the new form factors, I'm sure he's all over that being a hardware guy.

Amen. My— that's a good point, Craig. I just thought of my last word. I talked to Cody Willard about this last week. Usually my last word is very optimistic and bullish or something, but I am slightly, tiny bit worried about Nvidia. And Jensen Huang gave an interview with Dwarkesh Patel, who's a big tech podcaster. He's a pretty smart guy. But he was just a little bit— Jensen was a little bit defensive, was dismissive at times, quite argumentative. It just didn't come off as like secure, confident. I don't care what anybody says, this is the right way. And I, you know, he was very defensive when it came to talking about the potential for custom chips, which ties back into the Apple thing. Every customer is developing custom chips outside of the NVIDIA ecosystem. And China, he's lost, basically had to been forced to concede the Chinese market. So there are some risks there. And I'm still confident, mostly confident in NVIDIA, but I did find the attitude of Jensen slightly concerning there. And maybe he just had, was an off day. But I will just— I just want to put it in the public domain that I am a little bit concerned about NVIDIA for the first time in a long time.

It's interesting the signals that you pick up on because that's kind of impressive to me on part of you kind of watching it. And it's not just what's being said, but you're like, look how he's saying it and how he's answering. And good for you.

Yeah, it's interesting. You're not talking to somebody who woke up a loser.

Yeah, that's what he— one of the things he said. I don't know, have you watched the full thing yet, Cody? I haven't finished it.

I started it because you made me. Um, I'm a reader, not a podcast listener. My final thought is this: bought this at, uh, dollar store with my daughter the other day.

What is he protecting with the shield at the moment? Its location seems a little Knees.

Okay, knees. Business. Um, and then, uh, this is the future. I bought this from Tesla a long time ago. Not really any deep thoughts on that, just my final words. Thanks everybody, goodbye.

You gotta do the closing, Craig, you're up.

Oh yeah, you got to do the closing. How do we do that?

Just hit on stream, it'll do it.

All right, here we go. Thanks everybody. Thanks, Johnny. Thanks, Craig. Thanks, Bryce.