CEOInterviews.AI
Start App
Get an email when Paul Kedrosky gives a new interview
One email per new interview. Unsubscribe in one click.
Paul Kedrosky
Managing Partner, SK Ventures

Anthropic’s $1.5 Trillion IPO Is A Trap

📅 Nov 19, 2026 Prof G Markets 28 MIN 14984 VIEWS 41 SEGMENTS · 3 SPEAKERS
Ed Elson is joined by Paul Kedrosky to discuss what to expect from Anthropic’s first day of trading, and the stock’s long term trajectory. He also shares his take on the Nvidia scandal. Then, Joost van Dreunen joins to discuss the release of Grand Theft Auto VI, what makes the franchise so iconic, and why the game’s success is critical for the broader gaming industry. Finally, Ed checks in on New York’s pied-a-terre tax. Paul Kedrosky is a Managing Partner at SK Ventures. Joost van Dreunen is a Professor at NYU Stern who writes about the business of video games. Timestamps 00:00 - Intro 00...

Questions asked in this interview

7
  1. 1:10Any thoughts, takeaways, predictions for what this is going to look like?
  2. 5:02Would you say that that is accurate?
  3. 6:52But what is your expectation for this company and this stock over the next 12 months and perhaps even longer than that?
  4. 8:48What do you make of the business at this point?
  5. 10:10Is this what it looks like when a bubble pops?
  6. 12:14Who is most at risk as a result of that tightening cycle? I mean, who gets hurt?
  7. 12:55What do you make of all of it?
Ed Elson 0:00 ↗
It seems as though they are setting this up such that it cannot go wrong.
Paul Kedrosky 0:05 ↗
That is exactly what's happening. It's engineered for exactly that purpose. It has no relationship to the value or fundamentals of the business. It's purely an exit event. It's an engineered exit event partly for the series A, B, C, D, E, F, G, H investors and partly for the people who are coming in at the late stage and hoping to flip it to retail. But I also think you'll see a SpaceX-like 20 to 30% decline over the next sort of six months.
Ed Elson 0:30 ↗
Welcome to Prof G Markets. I'm Ed Elson. It is November 19th. Let's check in on yesterday's market vitals. The S&P 500 and the Dow fell again with Nvidia shares weighing on the indices. Nvidia slid another 3% following its earnings report. Still, the Nasdaq managed to post a gain. Meanwhile, Brent crude declined. So did the yield on 10-year Treasuries. And finally, companies close to Anthropic climbed in anticipation of its IPO. Amazon rose 4% and Broadcom rose 3.12%.
Okay, what's happening?
The moment for Anthropic's public debut has finally arrived. The company is set to start trading at $83 per share later today. That should put the valuation at about $1.5 trillion. And reportedly the IPO is close to seven times oversubscribed. As of last month, Anthropic was on pace for $81 billion in annual revenue, up roughly 1,000% from last year. But investors still have plenty to worry about. Anthropic posted a more than $500 million operating loss last quarter. And it has committed to spend roughly $900 billion on compute and infrastructure in the coming years. So, what will the public markets make of Anthropic? Here to help answer this question, we're speaking with Paul Kedrosky, managing partner at SK Ventures. Paul, Anthropic will finally go public. Many thought maybe it wouldn't or that it would be delayed. In fact, that was my prediction. That appears not to be the case. I got that wrong. This is happening. They're going public. Any thoughts, takeaways, predictions for what this is going to look like?
Paul Kedrosky 2:18 ↗
They're going to regret doing it the same day as GTA 6. That's my prediction. That was a terrible error. Who's in charge of that? Even Hollywood studios can get that stuff right. So, no. Although I will say, I mean, one of the things that's really startling to me, and I was looking at the data earlier today about the IPO, was comparing — and again, Google Trends data isn't great because apparently no one uses Google anymore — but the trends data, if you look at the level of it as a proxy for retail interest, is really interesting because you see that SpaceX was roughly about 4x the level of interest that Anthropic has at the same point in the cycle, day of IPO. Again, it's a very loose retail proxy, but it is interesting because I think a lot of people in the industry, in particular on the institutional side of finance, see this as a much more interesting IPO because it represents the first chance to get access to a frontier lab in terms of AI. And to see that I found really, really striking. And there's lots of ways you could look at it. I mean, retail interest is just one aspect of a successful IPO. I will say one of the things that I find interesting about it that frankly annoys me is this idea of it being oversubscribed. As I've said, I think maybe on this show before, is that being oversubscribed is required to go public. If you're not oversubscribed, you don't get to go public. So saying that you're oversubscribed, which they very noisily have leaked repeatedly to the point that it's like, okay, I get it already. Just, you know, run hot air balloons and tell me you don't have to keep leaking it anymore, is to say that that's just what you do. Because when you're the underwriter, you engineer the price, you engineer the allocation, you know the quantity. If you can't figure out a way to make sure there are more bids than there is stock, you need to get out of the business. So having said all of that, this for me is not much of an indicator in terms of what's going to happen. I still think it'll be successful, but that they feel the need to bang the drum so much with respect to it being oversubscribed is a proxy for retail interest because it's a way of attracting more retail investors. And the reason why you want them is because the institutions are getting the allocations in large part. So you need someone to flip it to. And so for me, it's a cynical exercise in making sure there's enough people waiting in the pond whenever you want to flip it after the IPO who can pick it up. And so by promoting how oversubscribed it is, it's a way to create this, you know, just-add-water feeding frenzy around it, and that's the way this should be interpreted. And turning it around, what the data says in terms of really dramatically oversubscribed IPOs is that they actually tend to underperform badly over the next 12 months. You have worldwide data on this — the US, China, or sorry, Hong Kong, Europe, and so on. So even insofar as it tells you anything, it mostly tells you that you're looking at an IPO that's likely to underperform if it is that oversubscribed.
Ed Elson 5:02 ↗
Well, this is the thing, because it does seem that this IPO — all IPOs are engineered by the underwriters. They're always priced such that the first day you get a good pop and bang, you have a successful IPO. If you don't have a pop, then something went horribly, horribly wrong. But this does seem especially engineered, and two reasons it seems like that to me at least. One, they have $26 billion coming in in investment coming in from a lot of their partners — Google, Amazon, Nvidia, all of these tech giants. So that helps the price. And then number two, the float, like SpaceX, is very small. It's around 5% of the company, averages 15 to 20% for most IPOs. So it seems as though they are setting this up such that it cannot go wrong. You'll have so much price support from the institutions coming in to support a relatively small amount of shares compared to the number of shares they could be selling in this IPO, which makes me think this is going to be — they're going to be fine, but it might not actually tell us that much about the company fundamentally. It might not actually tell us what the markets are really thinking. Would you say that that is accurate?
Paul Kedrosky 6:16 ↗
That could hardly be more accurate. That is exactly what's happening. It's engineered for exactly that purpose. It has no relationship to the value or fundamentals of the business. It's purely an exit event. It's an engineered exit event partly for the series A, B, C, D, E, F, G, H investors and partly for the people who are coming in at the late stage and hoping to flip it to retail or whomever that happens to be on the other side of it. So, it's engineered for that exit but doesn't tell us anything about what people think in terms of the prospects for the business. And then more fundamentally, we also know that when this phenomenon happens, it is highly predictive of underperformance over the next 12 months. We saw that with SpaceX. We've seen that with a host of others.
Ed Elson 6:52 ↗
So that gets to probably the more important question, which is what do we think is going to happen over the next 12 months? Because this thing will go out. It sounds like you and I both agree it'll be fine. It'll get that pop. It'll probably — people will say, 'Oh, good. Okay, good. It was a success.' By the way, anchoring from a much smaller number than previously reported, which was $2 trillion. So, we've reduced the price quite significantly. We'll see exactly what it goes out at over the course of the day. But what is your expectation for this company and this stock over the next 12 months and perhaps even longer than that?
Paul Kedrosky 7:27 ↗
Because that's a time frame in which you might see something closer to true price discovery and where the true feelings and emotions of Wall Street are actually reflected in the price. Yeah. So, I think you probably will see that kind of at least modest spike on the day. I'll hang my ass in the breeze and say that. But I also think you'll see a SpaceX-like 20 to 30% decline over the next sort of six months as price discovery actually happens that's separate from this phenomenon of artificial scarcity and an engineered price. And so you'll see something like that. And then as you look forward and you say at the current run rate what's a reasonable price, you know, 10x the current run rate wouldn't be too far out, would put you to $1 to $1.2 trillion, something like those kinds of numbers. It still looks, I'll say, rich at the price. And that's all not even talking about these two things we've talked about before. One is they're in this rapid transition from Anthropic 1 to Anthropic 2. Anthropic 1 largely leased compute and sold access to models. Anthropic 2 is building compute and is selling applications that sit on models. So you're in the middle of this really dramatic transition between two very different architectures of the same company. So the opportunities for revenue to look much less predictable than is implied by the valuation are dramatic. So I feel really comfortable with that kind of number.
Ed Elson 8:48 ↗
What do you make of the business at this point? Because as you say, they reduced their losses last year, but that doesn't really tell us much about what's going to happen in the future because they're going to spend roughly $900 billion.
Paul Kedrosky 9:00 ↗
So things are going to look very different, which tells me they're going to have to go out and raise a lot more after this IPO. And not only that, if you look at the data, they've been racing — in the last 18, 24 months, launching more apps, if you will, more verticalized products than they have new frontier models. And most people haven't noticed this, but it's really dramatic. This is a company already in the process of launching products that compete directly against people for whom they are distributors, Microsoft and others. So Microsoft distributes the product and yet competes with them. So they have to be very concerned about having their oxygen supply cut off by their own distributors as they race up market, and then at the same time as the people they're competing with are also users of their product and say, 'Wait a minute, am I providing you with data that helps you compete against me more successfully?' So the old game of being able to ingest lots of data to compete more effectively is already over. And so this is going to make it so much more difficult for them to pull this transition off for those two reasons: loss of distribution, and competitors or former customers who now see you as a competitor and are much more leery about giving you data.
Ed Elson 10:10 ↗
This, I think, brings us to what many would say is the more interesting news, or at least the more sensational news, which is what is happening to Nvidia and what is happening to AI at large. I mean, Nvidia has gotten clobbered. It's lost more than a third of its value. A lot of the AI names, many of them have been cut in half, particularly the neoclouds over the past six months. All of the semis, most of them at least, are in decline, down around 30% over about six weeks, which raises questions for Anthropic. I mean, we agree that this IPO will be fine, but if they have to go out and raise lots more capital in the future, then that means that they have to go out and find money from investors who right now appear to be very anxious about this whole AI thing. And so it does raise questions: can they do it? Is the euphoria, is the exuberance surrounding AI coming to an end? Is this what it looks like when a bubble pops?
Paul Kedrosky 11:15 ↗
It does. But I think there's a really unusual aspect to this, and this has become an overused metaphor, and I mean, I like to think I was one of the first to say it, but I probably wasn't, was that Nvidia is the kind of central bank of AI. This idea is so — it operated a fractional reserve system with GPUs as currency, and it tried to peg the currency back in the summer when it put a floor under it. That didn't work. It tried to run a fractional reserve system when it extended credit with no payment terms to many people in the circular financing cycle. What we've now entered with Colette as the interim CEO and Jensen on the out is what I would call in banking terms a tightening cycle. So we've entered a tightening cycle. So the cost of capital becomes more expensive. So to your point, it's not that you can't get the capital, but the price just went up. It became much more difficult to, and so that changes the underlying economics because if your assumption was I can have 20% margins and a 10% hurdle and this is my revenue nut by 2031, it just went up by 30%. It just went up by 30% because of this tightening cycle. That's what's changed.
Ed Elson 12:14 ↗
Who is most at risk as a result of that tightening cycle? I mean, who gets hurt?
Paul Kedrosky 12:20 ↗
Across the board, the neoclouds. And she tried very hard on the call last night to say we — she could have only done more, like maybe put in a heart emoji and say we still love neoclouds, but that didn't happen. So instead, she said, you know, we're still big supporters of neoclouds, but we're going to be much more discriminating in future. That 'but' was the telling 'but'. That's the one that told you that it's all about to change in terms of the cost of capital for the neoclouds because the central bank just decided the cost of capital needs to be different, and there's the group most at risk. And they are both the marginal providers of compute as well as already having the most expensive cost of capital, and it just got more expensive.
Ed Elson 12:55 ↗
What do you make of the entire Nvidia scandal? I mean, I would love to just get your top-line thoughts, especially this question of the return of Jensen Huang, which has a lot of investors very fired up right now. Many think that he has to come back. He is the essence and the soul of Nvidia. Others think no, this is a major scandal. He can't be CEO. What do you make of all of it?
Paul Kedrosky 13:18 ↗
Oh, I don't think he can come back. I mean, for me, this is anathema to the whole idea of board governance. I mean, why have a board? What's the point of having a board if it's something that the board has decided on its own was worth investigating? And to be clear, he didn't resign because he — or he resigned, he wasn't fired. He decided that he didn't want to be part of the process. But his actions, I think, are so difficult to defend from what we know in terms of his engagement with respect to what was going on in terms of loosely transshipment, but in particular just gaming US regulations with respect to, if not the letter of the law, the spirit of the law in terms of being a cloud compute provider to China via Southeast Asian data centers and clouds. And he said this very clearly, I don't know, it was six, eight months ago on Dwarkesh Patel's podcast where he defended it straight up and said, 'We're going to provide compute to China.' And this was a kind of regulatory arbitrage using clouds to do it. And the FT reported recently that something like 50% of their revenue this year, which they no longer report this way, now goes through their Singapore subsidiary. Granted, some of that — a significant fraction of that ends up back in the US again. But the point is there were things going on where he was essentially wrapping himself in the flag of nationalism and saying this is good for the US and it's also good for me. Don't worry your pretty little heads about it. Let us continue to do these things. So I think he was a very aggressive and brass-knuckle player trying to say that things that were good for Nvidia were definitionally good for the US. And I disagree.
Ed Elson 14:53 ↗
Paul Kedrosky is managing partner at SK Ventures. Paul, appreciate your time.

18 more exchanges in this transcript

Sign in free to read the rest of this interview. No card required.

Sign in to read the full transcript

Cite this transcript

APA, MLA, BibTeX
APA

Kedrosky, P. (2026, November 19). Anthropic’s $1.5 Trillion IPO Is A Trap [Interview transcript]. Prof G Markets. CEOInterviews.AI. https://ceointerviews.ai/interview/3333970/

MLA

Paul Kedrosky. "Anthropic’s $1.5 Trillion IPO Is A Trap." Prof G Markets, 19 Nov. 2026. Transcript, CEOInterviews.AI, https://ceointerviews.ai/interview/3333970/.

BibTeX
@misc{kedrosky2026_3333970,
  author       = {Paul Kedrosky},
  title        = {Anthropic’s $1.5 Trillion IPO Is A Trap},
  howpublished = {Interview transcript, Prof G Markets. CEOInterviews.AI},
  year         = {2026},
  month        = {nov},
  url          = {https://ceointerviews.ai/interview/3333970/},
  note         = {Speaker-attributed transcript with timestamps}
}