5 Stocks Are Carrying The Market — Here's How To Protect Yourself

Prof G Markets

Scott Galloway and Ed Elson dissect the financial realities behind Anthropic's upcoming IPO, highlighting extreme customer concentration and valuation risks that defy historical precedents. They then

Key takeaways

• Paul Kudrowski noted Anthropic generates roughly 25% of its revenue from just two clients, a concentration level that raises systemic fragility concerns.
• Ed Elson revealed that 80% of both OpenAI and Anthropic's enterprise revenue comes from only 1% of their customer base, amplifying dependency risks.

Main topics

• AI lab valuations and customer concentration risk
• Anthropic and OpenAI financials

Notable quotes

Market dynamics trump individual performance.

Conclusion

The episode equips listeners with a clear framework for navigating current market

Transcript preview

Speaker 2 (0:00) Today's number, $30,000. That's how much the world's most elite sheepdogs sell for at auction on a farm in England. True story, Ed. My sister was accused of bestiality. We'll all be a monkey's uncle. Speaker 2 (0:16) You like that one. Speaker 2 (0:28) Ed trying to look sexy. The flannel shirt. He thinks he's Grizzly Adams at a Pearl Jam concert. What are you wearing today? The same thing I wear every day. I'm rag and bone jeans, Panerai watch, and Brunello Cucinelli. Hello, Speaker 3 (0:43) ladies. There we go. Hello, ladies. That's pretty sexy too. Yeah. $30,000. Speaker 3 (0:51) Grr. Grr. Speaker 2 (0:54) Yeah. Yeah. No, I make a room sexier by leaving it. Speaker 3 (1:00) So $30,000 for a sheep dog. Speaker 2 (1:04) What do we think? Worth it? I asked a Scotsman what his body count was and he started counting and then he fell asleep. That is my favorite family guy thing where the guy is shearing the sheep and the sheep's like, oh, yes, you bitch. Shear that, shear that shit. Shear it. It is very good. You spoke to one of my like, early role models, Paul Kudrowski, yesterday. By the way, I listen to podcasts, which I don't like to do, but he's really good. It was a really good one. Well done. Paul's really good. Paul's a very, very smart guy with a lot Speaker 3 (1:42) of opinions on AI, which we will be getting into now. But first off, we haven't talked enough about me. Speaker 2 (1:47) About 25 years ago, I'm not joking, my first TV appearances were on Bloomberg. Someone named Laura... Gosh, such a cool woman who did content programming, reached out of nowhere and said, you're a marketing professor. Do you want to come talk about marketing? And I used to go to Bloomberg Television every week, and I'd be on with Tom Keene. That's how I met Stephanie Ruhl. That's how I met Sarah Eisen. And I used to do every week this thing called Business and Tech with Deirdre Byrne, who I love. She went to Fox. It's like college. Stephanie Ruhl went to MS. Now! And then, oh, and Eric Schatzker, he's still there. He's the only one that stayed. But it all started for me at Bloomberg Television. Speaker 3 (2:28) And Speaker 2 (2:29) what is this? How does this relate to our man, Paul? Paul was also a frequent guest. And I used to be both happy and bummed when I'd have to go on. They'd bring us as the two panelists in for like, you know, he and I were the panelists and the founder of Reddit for one. Speaker 2 (2:50) Facebook went public. It was supposed to go public at the opening of trading. It didn't go public to once. We had to sit there for three and a half hours trying to talk about tech. It was really painful and come up with new insight into the Facebook IPO because there were some technical problems. And Paul, it was like a love-hate thing. I love Paul. He's a nice man. I think he's super fucking smart. I was always so intimidated by him because I'd be like... The audience is going to figure out, I don't know what I'm talking about when I hear Paul talk. He's very smart. He's very knowledgeable. He's Speaker 3 (3:18) very smart. He brings the heat. He's aggressive. He's opinionated. He's everything you want in an investor, but also in a guest on a show. Speaker 2 (3:28) Yeah, he's really good. Speaker 3 (3:29) He told when we had him on a while ago, we've had him on a couple of times now. He said he doesn't like going on Bloomberg or CNBC anymore because he doesn't like feeling like he's in a Piers Morgan debate show anymore. It Speaker 2 (3:39) could be worse. You could go on Piers Morgan and have to put up with some some maggot douchebag who calls you un-American and desperate. That's nice. That's why I've worked this hard, Ed. Speaker 2 (3:52) God. I like Pierce, though. I like Pierce. Pierce Speaker 3 (3:55) is good. He knows what he's doing, which is his Speaker 2 (3:58) sensationalizing Speaker 3 (4:00) drama. But it's entertaining. It Speaker 2 (4:03) makes I Feel Stupider with Abby Phillips feel like real journalism. Speaker 2 (4:09) That's even more, yeah. But Pierce, he knows how to, he does good TV. It's like- When Speaker 3 (4:14) did CNN and Abby Phillips turn into like Jubilee Media? When did that happen? Speaker 2 (4:19) I think they nailed their eyelids to their forehead and made them watch 72 Hours Straight of Jerry Springer. Speaker 3 (4:25) Yeah. Speaker 2 (4:25) And they said, that's what we need. We need to turn this into Jerry Springer with a CNN logo. That's probably a little harsh. That's probably a little harsh. By the way, Abby texts me to get in my face. She texts me. She is not afraid to text me and tell me how wrong I am. People like her, though. They say she's a good journalist. That's my attempt to put my words back in my mouth here. Speaker 3 (4:51) It's also the highest rated show Speaker 2 (4:52) on CNN, which is a pretty wonderful reflection on the state of our society right now. Speaker 3 (4:57) Yeah. Keep trying to unsay what you said. I will quote Katie Martin. You can't put the shit back in the donkey. Let's get into our show. We have a lot to get into here. All right. Speaker 2 (5:10) is the time to cry. I hope you have plenty Speaker 1 (5:15) of the world at all. Speaker 3 (5:16) Washington is starting to take a closer look at the AI industry. Last week, the FTC opened an investigation into OpenAI and Anthropic over potential safety and consumer protection concerns. The news came during an already chaotic week for both companies. OpenAI announced that it would not release its latest Astra model due to safety concerns. And Reuters reported details from Anthropix S1, which showed that the company racked up an operating loss last year of $8 billion. So, Scott, a lot of AI news to dig into here, which is going to be extremely important for the public markets going forward. I think a decent place to start would probably be the Anthropic S1, which, as you mentioned, we dug into last week with Paul Kudrowski, and he had some interesting things to say. But I would like to get your take. Revenue last year grew 12x to nearly $4.6 billion, but their operating losses widened to more than $8 billion last year. To be clear, we don't know. what their financials look like right now in 2026. And that obviously matters a lot. But we did get some clarity on what financials looked like last year. And the reality is, as we have suspected for a long time, that the losses are enormous in the same way that they're enormous at OpenAI. Not as enormous when it comes to Anthropic, but still very, very large. And this is very relevant considering this company is imminently going public. Supposedly, it's going to happen right after the midterms. Scott, what did you make of these leaked, revealed financials? Who knows if they were actually leaked or if it was a setup? I don't know. What do you make of them? Speaker 2 (6:59) They were leaked by Anthropic, would be my guess. There's the valuation itself, and we'll come back to it. But what Paul said yesterday on, I was going to say your show, but our show, it was that the really... The number that jumped out to him, and I didn't recognize it, was I believe about 40 % of the revenue was coming from just two clients. A quarter, not 40%. Excuse me, a quarter. We always used to worry. I've always run small companies. I've never been able to scale to a big company. And we always had, and I've always been in B2B services, right? Strategy, analytics. And we always had this problem. of too much concentration across the small. Nike was like 22 % of our business at one point. P &G was 15%. LVMH was like, Estee Lauder. The five companies were like 60 % of our revenue. And I used to say, okay. That's a problem, but it's a good problem. It means these iconic brands are spending a shit ton of money on a small company and getting the types of contracts that only big companies get. But when you're asking for a $2 trillion valuation, you need to look more like meta. And that is, remember all of these, and I'm guilty of this, attempts to boycott meta, where people were so upset at meta. I think it was like five years ago, they said, let's start a boycott. And Meta has something like 300,000 clients and no one is more than a third of a percent. That is what you call robust. And robust is a really interesting term in business because it indicates fragility. Let's talk about the economy. The banking industry is not robust. If JP Morgan finds out it had some rogue whale trader that figures out a way to short circuit or get around some of their internal controls and makes an enormous bet on the debt market in Germany and it turns against him and they end up down $700 billion somehow using leverage and derivatives that got past their compliance or risk department. and Jamie's about to get flipped upside down, if he calls Donald Trump, Donald Trump has to bail him out. Speaker 2 (9:19) JPMorgan has become too big to fail. And they put in place a bunch of capital constraints and ratios, Sarbanes-Oxley, that are supposed to limit the likelihood of that. But the banking industry in America is not a robust business. There's a lot of small banks, but a disproportionate amount is concentrated amongst some of the bigger banks. They would argue back that that creates strength, creates operating leverage, better customer service, lower interest rates. Those banks do an amazing job. And JP Morgan right now is worth more than the biggest 12 banks in Europe, just that one bank. So there's this capitalism versus regulation. But back to the notion of robust versus fragility. The fast food industry is really robust. It's not fragile at all. If the biggest player, McDonald's, went out of business, you could just still find really cheap caloric intake that puts you on a path towards diabetes really early, really easily. There's a ton. of substitutes. And the reason why you want a robust industry is it not only results in a lack of systemic rest to the system if one goes out and consumers start bailing it out. The problem when you're too big to fail is you start taking stupid risks because your downside is limited because you know you'll get bailed out, right? And also the advantage of a really robust industry is there's more competition, which ultimately lowers rents and transfers capital back from the shareholders to the consumer, which is absolutely what we need. more of long-winded way of saying the most frightening thing about this is i don't think you've ever seen a company ask for 500 billion in market capitalization that had a fraction of the client and revenue concentration that anthropic highlights or articulates in its s1 uh i want to get your reaction to that and then i want to talk a little bit about the valuation oh Speaker 3 (11:10) i kind of 100 agree with you i think the problem here isn't necessarily the company per se. It's a startup. It's a relatively new company that is trying to figure out how to get clients, similar to the dynamic you're describing that you dealt with when you were starting your businesses. Some will say, well, that was last year. You're looking at the customer concentration from last year. But what we know based on the ramp spending data, which has been extremely informative, is that 80 % of both OpenAI and Anthropix enterprise revenue is coming from 1 % of customers. And so the question that you have to ask yourself, sure, that they're getting a lot of revenue. And let's be clear, the revenue numbers are incredible. I mean, supposedly OpenAI is... coming up on $70 billion ARR. ARR is a little bit of a bullshit number, but let's put that aside and recognize that these are tens of billions of dollars in revenue that these companies are raking in, and that is impressive. But there is extreme customer concentration risk here. And I've been going back and watching Mad Men, which is the greatest TV show of all time, and I am reminded of exactly this dynamic, because I don't know if you remember, but... uh, Sterling Cooper Draper Price in that TV show, they have a similar problem where they're extremely dependent on Lucky Strike. And then one day, uh, the CEO of Lucky Strike has a meeting with Roger Sterling. He says, sorry about this. I know we've been together for 25 years, but it's time to move on. At which point Roger Sterling says, you're going to kill me. And that's essentially what happens. It completely collapses the business. And this is what happens in business. If you become too reliant on a handful of customers, things can go wrong, and suddenly your business is no longer sustainable or structurally sound. So they are at immense risk of that. And that's okay. That happens. But they are asking for a $2 trillion valuation, which would make them more valuable than Meta, more valuable than Saudi Aramco, more valuable than Broadcom, more valuable than Berkshire Hathaway, more valuable than Eli Lilly. more valuable than JP Morgan, more than twice as valuable as JP Morgan. And so for me, the problem here, as usual, is that we're having price problems. We're having an issue of the valuation. And I don't think that the valuation is... properly pricing in the massive risks here. And I've said this before, but I'll say it again. I don't think that we necessarily have an AI bubble right now that is infecting the totality of markets. We can see that in, say, the NVIDIA valuation, which has actually come down quite significantly. But we do have an AI lab bubble where there seems to be too much. excitement, and too much enthusiasm about Anthropic and about OpenAI because they're new, they're sexy, they're doing interesting things. But let's be very clear, they still haven't figured out their business models by all accounts. And from what I'm seeing, they are still not profitable. In fact, they're very unprofitable, some of the most unprofitable businesses of all time, and the risks are still systemic. So I think that needs to be priced in. But the question is, will investors care about it? Or are these businesses just too sexy? then it's not going to matter. I'm not sure, but over the long run, I think it will matter. I'm Speaker 2 (14:25) just impressed I was able to find a protege that was more glass half-assed. You're like glass is broken and it's going to sever your thumb. Speaker 3 (14:36) I'm not. I'm not. On these two companies, again, for me, it's all about price. Just remember about me. I also bought Meta. I also bought Microsoft this year. I bought the SaaS company. So I'm not everything's, I'm not a perma bear. Speaker 2 (14:52) I've never seen a more important distinction or what will be a greater effort in IR history for people, for everyone to start talking about four or 12 months revenues versus trailing. Because the thing that is staggering about this company is its growth rate. And if it were to, it's supposed to be 10Xing again this year. And if it 10Xs next year, then all of a sudden its multiple on revenues goes from 435 to 7. At 435, you know, DeMotoran accurately called it, said that the math says Anthropic needs the entire AI market to nearly 4X just so one company can hit its number, which isn't, you know, which isn't a forecast. It's a monopoly fantasy wearing a spreadsheet. you know, needing 74 % annual growth for 10 years straight to justify your price tag, you know, you're asking investors to bet on a streak no company, including NVIDIA, has ever thrown. So that this valuation, built into that valuation, is this will be the most enduring, fastest growth company in history. Speaker 3 (15:53) That's a bold thing to assume. Speaker 2 (15:56) That's not pricing to perfection. That's pricing to hallucination, right? That's pricing to kind of a wing and a prayer. I think it is probably growing faster than any $1 billion plus business in history, including NVIDIA. Now, having said that, the asterisk on the asterisk is the thing that is fueling that growth appears to be a small number of enormous spenders that keep finding, I think it's specifically Alphabet and Amazon, that keep finding new ways to leverage AI. And as Paul referenced, how many of those types of companies have that type of business? to one, justify multibillion-dollar spends, and two, can leverage that multibillion-dollar spend into their businesses. How many businesses have the capacity and the need to spend one, two, five billion dollars on these types of services? So this is, either way, I would imagine, I think a lot of these marks, I would be dying to know, and unfortunately we don't have clarity into this, what is the volume in the secondary market? Because if I, you know, I've had a couple of these calls from people, oh, my brother or my wife works at Open Air. Sell. Sell right now. Call Setter, whoever it is. Sell. I mean, and call me and I hope I'm wrong. But, you know, 435 times trailing revenues, or if it's 300 in the private markets right now, whatever that is, this is an extraordinary company. And then the asterisk