Prof G Markets - Aswath Damodaran: Big Tech Has No Idea How AI Pays Off

Episode Date: August 7, 2026

Ed Elson and Scott Galloway are joined by Aswath Damodaran to break down the biggest takeaways from Big Tech earnings. He explains what the latest results reveal about the AI race, why he's becoming increasingly concerned about the Magnificent Seven's AI spending, and how he values the hyperscalers. They also discuss how much AI risk is already priced into the market and whether SpaceX's current valuation is justified. Subscribe to the Prof G Markets Youtube Channel  Check out our latest Prof G Markets newsletter Follow Prof G Markets on Instagram Follow Ed on Instagram, X and Substack Follow Scott on Instagram Send us your questions or comments by emailing [e-mail] Learn more about your ad choices. Visit podcastchoices.com/adchoices

powders, pills, plunges, everywhere you look, there seems to be a new wellness trend. We all want to feel good. And if there's a way to feel better, I think wouldn't you want to try it? I would. What's the cost of being well? And why are we so obsessed with it in the first place? That's this week on Explain It to Me. Find episodes every Sunday wherever you get your podcasts. Today is number 48. That's the percentage of U.S. adults who now support the idea of banning phones for the full school. Ed, why did the smartphone need glasses? Why? Because it lost all its contacts.

How are you, Ed? I'm doing well. I've got to tell you, I'm excited for vacation, which is happening for me in about two weeks. Where are you headed? Oh, wait, we talked about it. You're going somewhere fabulous. I'm going to Austria. I'm going to a wedding in Germany, and then I'm going to drive to Kitsbule in Austria.

Oh, Kitswills. I've been to Kitspiel. It's great, quite frankly, it's great for kids. I'm very excited. It'll be very nice. We'll go hiking. We'll play sports, we'll do saunas and swim. It'll be a very health-oriented vacation, which is exactly what I want right now.

It's a wedding, you said? Yeah, I have a wedding first, and then I've attached, sort of bolted on a vacation next to it. So I'm pretty proud of how I've kind of maneuvered this, because we have the wedding in Bavaria, which will be awesome, and then rent the car, drive about two hours into Austria, and then spend the week at a hotel. So I'm very excited about this one. That is exciting. You know what Austria's greatest achievement is. What's that?

Convincing everyone, Hitler was German. He's back. I was waiting for the Hitler. He's back. It's something to do with Hitler. That's actually a good point. Yeah, that's true. They did convince him.

He's an Austrian. That's right. Everything's a German. Nope, Austrian. Nope. Do you have vacation plans? The better question is, do I have work? work plan. I'm on vacation most of the time, as your team knows. Yeah, I'm here in Colorado,

and I head to... Oh, you're on vacation. Well, on vacation. I'm up at 7.30 doing these Joey Bagadona podcasts with you. But technically, I mean, my kids are out causing trouble and bored and doing hikes and shit like that. And I saw a bear last night, actually, in town, which was kind of, which was kind of fun. Brown bear? Yeah, a little brown bear. My son said, should we run? I'm like, no, you stay here. I'm going to run. It was very, very dangerous animals. I saw one of those TikToks about what animal kills the most people, and everyone's so freaked out about sharks.

Moose kill more people than sharks. Yeah, moose. I've heard that moose can be quite aggressive. Yeah, and hippopotamai and crocodiles. Snakes kill a shit ton of people, I think mostly in India. This is what you've been doing on your vacation? Well, of course, you know, the animal that kills the most people, right? No, I don't. What is that?

The amateur answer is a mosquito, but the animal that kills the most humans is other humans. So it's true. We're the problem. We're the problem. The amateur answer is the mosquito. How many times have you had this conversation? I think it's fascinating. Have you been through this? It's really interesting what animals are dangerous and which ones aren't. Should we turn this into an animals podcast? Maybe that's what this should be. Oh, I would love to bring on.

I like that guy, the dog with the Oscar guy, who goes into Holmes, and then establishes his dominance over the chihuahua that thinks it's a... Yeah, Caesar Milan. Is that he talking about? Yeah, yeah, he was really good. Yeah, no, he's great. Despite the fact you hate dogs. That's the underrated side about me. I do prefer cats.

That's not going to help our download. You had a dog, but not a cat growing up, right? I did have a dog. And as everyone knows, he was okay. He was an average dog at best. I think it's more the owner that was the problem here. That's fair enough. I think that's probably true.

All right, well, we've got a very exciting interview to get into here. We recorded this conversation live on Substack earlier this week. And if you'd like to catch the next live stream, we'll be doing plenty more. Then you should head over to profgmedia.com and subscribe to ProfG Plus. Without further ado, let's get into it. Today, we are joined by the one and only dean of valuation professor. Aswath Demoderin for our quarterly review. We will break down the latest earnings from big tech. We'll get his take on SpaceX after its blockbuster IPO and we'll discuss whether we should

be worried about the growing debt behind the AI buildout, among plenty of other topics. So let's get into it. Professor Demoderin, thank you so much for joining us. I'm going to launch us right into it. I want to start with big tech earnings. It seemed pretty good across the board. Microsoft revenue up nearly 20%, Amazon revenue up 20%, meta revenue up nearly 30%. The response was kind of mixed. I have some underlying questions about these earnings, but I'd love to just start with your reactions to what we saw in Q2 from Big Tech. On revenue growth, you're absolutely right.

I think the companies delivered more than expected in terms of revenue growth. That's good news, at least in the near term. on the earnings level, what you notice is it's happening under the surface. These companies are so big that you don't see them. The marginal returns that these companies are making, basically the incremental earnings relatively incremental invested capital, continue to show a shift in business models, which I think investors need to keep their own. It's neither good news, no bad news.

These are very different companies than the company's investors were investing in five years ago. And I think we need to talk about what's happening at least outside of Apple, and what's happening to Mag 7 that's changing these companies and changes what investors need to look at going forward. I mean, the thing that jumped out to me, if we just go straight down to the bottom line, if we go to the free cash flow, META's free cash flow came down 91%.

Amazon's free cash flow went negative. Google's went negative, which is something that I don't know if anyone thought would ever happen. Certainly wouldn't have predicted it several years. just due to how much money these companies make and somehow they found a way to spend it. Does that concern you? It's something that needs to be thought about more seriously. These are long-term shifts.

It's a one-year change. You might say, okay, they had a big investment this year. They're going to go back to being cash cows next year. I don't think that's going to happen at these companies. The one company in the mix that's had experience with this negative to positive cash flow and back again is Amazon. So in many ways, if you're going to pick a company that's equipped to deal with negative cash flows to be Amazon, because they've seen this movie before. They've lived through it. For their 25 years, you look at them, they're in and out of cash flows, and they've

found a way to always get back. For meta, for alphabet, and for Microsoft, this is new territory, something they've never had to deal with. And the question is whether they're equipped to deal with a very different kind of company going forward. And this is in the top down. These are more capital-intensive businesses. As anybody who's run a capital-intensive business will tell you, it's a very different business model, a much more difficult business model to generate value from

than the models that they used to use pre-AI. When you said investors need to be cognizant of the fact that they're investing in much different companies than they were five years ago, can you give us a broad overview or get as specific as you want? What type of companies were these five years ago and what type of companies are they now? These companies five years ago,

if you asked me what their invested capital was, I wouldn't even have cared. Because you knew that they could generate revenues and operating income with very little additional invested capital. Outside of acquisitions, even with R&D, consider these companies generated returns of 70, 80, 90% invest capital.

The only survivor from that group is Apple, which still continues to deliver that kind of return and analysts are not happy with it because it's not investing. The other companies now are the equivalent of manufacturing companies. They're building huge capacity for whatever, AI products and services. And like all manufacturing companies historically, they're now going to be judged on whether they can deliver the earnings on this investment, something they've never had to do historically.

So measures like return and investment capital, it used to be not that useful with tech companies now come into play. Questions are, are you earning more than your cost of capital? A laughable question five years ago with these companies now becomes a relevant question. And I think that is the question on which these companies will live or die. If they can manage to deliver returns that exceed their cost to capital, I think they can come out on the other side as more capital-intensive but still valuable companies. But if they fail, markets are punitive on companies and invests a lot of capital and can't deliver the earnings to justify that capital.

My sense is when the market gets these earnings, it's not a very important. the earnings. It's about the capax and the market's ability to discern and return somewhere down the road on that capax. And if you were on the board of one or more of these companies and you were head of the, you know, the finance committee or the auto committee, obviously every company struggles with the tension between investing for the future and trying to build moats around your business while recognizing, you know, while not getting too far out in front of your skis. Where do you think that kind of that fulcrum or that tension is right now? As you look at these companies, do you agree with the markets right now or recently that the CAPEX has, quite

frankly, gotten a little bit out of control? Or do you think that these guys are in a unique position to do it, so why not do it? I think the lesson that Facebook should have learned from the metaverse investment fiasco is investing is easy. Spending money is easy. But spinning a narrative that markets get of why you're spending the money and what your business model is going to be is just as critical. As an investor in these companies, my concern is not that they're spending money. I think they can afford to spend the money. I can see that they're going for growth. But none of these companies is enunciated what exactly the business model it is that they hope to deliver. I mean, at the very minimum, are you going for scale with low margins? Is this the kind of business

I should be looking at? Are you going to be looking at? Are you going to be? for premium products with high margins and niche markets. What is it exactly you're planning to do? And for the moment, at least, that's not there. And maybe they don't know, but then they need to be open about the fact that they're trying stuff out just as much as the rest of it. It's scary for markets. But you know what?

If you don't say something, markets fill in the vacuum. The fact that you're not being open about your business model, markets look at that and say, hey, maybe you don't have a business model, which is one reason markets have turned increasingly skeptical about the CAPEX. Because you remember only on two years ago, when they started the Cappex was all good news. Look how much money they're spending. The assumption was, hey, they're smart companies. They'll figure a way out to make money. But markets are recognizing that you can be a smart company. But you've got the situational awareness, I hate to bring that in, component of you're smart and perhaps you're too immersed.

in this space to step back and ask the objective question of, is there really a business here that can justify, not a billion, five billion in CAP-X, but tens of billions of CAP-X. And I think those questions are only going to get louder. If I were on the board of these companies or the top management, I'd be thinking seriously about the business narrative end, and not just throwing out the CAP-X numbers and leaving them at that,

because markets are going to continue to respond negatively to big-cap-cap-ex numbers, numbers without a story backing the capex. There's an outlier here. We have a tendency to talk about big tech as if they're all one amorphous blob. And the real outlier, I see is the following. Everyone is spending between that we're talking about today, 150 to 200 billion in capax. But Apple is at 11 billion.

Apple has made a distinct decision to pursue a dramatically different strategy. As far as I can tell, they've said, we're not going to engage in the CAP-X wars of AI. And, I mean, it strikes me as a very big bet. Like, one of them is wrong. Thoughts on that? I agree with you.

This is going to be a classic case study 10 years from now as to whether it's better to weight out the uncertainty and then decide what kind of factory to build rather than build a factory first and worry about what can be produced from the factory. I mean, the analogy I keep coming back to is, is this factory building exercise, which is Apple is saying, we don't know enough about this space.

We don't know yet whether the kinds of products and services that AI will deliver will be low-cost, high-scale products, which will require a very different kind of factory than this premium product, high-margin, lower-scale business. And I think even within the LLMs, you see this fight, you know, with the OpenAI versus Anthropic, Anthropically is going for the premium pricing strategy. And Open AI, surprisingly, saying,

hey, maybe the big market here is to sell stuff for at a lower cost and sell at scale. And the Chinese are, of course, waiting on the sidelines to throw the story into complete turmoil because they can come in. And this briefly, we saw this with Deepseek a couple of years ago, coming in and shaking up the story

to alter everybody. The only problem for Apple is they're facing the side costs of the huge AI CAPEX in the sense that chip costs have gone up and they're facing, I mean, part of the reason they got punished so much was because they had to raise the prices of almost every single device because everything has become more expensive to make. So much as Apple would like to be completely on the sidelines, at least for the moment, they cannot be because they get dragged into the space because of what everybody else is doing. But I think you're right. And I think it reflects Tim Cook's personality, and it'll be interesting to see if continues under the new CEO of saying, look, you know, jumping in with both feet into things you don't know is not the greatest way to

make money. He's a cautious person. And an address have taken issue with Apple for being cautious. And I think it kind of shows up in the way he's chosen to run the business. But I think it indicates why CEOs matter, because a much more ambitious CEO at the top of Apple would have probably charted a different path, and you'd have seen Apple with tens of billions of AI investment as well. Aswood, you mentioned this idea that they haven't, none of these, the hyperscalers have really articulated what the ROI on these AI investments are actually going to be. And when they do report their ROI, it's generally in kind of these as vague a metric as possible. Like Amazon says, oh, we generated $25 billion in AI, ARR.

And it's like, okay, why are we doing ARR? Why don't we just, like, hear the actual revenues? But the worst of all is meta, they're not saying anything. And Zuckerberg was literally asked the question, like, how are you going to generate the return? And he just filibustered. He didn't answer the question. He didn't talk about the cloud plans.

You said something really interesting. You said, maybe they don't know. Is that possible? If this is a trillion-dollar bet, could they really not know? I think they truly don't know. And I think that they're afraid to say that. But I think in this business, they're better off being transparent about what they don't know.

What I'd like them to do is just as they've done with the cloud business, which is clearly a money-making business to that, is to have an AI division, a separation of AI, where they tell you how much they're spending, how much money they're losing, think of it as a startup that they've created with a huge amount of venture capital investment and say, look, we're making a bet

on what we think is going to be in the growth space, like a venture capitalist. We can afford to do that because we have the capital to do it. But like most venture capitalists, we're venturing into the unknown. And they think markets will punish them for being honest.

But I really think markets would welcome that honesty because I think markets increasingly, as they listen to vague answers to questions, said, these guys have no idea what's happening. You know, they don't know what's going to happen, and they're trying to act like they know more than they do, when in fact they don't. It's better to be seen as trying to find an answer than acting like you have the answer,

but you're not willing to give the answer to markets. So I think more transparency would be good for these companies, but I'm not sure they will take my advice on that. It seems that the lack of transparency and the unwillingness to answer the question, to your point, it makes me more anxious. It makes me think that they're not telling us something because if they tell us the truth, then the business models don't work anymore. And there's one piece of data that I'd love to get your reactions to.

Something I'd love to know the answer to is what share of their AI revenues are coming from Open AI and Anthropic to very, very big companies who, whose financials we know to be shaky at best because they're highly unprofitable companies. We don't know the answer to that question fully because they haven't told us, but there have been some estimates from some Wall Street research. Barclays estimates that Open AI and Anthropic makeup, 73% of Amazon's AI revenue. Wells Fargo thinks that Open AI and Anthropic makeup,

74% of Microsoft's AI revenue. UBS had an estimate for Google as well. I mean, the point being highly reliant. How big of a problem is that if all of that is true? It is a big problem because almost all of the revenue you're talking about is intra-company revenue building the factory, building the architecture. It's not revenue from end users. And that's really the part that we're uncertain about, right?

I mean, you can keep spending more money on the factories. Invidia sells chips and you pay for the... I mean, the intra-company revenue just reflects the fact. collectively of building the biggest architecture businesses ever known. But to do what? If people are not buying your end product and services, what difference does it make the degenerated revenue from each other? So I think that it would be useful to actually get a sense of that end revenue.

That's the part we're all seeking out is Anthropic might be the one company that you can talk about end revenues because the LLMs and what they generate. It's a fraction of what you think of as total revenue from AI, and it's a small fraction. And for this to be a healthy business, that's got to be the driver. The architecture can't be 80% of your revenues. In a healthy business, the architecture has got to be 10, 15, 20% of your revenues. The rest has to come from end customers. End customers can be businesses, they can be individuals, but that's not what we're seeing

right now in the AI space. And we're not getting a sense of whether that's building or not, other than through anecdotal evidence, which is the worst kind of evidence we can get. Of somebody saying, I use Claude and I save $300 million. Hey, that's great. But what does it tell me about collective revenues? And $300 million is a drop in the bucket when you're spending hundreds of billions in building this architecture. So I think more transparency all the way around would be helpful here. because, and I speak as somebody who's an optimist on AI product,

I think that there is a market out there. It's going to be a big market. But I'm not getting any sense of clarity on that market from all of these companies reporting on that space. And I wish I had more clarity. And I think, I know, the Pellantir earnings are in a sense. It's one of the few companies that can actually say, look, we're making money on selling stuff to people with AI.

built into it, rather than selling to other companies, building more of the AI architecture. So I think that that's the place where I'm looking for more clarity is that end user revenue. And I'm not getting that from any of these companies yet. We'll be right back after the break. And if you're enjoying the show so far, send it to a friend. And please follow us on YouTube, Spotify, or wherever you get your podcasts. Support for markets comes from Anthropic. Your day-to-day task can be impossible to wrangle with data and to

do lists spread across different tabs in different windows. But with Claude, you can have access to all the information you need to face the complex problems your day brings and solve them. Cloud is the AI for minds that don't stop it good enough. It's a collaborator that actually understands your entire workflow and thinks with you. Whether you're debugging code at midnight or strategizing your next business move, Cloud extends your thinking to tackle the problems that matter. You can delegate tasks of all sizes from tedious bug fixes to new future ideas, then handle the other work while Cloud runs in the background. Cloud has checkpoints that let you save progress

and roll back to any previous state instantly. Plus, you can set up workflows to run on a schedule such as daily reports, weekly summaries, and recurring data polls that can help your day-to-day operations. For problems worth solving, get started with Clod at cloud.aI. Slar. Markets. That's clod.aI.S. Markets. And check out Cloud Pro,

which includes access to all of the features mentioned in today's episode. cloud.a.ai slash markets. Support for the show comes from BCX, the public ticker for private tech. For generations, American companies have moved the world forward through their ingenuity and determination. And for generations, everyday Americans could be a part of that journey through perhaps the greatest innovation of all, the U.S. stock market. It didn't matter whether you were a factory worker in Detroit or a farmer in Omaha. Anyone can own a piece of the great American companies. But now, that's changed. Today, our most innovative companies are staying private rather than going public. The result is that everyday Americans are excluded from investing

and getting left further behind while a select few reap all the benefits. Until now. Introducing VCX, the public ticker for private tech, now available wherever you buy stocks. VcX by Fundrise, gives everyone the opportunity to invest in the next generation of innovation, including the companies leading the AI revolution, space exploration, defense tech, and more. Visit getvcx.com for more info. That's getvcx.com. Carefully consider the investment material before investing, including objectives, risk, charges, and expenses. This and other information can be found in the fund's prospectus at getbcx.com. This is a paid sponsorship.

Hi, everybody. It's Megan Rapino. I've been thinking a lot about this one question I've been asked over and over. A question about the choices I've made, the colors of my hair, the things in the world I've spoken about, and the things that I haven't. I've heard this question asked so many different ways, but it always came down to why are you like this? And as you know, there's no simple answer because people are not simple. We're messy and complicated and contradictory and layered.

So on my new show, I'm sitting down with the ultimate disruptors, the athletes, the artists, the activists, and the architects of our culture who looked at the way things were and asked, why do we do it like this? And can we do it differently? My hope is to give us a little more space to that question and to the person on the other side of the mic,

not to tell us their answer or read from their script, but to take us on their journey. Check out my new show, Why Are You Like This, on YouTube or listen in your favorite podcast app. New episodes drop every Thursday. We're back with ProfG Markets. It seems as though these big tech companies were looking for a growth engine.

They saw AI happen. They invested huge amounts of money into basically two companies, Open AI and Anthropic, then Open Air and Anthropic turned around and sent the money back to them, buying all of the compute from the data centers that they're building, and that is now how they grow.

And that's what we're seeing when the numbers explode. That, to me, signals something very unhealthy that we are, and I thought about you a lot when I was looking at these earnings, the idea of the mature company versus the young company. And it seems as though these companies are trying to sort of artificially inject this Botox into themselves to pretend that they're young again. Is that not what's kind of happening here with Big Tech?

They are middle-aged companies. And I've been saying that for a while for the Mag 7, middle-age is not bad. They're in great shape. They make a lot of money. They're good middle-aged companies. But you're right. Nobody wants to be a middle-age company,

especially these companies, with a history of growth. And there are two things that drove them into AI. One, of course, is greed, the fact that there's a big market and they can find growth there. But I think you can't underestimate the fear of fact that the fear of being left behind of history books saying, alphabet got into AI, but meta was slow to join in.

You can almost see that playing in the way they talk and the way they invest is they don't want to be pointed to as the company that's not keeping up with everybody else. There's an element of the emperor's new clothes here, right, which is, you know, we talk about AI. It's going to be a big market. We all buy into it. And they kind of hang out with the crowd where this is accepted wisdom.

You know, conviction that AI is going to be a huge market is deeply embedded. You know, and I think when that's deeply embedded, you don't want to be the company or the CEO of the company. And I put it right to the top here, the CEO of the company that falls behind. and I think that's a big part of what you're seeing in this race to spending is not that there is a belief that they're going to make money, but the worry that they would be left behind in this race and be the company that didn't catch that race.

I mean, we often learn the wrong lessons from looking at history. A lot of chip companies look at Nvidia, and they say only we'd been like Nvidia, think of how much money we could have made in the AI chip business, or and I think that the invidia lesson for a lot of companies is if you bet big on a growth business and it works out you're going to be this insanely valuable company but that's betting that's not investing you can't run a company betting on the odds that you have right now and collectively that's why I think they're going to collectively I think they're overinvesting

I don't think there's a question the question is whether there'll be one big win and lots of big losers or how big the losers will be or how this business will shake out. With the worry that somebody outside the space who didn't invest like they did could be a late entrance into space, but because, you know, it's like, you know, every time I take the train from San Diego to L.A., I curse the fact that U.S. railroads were built 100 years before everybody else because they were built for trains that could go only 25 miles an hour. And by the time you got around to putting fastest speed, you couldn't take, my worry is you're building air architecture,

and you might find out three years from now that you've been building for the wrong kind of product and service. And that's the charitable view you can take of Apple, is they have the money to spend if they wanted to, and maybe they'd rather wait for this turmoil to clear up. and people to find that end game before they jump in with tens of billions. But I think, I mean, uneasy is the word I was used. Worried might be, you know, might be the follow-up to that.

But I'm uneasy about the way they're spending money. And if somebody owns five of the max, I don't own NVIDIA anymore or Tesla anymore, I see the intra-company both investing and financing. That's the other side of this, right? in addition to the intra-company operations, there's intra-company financing with Invidia often financing customers who buy from them, and intra-company investing,

something that when I first started valuing companies, I used to be happy to be able to value U.S. companies because cross-holdings were not common here. I hated value in family group companies in Asia because to value one company had to value four companies. I worry about the fact that to value any of these companies in the near future, I'll have to value three companies. To value Microsoft, I'll have to value Open AI because of the big position it has.

So the intra-company investing, financing, and operations creates noise in the process. The noise is just a fancy word for, we don't know what's going on under the surface because of the intra-company stuff. On this point, when we look at the net income for Amazon and Google, both of which exploded, almost tripled in Amazon's case. 85% of its net income, Amazon's, was attributable to its paper gains, its unrealized gains in their stake in Anthropic

and Open AI. And for Google, looking at their stakes in SpaceX and Anthropic, that number was 87%. So their earnings have been massively inflated by these gains in these AI startups, which they haven't even actually realized. So we didn't even know if there's value are fair or even make sense. We don't know how they were valued either. And this is something I

pointed out on Twitter recently. As a result, their PE multiples have been massively skewed. They've come way, way down, somewhere below 20 times earnings, because their earnings have exploded. And I worry that that means that the metric has been compromised by AI because it looks cheap, but you're not getting the full story. We made like our own little adjustment where we kind of stripped out the AI startup gains and you pointed out that actually we need to be including their value in the numerator. Could you talk a little bit about how we can actually value these companies now that they're so in bed with the AI Frontier Labs? When I talk about pricing metrics in class, P, EV to EBTA, EV to sales, I mean, the whole,

there are dozens of multiples. One of the first rules I start with is a consistency rule, which is what's in your numerator should be what's in your denominator. So let me give you an example, you know, enterprise value to EBITDA, widely used from capital-intensive businesses. What's in the numerator is the market value of equity in that net of cash. And people say, why do we net cash out? Because the income from cash is not part of EBTA. And then I point to a danger with EBTA,

when you have cross-holdings. When you have cross-holdings, here's the problem. If you have minority holdings and other companies, your market value reflects those holdings. Because the market knows you own 30% of this very valuable company. So what happens is your market cap inflates, your enterprise value is higher, but your EBITDA does not include the earnings from that crossholding because it's viewed as a non-operating. It shows up below the EBITDA line. So I say even with enterprise value to EBITDA, you should be netting. out the value of the crossholdings because otherwise you inflate the, you think that these

companies are more expensive than they are because the EV is inflated by including them, but the EBITDA does include it. And people say, that's a pain in the neck. And they say, welcome to reality. A reason people like to use multiples, it's a shortcut, right? You're not asking the in-depth questions you need to ask to value companies. I need a shortcut. And that shortcut but historically has been the P.E. ratio. But for 20 years, I've argued the P.E. ratio is the most dangerous of all multiples to use because it's this mess, right? There's a numerator that's just the market value of equity, and a denominator includes everything. It includes interest income from cash. What if you're making a huge interest income from cash, and you're including it,

and that's part of your denominator. You're mixing up a business with a cash holding, and In this case, a crossholding, and you're trying to come up with a consolidated multiple. So the lesson I think that you get by looking at the Google in the alphabet is don't trust net income. Net income is a deadly number at these companies because of the mess that goes into it. Climb the income statement. Look at the operating margins. Look at the interest expense. Because in a sense, you worry about things like financial expenses and other income, but separate the two.

The operating income is going to tell you what the operating business of these companies are doing. The rest of the stuff is telling you what the other investments are. Now, I'm surprised to see the marking up. I mean, because on the Amazon income statement, it says gain from sale of asset as opposed to the marking up component. The reason I'm surprised is when you hold an investment for trading, you have to show the marking up or marking down. SoftBank is a classic example. When you hold an investment as a strategic investment, is an investment that's going to be part of your business in the long term,

you generally don't do that. You hold that original value, and any income you show will be the actual income or loss from that holding. That's what happened with Microsoft. So I am confused about what the accounting is, and if the accounting is they're holding it these companies as trading investments, that's a very revealing statement. Because you told me this investment in Open AI

was because you wanted to build the AI business in the long term, not because you wanted to make money like a venture capital, buying Open AI at a low price and selling at a high price. So again, it goes back to this. Once you create cross-holdings, you create these follow-up questions.

What's the motive? Why are you doing it? And I think that's what I mean about transparency. Tell me what the end game here is. What is it that you're hoping to get from an anthropic investment or an open AI investment and frame it in terms of that end business, that this will give you an advantage

on that AI product and service business. I'm willing to listen, but don't tell me you're doing it because you want to make money because you're not a trader. You shouldn't be doing this if your objective is to just make money on another player in the AI space. But if all of Wall Street were as rigorous as you, then maybe we'd be okay.

But I think the trouble is that people like shortcuts. people don't want to do all of the homework that you're describing. So they just look at the numbers. They go, oh, it's okay. The revenue exploded. The net income is up. I don't think it's rigor, particularly that keeps it apart. It's the fact that you're required as an analyst to react in real time.

I'm glad that I'm not there at 4.30 after an earnings call where somebody says, what do you think about that earnings call? I mean, I need a little time to digest what's in that statement. I need to look at the footnotes. I need to see the breakdown. But I think we live in a world where instantaneous reaction is required of us. It's part of your job.

So I cut them some slack on what they're doing, but I faith that eventually markets kind of figure it out. And I think that that's going to be the end game is, now, I did strip out the Google and the Amazon earnings from the effect on earnings. And they're pretty good quote. is even without it, and I think it made more sense for them to say, this is what we did without R.A. And that's what I meant about separating what's happening with AI from everything else,

is I don't think markets would punish them if they did that. I think markets would actually reward them for transparency. I mean, it's one of the reasons I think Jeff Bezos was cut so much slack by the market for so long at Amazon. Because he was open about what he was doing. He said, look, I'm building, I mean, I call it the field of dream story, which is if we build it, they will come. He was open about the fact that Amazon, in its early years, was building revenues, that margins would look terrible, that they'd give away shipping for free because they had an end game.

He brought people into the end game. And it's that belief that allowed Amazon to do what it did. So that's why I said if there's one company in this mix that should have experienced, having gone through this before it's Amazon. So I'm going to get my cues on whether Amazon starts to become transparent earlier than the rest because I would expect them to. So there has never been this level of CAPEX in emerging technology that didn't ultimately result in a fairly serious correction, if not a crash. Whether it's the railroads, the electric grid, you know, the highways, whatever with the steel in the ground, infrastructure in 99, whatever, you know, there's a. always a correction. And also, to be fair, the technology and many of the companies survive

that correction and go on to be big winners. But the thing that gave me the sense that we might be closer to that correction moment than further is when Meta or when Zuckerberg announced they're trying to sell their compute. And the same with Musk. And my sense is that, I mean, If you look at the beginning of the year, the narrative was around compute scarcity. And now both Musk and Zuckerberg are trying to spin it as look at the premium we're getting for the infrastructure we've built. And what I see in that is that the AI demand curve has been vastly overestimated. And now you have essentially hundreds of billions, if not trillions of dollars, in CAPEX, all going to only two sources of demand creation, open AI and anthropic. it feels like we have all of a sudden pivoted from a supply crisis to potentially a demand crisis.

Your thoughts? When I was looking at SpaceX prospectors, and I noticed that they were making more money by leasing out their data centers to others, in this case, Anthropic. Then they were making, you know, I thought it was actually at, you know, seriously at odds of the AI story they were telling in the prospectors of this huge market. 28 trillion. And I think in the aggregate, what you're pointing to is the fact that you're making more money by selling into the air architecture space, and I include LLMs in this, then from talking about air product and services is very revealing. It tells me that you're not as confident as you claim to be that there's going to be this huge. If you were really confident that there was going to be a huge market, you wouldn't want to lease the space out to who could

be potential competitors in that market. The fact that you're doing it, I think, is a sign that at least for the near term, you don't have as much faith as you claim to have, that AI is going to be as big as it is going to be. I take that as one data point, and then I take the fact that the amount of tokens being consumed from Chinese LLMs or AI infrastructure companies has, the data I've said, it's gone from 8% share in January of 25 to somewhere above 50% now. it feels as if the cracks are really beginning to emerge in the whole narrative here. And that China is potentially engaging in AI dumping, trying to do to our market,

what they tried to do to our steel market 30 or 40 years ago. And doing in kind of 20 weeks to Silicon Valley, what Japan did to Detroit over 20 years. Am I overstating the threat of these inexpensive LLMs out of China? No, I think China is just ahead of the game in seeing that the AI product and service market, this mythical market we keep talking about, is going to bifurcate. There's going to be a premium component of the market, primarily business products and services, which is high margin. And there's going to be a big component of the market, which is going to be a low margin, big scale market. And China's clearly putting its stakes and saying, that market, we're going to go after because we're equipped to go after it. So I think China in many ways is taking a look at that end game playing out.

And I think the real question is in that end game. So let's play it out. Let's suppose the end game, the AI product and service market turns out to be looked at. Let's play along with AI optimists. Let's say it's $6 billion, $8 billion, $9 billion, even $10 billion. That by itself doesn't create valuable companies because you haven't told me much about the business models that will be used to generate money in that market. If 90% of that market is low margin, large scale, you could be a $10 billion market.

But the companies in that market are not going to make much money on those trillions of dollars because your margins are going to be single-digit margins. Now, one of the most revealing components of the Anthropic success story was how much it costs Anthropic to deliver the products and services that they charge $6,000 an hour for. This isn't software. where the unit economics are amazing. The unity economics are struggling in a business that's still evolving. I think the question, though, is what's the catalyst that's going to create a major correction?

This might be one of those things where you get multiple catalysts and corrections along the way. Rather than a big time, you know, like the dot-com bus, unlike the dot-com bus, something that happens, staggered pain. I'm not sure which is worse to get the pain at one go and clean up and move on, or have staggered pain where individual companies get into trouble and markets go through these cycles of correction and hope. Where you come back a little bit, then you have a correction again, but I don't see an individual catalyst that's big enough to, for a moment where people are, oh my God, the air market is not going to be as lucrative as we thought it was. Because it seems like every time you get something that has the potential to do it,

there's still money on the sidelines that jumps in and says, we need to be in AI because everybody else needs to be in AI. Now, I think that last week after the blow-up situation, when you had all that selling of force selling, the next day you wake up and there's a 10% jump in all of the stocks, clearly money coming in saying, we've never been in AI, we need to be there, because everybody else is there.

I think I know whether this is being accentuated by social media and the awareness of other people make money. I don't know. But that factor still seems to be strong enough to overcome the catalyst effect. But at some point in time, the catalyst effects are going to overwhelm that momentum effect. We'll be right back. And for even more markets content, sign up for our newsletter at profgmarkets.com. presidential races are always messy, but 2028 is shaping up to be our messiest presidential campaign yet.

There are a ton of names that are rumored to be thinking about throwing their hat in the ring. All righty, here we got Trump. We got Kamala Harris. We have Marjorie Taylor Green. So this week on America, actually, I wanted to hold a little fantasy draft. Of course, we got to have AOC on the list and stud George Clooney. And to do that, I wanted to invite. two of the messiest people I know. Box journalist and podcast host Kara Swisher. Let me tell you.

Here's the charm of a cyber truck. An independent journalist and political commentator, Don Lemon. This could go really fast because I got boom, boom, boom, boom, and I'm sure Kara's the same way. I'm a Stead Herndon, and this is America Actually. Catch us every Saturday on YouTube or wherever you get your podcast. We're back with Prof G Markets. Do you think that opener, you mentioned the unit economics,

Charging all of this money, but then they spend way more money delivering that product. Do you think that Open AI and Anthropic ever figure out the unit economics? Like, I'm just going along with, I guess they will, but I don't know why I should believe that. The way AI is set up right now, I don't think there's going to be that much give on the unit economics in the near term, right? You still have to build. I mean, the physical infrastructure seems so expensive. Unless you can figure out a way. to build data centers at one-tenth of cost, right?

This is not the kind of investment where scaling up is going to help you that much. You have to figure out how, you know, so none of the elements are easily going to bend to economies of scale argument. Maybe I don't know enough about AI. Maybe they have some secret source that's going to allow them to do this.

But I don't see it. I think every time I open Google and AI helps me out there, it seems like a marginal cost is being created somewhere along the way that I'm not seeing. And that marginal cost is not decreasing because tens of millions of people are using Google at the same time because each surge seems to provide an additional cost. So, as I said, I am not enough of an expert in AI to see where these economies of scale will come from. But I don't see them yet.

I don't see anything in the architecture that bends easily to scale where you, you can say the costs are going to decrease just because they get bigger. I don't see it either, and I feel as though we're just being told to trust that it's going to work out and trust that the unit economics will make sense eventually. But if they don't, the whole thing collapses. On the premium side, it starts to shrink, maybe not collapse, but shrink. McKinsey might be willing to pay $9,000 or $90,000 and $500,000 for an AI agent. There would be premium products where you can charge a premium price even over the high cost,

but it'll mean that that end market is going to be more premium product. The economy is on scale argument might work better in that lower margin market where you don't need Nvidia chips. You don't need expensive data centers. And, you know, the reason I say that is I see a lot of stuff that I saw 10 years ago that was labeled as machine learning or, you know, look at this macro we built for Excel. I see the same stuff today marketed as AI. And my question is, why do you need any of this AI?

You don't need data centers. You don't need Nvidia. You could do this with the traditional computing power and without the access to data. That's the part of AI that I think is going to be the scalable part. We have low margins, but the scale takes care of it. So it's not that the AI product and service market will not exist, but it'll become almost primarily a mass market, low-margin business. And that's why the architecture you're building might not be right for that.

It might be too much for the premium part of the business where you need these high-end chips and data centers to feed those products. We're describing a lot of AI anxiety, and as you can tell, I'm pretty anxious about this myself. How much of this anxiety, in your view, is priced in right now? I get the sense that it's somewhat priced in, certainly priced in in the case of meta, in my view, perhaps not the others. What do you make of these valuations?

Do you think that the markets are telling us that they see what we're saying? They seem to see it, and then they seem to forget it, right? You take the meta example, right? Over the last week, it's made up about 70% of what it lost in the previous six months. So it's not like the lessons are sticking because, you know, all you seem to need is some other distraction along the way. And that's why for the moment at least, the catalysts are not working because they have a temporary effect. One company gets hurt, but you're not seeing it ripple into the other companies. It seems to be company

specific and time specific. So I think that collectively, I think, anxiety kind of ebbs and flows, but it doesn't seem to be at a level where it permeates into permanent pricing changes yet. And I think maybe that reflects the fact that those who are anxious are in the wrong. So I'm always open to the possibility that maybe we're missing a really big story. Maybe Leo was right about this being the ultimate winner. It's going to happen in 2027. And I'm always going to leave that door open because I've learned that you can't ignore somebody just because you disagree with them. I have to be willing to listen. So I try my best

to read as much as I can from the AI optimist. I'm looking for a story that explains economies of scale. But so far, the story still seems to be focused on fuzzy end games and how great an AI agent has worked in an individual business. I'm not seeing the aggregate numbers from any of these stories that lead me to say, okay, there's something here that I should be taking a closer a look at. So it's going to be a race between whether those stories come into play or whether the skepticism builds up to a point where people stop believing. So are you comfortable with current valuations of big tech, Microsoft, Amazon, Google, meta, will leave Apple aside because if they're not getting into AI. Are you comfortable with them, these valuations, or are they too

volatile to even have an opinion? I can live with them. And it's because of my framing. I framed them based on what I paid for them. I paid for them a long time. I know. I know. know this sounds irrational, but if it keep comparing the price to something, you know, that I could have got it for six months ago, a year ago. So I'm willing to accept 20% right down in those investments and accept it as part of the long-term costs and benefits. Now, the reason I do it is, if I think there's a shakeout in the AI space, these companies will be impacted, but the lesser companies in the space are, I mean, none of these companies have net debt ratios that are beyond the single digits. Their debt is completely manageable. And I did net to debt to EBITDA for all of

these companies, one and a half times EBITDA. So basically next year, they stopped investing in AI. The year and a half of EBITDA from their regular businesses would pay off the debt. So debt is not my concern of these companies. But there are a whole host of lesser companies where debt is a much bigger component. There, an AI meltdown in the story will be catastrophic. Now, just as Amazon got hurt after the dot-com bus, but it's the best thing that happened to them, because it wiped out all of their dot-com competition, in many ways the Mag 7 might be, you know, this is a very sinister view of this whole thing. Maybe they're wishing for an AI correction.

They'll be punished, but the rest of their competition will be decimated. Now, and remember, Open AI and Anthropic can't survive on their cash flows either. So maybe there's a, you know, if you're a... If you were looking for a story with villains that knew more than they did, maybe this is the end game for them. Is they hoping for an AI shakeout where they come in and pick up the pieces at bargain basement prices? Because they're going to survive.

I mean, I don't have worries about failure rates with these companies. I just worry about their prudence and what they're doing in terms of AI investing. There's been a lot of concern around they have $1.4 billion in, disclosed debt. But the debt you don't see is now bigger than the debt you can see. So, for example, META is carrying $420 billion of off book versus $140 billion on book. But I don't entirely know if that's a feature, not a bug. Are these companies taking advantage of their credibility in the marketplace to offload from their shareholders some risk? Or as it's just an accounting trick to create opacity around the actual amount of debt they have.

Now, I'd be interested to see what the recourse on the debt is. If you're lending money on a data center that meta is a player in, and the debt has recourse only against debt data centers, revenues, and assets, then you're not going, I mean, in a sense, as a meta shareholder, it's clearly not something I want to see happen, but it's not something that's going to impact me. No, I would be, you know, again, I want to go back and look at the accounting rules as to what happens when you take debt that's off balance sheet debt, where there is recourse against the parent company and whether you can get away not revealing that as part of your debt.

And maybe there's this iceberg of debt that you're not seeing the underneath that could be, you know, but I think that even, I mean, let's bring the risk. recourse dead in then. And let's see. I mean, I'd like to see full disclosure of that debt that's not on the battle sheet. Because even then, I would wager it's three, three times. We're not talking about heavily levered companies in the sense of companies that are, you know, that are going to be dragged down by the failure of these, of AI. But I might be wrong on that. Maybe I need to do more of my homework digging through that debt. But that would require some bending perhaps even breaking of accounting rules to be able to get away with it.

So, you know, I'll do another read of the footnotes to see if there's something in there. But, no, at least for me, the worry with the Mag 7 is not so much the debt, but the investment paying off, whether there's enough for return. The worry with lesser AI is whether they can make it to the other side. And their worst-case scenario is that AI turns out to be an incredibly big market, but they don't make it there. They fail because the debt comes due and they have to sell themselves

to one of these other companies at a fraction of what they should be charging. And that's a very real possibility in this space. And if it does happen, there's going to be serious side costs for the rest of us because debt going down is always going to create side costs. Have you looked at Open AI and Anthropics

what very little we know of their financials and would you ever value those companies I will because I valued SpaceX with, you know, let's face it, 80% of SpaceX value, at least the numbers, the story was about XAI. So it's an XAI story embedded in a space launch company. So when the prospectus comes out, I plan to value them ahead of the IPA. And it'll be interesting to see as you move from SpaceX, whether the sum of the shine has come off the story because of what's happened. It's SpaceX, where people have them, you know, maybe the anthropic IPO is not going to be a trillion. Maybe it'll be 800 billion. Maybe Open AI is not going to get the price it wanted. So I have a feeling that they're revisiting the story because what they did with XAI was too sloppy. It really didn't stick.

And they need to get there, you know, and I think there you're going to look for more specifics because, you know, just telling me you're a great LLM, you have amazing agents. AI agents doesn't do it for me. You need to show evidence that this is actually sticking at a business level, that you're making money. There I need to see the unity economics and evidence that there are economies of scale you can point to. This is what it costs us two years ago, a year ago this year. Because that's how you back up in economies of scale story. Just don't give me the words.

Show me the numbers. Would you ever value Open AI without the full story? Would you ever undertake that? Because it seems as though, and you pointed this out in our exchange on Twitter, like, we need to put the value in the numerator. But that kind of means valuing open AI. In order to value big tech correctly, I need to value open AI, but how can I value open air because I know nothing about it?

Would you ever try to do it anyway? Absolutely. I mean, in a sense, you have no choice but to tell the full story. The question is, is it a story you're entirely making up based on clues, you're getting as well, which is a very danious, which is what I did with X-A-I. There was no story in the prospectus. There were just numbers thrown out of thin air. There was no story from the management of what they planned to do.

So I had to write the whole story. That's always going to create more uncertainty. So it's not a question of whether you can tell a full story, but what's the basis for the story you're telling? And the case of open AI, unless they fill in the blanks, I am. coming up with the story based on my limited understanding of AI. So I'll give you an economy of scale story that's not very strong. And I might say your costs are not, you might disagree with it.

But part of the reason I think it's critical that investors flesh out their full story is then companies are forced to respond, right? So if Open AI feels that they have true economies of scale and the story being pushed into the valuation is there are no economies of scale. get the costs are not going to go down, then show me the data that you have that tells me that I'm wrong. Because as long as we let these companies get away with these diffuse, total addressable markets, trust me, and you don't even tell a story. Use a pricing metric and say, it's okay because there's a big market out there. There's no incentive on the part of companies to tell the

fuller story. So I think I will try to tell a full story, but I'll be open about the fact that that much of the story is my story based on little strands that I've pulled out of different places, some from the company, some from people who know AI a lot better than me. But I'll always tell a full story and people will take issue with me saying, your story is wrong. And I say, absolutely, I know it's wrong. But what's the counter? Where is the counter narrative? You can tell me a narrative is wrong, but you have to beat it with a counter narrative. And that's good because it extracts the counter narratives from not just the company, but from other investors who disagree with me. But it seems as though there is almost no price discovery in the Frontier Lab world at all,

except for, and this is how we'll end, SpaceX, where as soon as it went public, it went up, and then it came crashing way down, cut in half, within a couple of months, maybe less than a couple of months. Let's just get your reactions to SpaceX. What happened there? And I think in many ways it shows you the power of hype, the power of momentum, and how social media is added to that momentum factor, which is, I mean, this is the ultimate social media experiment because you think of Grog and X, and basically you've got this. And social media is like riding the tiger. It's great when you're on the back of the tiger, but sooner or later you're going to slip up

and end up being its meal. So I think it shows you both the upside and the downside of social media driving prices. But I think as more numbers come out of the company, as the earnings, much as we take issue with the earnings we're seeing from the max seven, we extract information from it that is useful in kind of finessing our story.

So once you go public, that is something that these companies will face is now in addition to telling the story, you've got numbers that come out, and either your story is consistent with your own numbers, or people are going to look at the inconsistency. So, no, I think that there are more tests coming, and I think for SpaceX and these other companies as they go public, it'll be an experiment worth watching. It'll show up in the market press. I'm fascinated with 1999, because I think it was the year Ed was born.

That's true. But also, we remember that, Aswath. And I understand, you know, things are different this time in quotes. But it just feels eerily reminiscent. First, the B to C guys took a hit. It feels like OpenAI is under real fire around its business model right now. And then everyone said, okay, no problem. We'll go to B2B.

Everyone piled into Anthropic, which had greater share in the enterprise market. then when that didn't live up to its expectations, they started going after the infrastructure guys. It feels like we are midway through the exact same cycle with the domino's beginning to fall. This feels more similar than not to me to the 99 and then ultimately the 2000 implosion. Where do you see the parallels or not from 99?

No two corrections ever work out the same way. So, I mean, part of me doesn't want to see a correction because of what it'll do to every portfolio in the U.S. But part of me just, you know, from a market observer standpoint, I am interested and I'm much more aware now than I was in 2001 of the catalysts that created. Because in 2001, if you remember, it wasn't a single catalyst. It was just a collection of small things happening.

Now, as you said, you know, not even domino's falling, but trees falling in the forest. Eventually, you wake up, oh, my God, half the forest is gone. So, you know, I'm keeping tabs as I can, you know, whether it's on the investor's side, we see a meltdown of a hedge fund shutting down because it made too much of a better AI. Or on the company side,

where you see companies stepping back from the brink and saying, we screwed up, we're going to write off that. Now, we still haven't seen a major AI CAPEX right off from a company yet, because that'd be the ultimate admission of, hey, guys, we screwed up, we've admitted, we screwed up, and we're not going to do this anymore, right? Whether that'll happen after the correction or before the correction or whether it'll trigger the correction. So I'd keep my eyes on accounting revisions and, you know, restructuring

charges and look at what's being written off, because that might start to give us as an indication of how this correction will play out. Aswath Demodran is the Kirshner family chair in finance education and professor of finance at NYU's Stern School of Business where he teaches corporate finance and valuation. You can read his research on his blog, Musings on Markets. Aswath, thank you so much for joining us today. And to our live audience, thank you for tuning in.

We will see you next time. Thanks, Aswa. Thank you. This episode was produced by Claire Miller and Almond. Alison Weiss and engineered by Benjamin Spencer. Our video editor is Jorge Carty. Our research team is Dan Chalon, Kristen O'Donoghue, and Mia Silverio. Jake McPherson is our social producer.

Drew Burroughs is our technical director, and Catherine Dillon is our executive producer. Thank you for listening to ProfG Markets from ProftG Media. If you liked what you heard, give us a follow and join us for a fresh take on markets on Monday.