The Wobbly House of Cards (EP. 483)
Animal Spirits Podcast
The episode explores the massive scale and interconnectedness of AI hyperscalers, debating whether they represent a fragile 'house of cards'
Key takeaways
- The combined market cap of major hyperscalers like Nvidia, Microsoft, and Alphabet is estimated between $20 trillion and $25 trillion.
- Nvidia's forward P/E ratio is currently lower than the broader stock market's forward multiple, despite AI hype.
Main topics
- AI Hyperscaler Interconnectedness
- US Household Wealth Trends
Notable quotes
The market does plenty of worrying for me.
Conclusion
While systemic risks exist due to tech concentration, massive corporat
Transcript preview
Speaker 1 (0:04) This podcast is sponsored by Nuveen. As year-end approaches, investors are assessing their portfolios and considering how to move forward in 2027. Nuveen offers custom portfolio insights and guidance, helping investors deconcentrate exposure, offset capital gains distributions through taxless harvesting, and explore municipal strategies that can offer tax-advantaged income. From helping to manage risk, improving tax efficiency, and finding new opportunities in key asset classes, Nuveen's expertise across public and private markets helps build portfolios that seek resilience for what's ahead. Nuveen, positioning for the year ahead. Visit nuveen.com slash positioning to learn more and to sign up for our custom portfolio analysis. Investing involves risk. Principal loss is possible. Private market investments may not be suitable for all investors. Nuveen does not provide tax advice. Diversification does not ensure profit or protect against loss. Speaker 2 (1:01) Welcome to Animal Spirits with Michael and Ben. Michael, we've got a huge doc today this week. Speaker 1 (1:05) You are looking very GQ today. Speaker 2 (1:09) Well, it's a great time of year. It's bomber jacket season. Speaker 1 (1:12) Oh, Speaker 2 (1:12) I do love the bomber Speaker 1 (1:13) jacket. I have one. Speaker 2 (1:13) I get like two months of the year in Michigan where I can wear a bomber jacket. It's September and April, essentially. Well, Speaker 1 (1:21) you Speaker 2 (1:21) look great. Thank you. Taking advantage. All right. I want to offer, you know, I like to do both sides of things. Right? So I think the one AI-related, there's a lot of AI in the doc this week. It has to. Sorry. It's like the biggest story of our lifetime right now. We have to talk about it. So the FT has this huge new tool that allows you to see how interconnected the whole hyperscaler thing is. So it's kind of cool. You can click on one company and see who are they tied to. And they're all tied together. And they look at Anthropic, OpenAI. Alphabet, Amazon, Meta, Microsoft, Oracle, SpaceX, and Nvidia. And they threw Corweave in there too. So they talk about how interconnected all of these big companies are. And the idea is, okay, this is all a house of cards. Because I looked at, I think the combined market cap for these companies that they list is like 20 some trillion dollars, 22, 25 trillion dollars, something like that. So a lot of people think, okay, if everyone's in the pool, when this house of cards falls, this is going to be really, really bad. Speaker 1 (2:23) Well, because it's like, You invest in us. We'll loan money to you. We'll co-invest in this together. Speaker 2 (2:29) Yes. Yes, the fact that your level is rising helps our earnings and our cash-free cash flow is going to your profits and it's all interconnected. So Speaker 1 (2:38) the flywheel is spinning, but once it slows down or potentially reverses, we're all going to die. Speaker 2 (2:42) But my other side of this is these are the biggest companies in the world. Why isn't this actually a good thing? That they are all doing this at the same time. Why couldn't this essentially be, they are big enough and strong enough as companies to keep pushing through all of the worries people have and go, you guys seriously think we're going to slow down? We're worth like 20 some trillion dollars. What do we care? Speaker 1 (3:10) It sure does seem to be the consensus view, what you're describing. Bloomberg ran an article over the weekend with the headline, AI's wobbly house of cards puts markets and US economy at risk. They quoted Jim Morrow, the CEO of a Boston investment firm, Caledon Capital Management. He said, people may not fully grasp just how wound up the market and the economy is in all of this. There are just so many things to unravel if it starts. Actually, With all due respect, Jim Morrow, I think people are extremely, they do fully grasp. This is all we're talking about. I think everyone gets it, right? Everybody gets it so much that the AI trade, tell me if this is fair, the AI trade is trading at 16 times forward earnings. I'm talking about Nvidia. So NVIDIA is the AI trade. Now I know there's CoreWeave and NeoClouds and whatever, whatever. I understand that it's not, there are semiconductors, there's other things. But if you just want to zoom in. Speaker 1 (4:15) First principles. Is that even right? I hate that term. You know what? I said it. I hate it. I hate people who say it. It is stupid. It is filler. I said it to be ironic. It makes you sound smart when you say it, though. It's nonsense. It's a jack-off term. That's what it has been. The AI trade is trading at 16 times forward earnings. Speaker 2 (4:34) It is wild that NVIDIA now has a lower forward PE than at any time this entire decade. Speaker 1 (4:39) The market, the stock market is trading at a higher forward multiple. than Nvidia. There is so much doubt. Maybe warranted. I don't know. But there is so much concern that this house of cards is going to fall upon Speaker 2 (4:56) us. I'm not saying I necessarily believe this, but people always talk about how this is like the railroad build-out in the dot-com bubble. Those are the two biggest analogies people use. The railway build-out, which I did some pretty extensive research for one of my books. It was all funded by the public. They were the ones who were putting- Was it Speaker 1 (5:15) extensive? Did you go to primary sources? Yes. You revisited newspapers from 1873? Speaker 2 (5:22) I Speaker 1 (5:22) read like four books. It Speaker 2 (5:24) was funded purely by the public, essentially. The public put all the money in. It wasn't even the government that did the build-out of the railroads. It was the public who funded it, and they lost their shirts when all the railroad stocks went under. And that was just a lot of fraud as well. The dot-com bubble, that was also kind of the public. All those IPOs were being funded by individual investors. Today, this is being funded by free cash flow, first of all, and now potentially debt from these companies. These companies are the ones that are funding this. It's not like the public is holding the bag. Well, you could say, well, index funds, but still, that's the big difference now is that these companies, they're not relying on the public for funding. Speaker 1 (6:04) Ben, maybe chart of the year. Also from that Bloomberg article. And we've seen this before, but it's just, I'm staring at it and my face is melting. It's unbelievable. Yeah, but this one Speaker 2 (6:13) has a purple line on it. The other one's not a Speaker 1 (6:14) purple line. That's true. So we're looking at the CapEx of Amazon, Meta, Microsoft, and Alphabet, estimated to be a trillion dollars next year, overlaid with their trailing 12-month free cashflow. And this... Speaker 1 (6:31) was supposed to be the thing that put the emergency brake on the stock market. Holy shit. The reason why all of these companies were such amazing stocks is because they're incredible businesses. Just free cashflow out the ass, asset light, now they're asset heavy, a total re-rating, 40 % of the index. This was going to slow the stock market down. And then it's sort of, I mean, I guess it... The max seven sort of stalled out in the first half of the year. They did. The investors were digesting what this means. Okay. But we got past that narrative. And then I think a couple of weeks ago, we were like, dude, there's nothing else to talk about either. Slow down, keep going. But like, I feel like we're sort of in this in-between world where we're just having the same conversation every week. And then we got the slowdown potentially, or at least the whispers of it with the guy that said, the guy that left open AI and said, We're all going to die, 10 % chance. And now it's supposed to crash the stock market. The slowdown, that was supposed to do it. And what has the stock market done since? Look at the semi-trade. It's ramping back up. What is it going to take? The stock market is not going to come down until and unless, I don't even know what, until everyone discovers that it's a bubble. Like, I don't know, Ben. Speaker 2 (7:52) So I guess the Fed would be one thing that you could say could prick this bubble. The Fed keeps raising rates. Maybe that's pretty much every bubble in history. That's what finally got Japan. But getting back to the free cash flow, if you look at just that chart, you go, well, OK, they're going negative forever, right? They're not there. Where is their free cash flow coming from? Torsten Slock says consensus for these hyperscalers is they're going to grow operating cash flow from $600 billion in 2025 to $2 trillion in 2030. Like the return is going to start coming. That'll work. Right? Getting back to the size of this stuff, Steve Ratner did this chart. He showed the projected IPOs for OpenAI, Anthropic, and SpaceX. SpaceX already happened, obviously. At $5.2 trillion to those three companies. Now, this is not inflation adjusted, but it's bigger than all tech IPOs from 1980 to 2025, which is almost 4,000 companies. This could do it. Speaker 2 (8:46) But again, I keep coming back to the size of this stuff. I don't think people really have wrapped their heads around how big and powerful these companies are. You don't? Because I get there was a handful of big companies in the dot-com, but it was all these IPOs of these companies who had no power to do anything. They didn't make money. They didn't have profits, essentially. Speaker 1 (9:11) Well, I mean, that's not true because the biggest, Cisco, one of the biggest stocks back then, was certainly a legitimate company. But not as powerful as the companies are today. Correct. Cisco back in the late 90s. Speaker 1 (9:25) does not compare to the impact that all of these companies have on our daily lives. It just Speaker 2 (9:30) doesn't. Speaker 1 (9:31) So I'm still having Speaker 2 (9:32) trouble getting used to having conversations with AI, right? I'm talking to Claude and I'm running all these numbers through Claude and saying like, all right, what's the best, what's like the base case scenario for Wall Street? And what do you think the base case is? So they gave me three scenarios for how this plays out. Number one is you get like a trillion dollars in revenue. The ROI is huge and the bull market is justified. Okay, that's the Goldilocks scenario. Two is the technology works, but it's commoditized. The economy wins, but they don't get as big of an ROI on the spend because it just turns into a commodity immediately. And three would be, okay, the growth falls short and then things get written down. That's the bad. So kind of, right? I promise Speaker 1 (10:15) we're not going to make this a habit of asking Claude what he thinks about the stock market. True. You promise, Speaker 2 (10:21) Ben? You will never Speaker 1 (10:22) do that again? Speaker 2 (10:23) everyone's doing this. But don't you think that those are the three main scenarios people are thinking through right now? And it's like the easy muddle through one is, okay, this is just, everything becomes a commodity. In that situation, doesn't the rest of the stock market win and the mag seven kind of underperforms? Speaker 1 (10:38) I don't know. Too many variables to make that prediction. But I do think that that is where consensus landed on the conversation that the whole world was having last week about we need to slow down. It was like, this is such obviously. Such obviously nonsense that the frontier models are afraid of the open source ones catching up. And look what Meta just did as an example with their Muse agent. Speaker 2 (11:05) So Speaker 1 (11:05) this was my next topic. Speaker 2 (11:07) Yesterday, Meta was up almost 12%. And it's kind of crazy that for as long as we've been in this AI boom, that AI news can still impact a $2 trillion company this much. Like look at the stock price of Meta in the last, I don't know, two weeks. It Speaker 1 (11:25) doesn't just demonstrate how much. Speaker 1 (11:28) Not confusion, but just how early we are. There's no, the range of outcomes is 10 miles wide. Speaker 2 (11:36) So this guy on Twitter posted that he had a flight delayed seven hours. He asked Muse to file for compensation. Five minutes later, he had $250 credit in his Delta account, found and booked him a new flight. It just figured everything out, even responded to support email itself. This shit feels like magic. To me, this feels a little bit like, remember how nervous people were to pay with a credit card on the internet when it first came out? To me, this is what working with an agent feels like. So I played with Muse a little bit today. It says like, here, with your approval, I can update your calendar. I can make purchases. I can connect the apps. I have my own computer, the web browser, so I can keep doing things when you're away. And I asked it, what can you do for me for travel and all these things? And it does sound kind of magical, but it's like, hey, I can manage your inbox for you. I can add or schedule events for you. I can go find something you need to buy. All this stuff. It feels like there's going to be certain people who would, the tech people are going to adopt it immediately. Speaker 1 (12:32) So there's going to be Speaker 2 (12:33) other people who are going to be very worried Speaker 1 (12:35) about, really, I'm going to let you into everything. This morning, I got a text from instinct, which is, uh, the other AI assistant. And it said, heads up from the inbox. Your son's second tryout today is at 6 PM at the blah, blah, blah. And he needs to wear the reversible Jersey from first tryout. It's not on your calendar. Want me to add it? So I walked downstairs and I showed Robin. I said, hey, you probably should download one of these. It's super helpful. She's on the phone with her friend. Her friend said, I'm not doing that. And I made the point, like the credit card. This is just the way the world is. And there's going to be a lot of people that resist it. Probably in five, 10 years, everybody will be using it. I feel like the AI stuff is quickly becoming my personal fantasy football. Meaning, so Chris gets very excited about these things, as do I, but he has to share every. detail of his Speaker 2 (13:25) journey. Speaker 1 (13:26) And I said, dude, guess what I did today? I said, I promise you, I will not share the things that I'm doing with AI that are very exciting. If you stop sharing with me, I don't, right. It's like, I don't care about your gambling. I don't care about your fantasy football. I'm glad that you're having fun. I'm having fun too, but it's your own fun. Speaker 2 (13:44) There is going to be so much of that. Can you believe what AI did? Let me just Speaker 1 (13:48) stop you right there. Speaker 2 (13:49) We're Speaker 1 (13:50) all doing the same thing. Speaker 2 (13:51) So Ben Thompson wrote about it for Stratechery today. Why do we say Speaker 1 (13:56) Stratechery? I think in my defense, I say it because you say it. But it's obviously, tech is obviously tech, right? Because Speaker 2 (14:02) I'm holding on to Will Ferrell as George W. Bush from SNL saying Stratechery. I'm putting Stratechery and Stratechery together. Okay. So he says he thinks Meta's launch of Muse is a bearish signal for the Frontier Labs. Essentially saying this is a very approachable personal agent product and the LLMs, you're not really wedded to them in any way. But if you have your own personal agent who's in your inbox and knows your stuff and knows everything about you, that's how you get. So he's saying this is kind of bearish for OpenAI and Anthropic. And his sort of conclusion was. they need to start having, creating products with moats. Like, okay, you've done this. Now, where are the agents that give you the moat to keep people there? Speaker 1 (14:49) Yeah, I feel like the switching costs to this are like nothing. going between, if you go from Instinct, which by the way, I didn't, is Instinct one of these companies that raised a couple hundred million dollars that might be out of business overnight? I'm, I know nothing. So this is pure conjecture, but just again, the thing about how fast things are moving, how narrow the moats are, because I don't see why I can't switch from one to the other or back to Anthropic if they do something for me. Well, but his point was Speaker 2 (15:17) once you get your own agent and it starts knowing you and you have it locked into your inbox and your accounts and your apps and all this stuff, it's, it. creates more of a moat. I don't know if that's true. I don't know how he's partisan. It's hooked into your Delta account and it knows all of your preferences for Delta. You're right. He's saying that does create some sort of moat maybe. Maybe you're right that it's easy to switch from one to the other. I would imagine once you have your own agent, it's going to be hard to switch. It's like having a dentist. Who wants to switch their dentist? Speaker 1 (15:44) That's a strange analogy. Dentist. Have you ever tried to switch a dentist before? It's not easy. Confession. You got to send your x-rays to the other one. So my father is a periodontist, or he was. He's retired. A Speaker 2 (15:57) periodontist. Which is, sorry, someone said when I said braces are a racket and you defended him. No, no, no. That's Speaker 1 (16:04) an Speaker 2 (16:04) ortho. I have no Speaker 1 (16:04) allegiance to orthos. It's all in the same mouth area. So I am a terrible dental patient. I have very sensitive gums, which is probably not a great thing. So I read, I don't get my teeth cleaned nearly as often as I should. And when I say nearly as often as I should, it's probably been two years. I do floss. Michael. Multiple times a day. But I hate going to the dentist. Well, no one likes going to the dentist. No, I hate it. I'm sweating right now thinking about it. Speaker 2 (16:32) Just talking about it. I didn't. The first three years of my job, I had no health insurance, no retirement plan, and no dental insurance. I was like on my own. It was a very small company. And I didn't get my teeth cleaned for like three years. And the first teeth cleaning afterwards when I got a new job was horrendous. All right, let's move. Different topic. I can't take it. Although, Ben, you saw me squeamish at Future Proof. That's true. There was a booth at Future Proof that gave IVs for energy boost or pain. And someone said, you guys have to get an IV. It's magical. And you were doing this, looking away, because you couldn't see the needle go in. And I got to tell you, I'm someone who... This is going to sound very odd to say. I feel like medicine doesn't have an effect on me. Like minor medicine. Like if I take ibuprofen or Tylenol, I feel it doesn't, I never feel better or worse. I feel the same. You know, that's an interesting take. Speaker 1 (17:26) So I wouldn't go as far to say as medicine doesn't impact me because that's a bold claim, Cotton. But I always feel the same too. Like I, so I was, I was a pothead for a long time and I no longer am. And it's been. several years, but it was just part of my life. Self-medicine, I suppose, not a great thing. But I was talking about it with somebody recently. I said, do you feel different? And I said, I don't know. I don't think so. I think I feel the same. That can't possibly be true because marijuana is terrible for your overall everything. But I kind of feel like I always feel the same. But my point is Speaker 2 (18:05) - Speaker 1 (18:05) Not Speaker 2 (18:05) to steal Speaker 1 (18:05) your thunder, but I'm the same, same guy. Speaker 2 (18:07) Yeah. If it's like big medicine, then yes. But cold flu medicine, none of that. I feel like none of it touches me. But the guy at the IV, because people said, hey, get an IV. We're in the sun all day. We're drinking at night. You know, we're eating, like we're kind of tired. We're doing a lot Speaker 1 (18:20) on our feet. Yeah. It was a very busy Speaker 2 (18:22) time for us. And someone said, get an IV. It'll make you feel better. And so we did the B12 one. And the guy said, give it an hour and it'll kick in. And honestly, an hour later, I felt invigorated. It actually worked. I couldn't believe it. We hopped in the Osh. Yeah. Maybe that was part of it too. Nature. Speaker