In short: Referenced only — Talkington (16:14): the recent "peak doomerism and peak regulation" came as Anthropic and OpenAI "really want liability coverage from the US government, which I don't think they're going to get."
In short: Referenced only — named with ChatGPT and Gemini as the assistants Meta "flanked" in reaching the consumer first, and (via the Sep-14 video) Dario Amodei "begging for regulation."
9:38But even if that happens, that doesn't change the game for Meta. Meta already understands that they're going to be competing with ChateBT and Anthropic. So that's not going to be any surprise. And I believe that there's a lot of good things coming to Meta. See, Muse is just a beneficial aspect of the story.
In short: Eisman: with OpenAI it is ~70% of hyperscaler AI revenue (25–35% of total cloud revenue); "anthropic is going to go public… they need to get their S1 out fast because token maxing is ending" — they want to list on second-quarter numbers "with token max as at its peak," not third. "But they're in better shape regardless. They're in better shape than open AI." The ecosystem still hangs on both.
Anthropic, maker of the Claude AI models, is private but expected to list soon. Eisman thinks it is in better shape than OpenAI, but notes that the two labs together supply most of the AI revenue at the big cloud companies, so the whole AI chain leans on both of them.
His timing point: the recent surge in AI usage ("token maxing") is fading. He expects Anthropic to rush its listing paperwork out so investors see the strong spring numbers, before the weaker summer quarter is reported.
17:35— That's done. Yeah. And so I don't think they want to go public having reported third quarter numbers. They want to go public having reported second quarter numbers with token max as at its peak. But they're in better shape regardless. They're in better shape than open AI. And if open AI gets in real trouble because the entire ecosystem is just dependent upon two companies.
In short: The centre of the pacing debate, now read through its IPO. "The AI debate took a sharper turn this weekend with Dario pushing for a slower pace of frontier model development," followed by Anthropic/OpenAI/Google talks on an industry standards body. The capex link: "that pace might slow down if OpenAI and Anthropic both delay their IPOs because they need to raise money to spend on all this capex." He prints Burry's rebuttal — "IPOs need hype and puffery… it could be a cover for a real uncontrollable slowing as these IPOs look to be pushed out further… into the end of '26 and '27" — and Trump's ("Dario… who is now pretending to be a perfect little angel"), with his own middle view: "the AI companies are self-serving, but I do think there needs to be… at least cybersecurity oversight and someone looking at safety." Also: Palantir, Nvidia and Booz Allen "reportedly restricting Anthropic's Fable model for sensitive work" over data retention.
Anthropic is a private AI company preparing to list its shares on the stock market. After its chief executive called for the industry to slow the development of the most powerful AI models, Anthropic, OpenAI and Google began talks on an industry body to test and audit AI, and Microsoft's chief executive welcomed "deliberate pacing."
Singh lays out the competing readings. The investor Michael Burry calls it self-serving: companies about to go public need hype, warning that your product is dangerous is a form of hype, and talk of slowing down could be cover for a real slowdown as their share listings slip into late 2026 or 2027. President Trump dismissed the whole idea as a hoax. Singh's own position is in between: the companies are self-interested, but some safety and cybersecurity oversight is still needed. The market angle he keeps returning to is money — if Anthropic and OpenAI delay their listings, they have less to spend on computing, and AI spending growth could slow.
Full passage: premium transcript (PDF).
In short: Sam: Dario "saw the risk-factor section of his [IPO filing] and literally lost his" composure; METR is "affiliated with Anthropic and effective altruism." RPK: the labs are chasing a "Mag 1" platform play that may give way to fragmentation. Referenced only — no position.
33:09So it's not like we need to really reinvent the wheel to do this. So you had these titanic forces of closed-source models and open-source democratization of technology battling out for share of voice, but I think Dario saw the risk-factor section of his [IPO filing] and literally lost his —
In short: Cited as evidence, not a stance — named with OpenAI as the private stakes whose appreciation is inflating reported S&P earnings.
3:36Look, why would you be a bear on stocks right now? Well, you might have a question mark on broad S&P 500 earnings. I think that's generally a valid question mark because there is some element of, in the non-dividend payers, reporting in your net income the appreciation of your OpenAI and your Anthropic stock.
In short: Passing mention — Claude is "not" what Muse is; Anthropic is expected to follow with its own agent.
5:04It'll actually search online for any applicable way to lower the price for you as well. And this is just a preview of the new agentic world that we've entered into. Muse is just the beginning. We know that chatbt, we know OpenAI is going to be coming out with their version of this very soon. Same with Anthropic and same with all the rest.
In short: Dario Amodei's call to "slow down" is "false on its face": "Anthropic is going to go public this year… maybe in a month or two. It can't slow down. What would it say to investors on the road show?" The labs see "a price war coming," with no pricing moats, so they are fomenting hysteria to win regulation "that will foster an AI duopoly." "It's unclear to me how to keep a shell game going in an IPO process that requires transparency."
Anthropic, the private company behind the Claude AI models, is expected to sell shares to the public soon, perhaps within a month or two, Eisman says. Its CEO has been warning that AI is dangerous and the industry should slow down. Eisman thinks that makes no sense coming from a company about to pitch fast growth to new investors, and one that has promised to buy huge amounts of computing power.
His explanation: AI models are becoming interchangeable, and cheaper free ("open-weight") models keep winning customers, so the big labs face a price war. Scary headlines could lead to government rules that only the biggest labs can afford to follow, which would protect them from competition. He doubts that approach can survive a public listing, which forces a company to disclose its real numbers.
15:17If they were to slow down, they could not fulfill those commitments. Also, Anthropic is going to go public this year, not just this year, maybe in a month or two. It can't slow down. What would it say to investors on the road show? Are they going to say that growth has been great, but now it's going to slow to a crawl? A slowdown contradicts their entire growth narrative? The only way to fulfill all those commitments is to not slow down.
In short: Its IPO timing and safety warnings hang over the AI trade (16:03). Ethridge: it "doesn't help" when the CEO of a top lab says "we have concerns over the safety of the product that we were planning to IPO in a couple of weeks." Goldman, he expects, leads the prospectus.
In short: "Can Anthropic go bust? Sure. But that's an equity problem, not a bond problem" — Anthropic isn't the borrower; the hyperscalers borrow to run its products. Cheap Chinese open-source models would cut the hyperscalers' costs, not their ability to pay.
The host worried that free Chinese AI models could make Anthropic's paid models obsolete and sink the AI debt pile. Bassman's answer is that Anthropic isn't the one borrowing. The big cloud companies borrow to build the computers that run Anthropic's models. So Anthropic can fail, and he says it could, but that would wipe out its shareholders, not the bondholders of Meta, Google and the rest. If AI models get cheap, the hyperscalers' costs fall and they can pay their debts more easily.
16:17The other guys are borrowing the money to use Anthropic's products. So, can Anthropic go bust? Sure. But that's an equity problem, not a bond problem. And I think the hyperscalers, I think Meta, Google, Amazon, Microsoft, Oracle, they have plenty of cash flow coming on through to fund, to make the coupon payments and pay these things back over time.
In short: One of the two labs the whole chain rests on — "such a huge percentage of the entire chain" — and the stronger of the two by implication ("between the two, OpenAI is the weaker company"). Lumped with the labs "trying to manufacture a crisis" on safety to win regulation-built moats; a host notes Anthropic "has had similar instances" of agents breaking out.
Anthropic, the private maker of the Claude AI models, is one of the two companies Eisman says the AI boom depends on. He sees it as the stronger of the two. But he groups it with the labs warning loudly about AI dangers, which he thinks is really a push for rules that would protect them from cheaper competitors.
3:06What I would say is that. This is how I sort of think about the chain. — Sounded like you just described, that we're at some type of tipping point for AI if they're desperate to create a narrative because things aren't going. — Well, I think we're potentially closer. But I think the issue is that at the end of the day, the entire AI chain, from Nvidia to the hyperscalers to Anthropic and OpenAI all depends on the future health of Anthropic and OpenAI, because there's such a huge percentage of the entire chain. So if something were to
In short: He mocks the claim that it is "profitable now if you remove the cost of training the model," since that is "your cost of goods sold." He reads Dario's regulation push as a bid to be a utility-like monopoly with no need to compete: "either one or the other," a business case or the AGI moonshot. The IPO is being pushed back, and the S-1 is still undisclosed.
Anthropic is a leading AI lab heading for an IPO that has been pushed back. Deluard ridicules its claim to be profitable "if you remove the cost of training the model." For an AI company, training new models is the core cost of staying competitive, the equivalent of a factory's raw materials.
He suspects the push for AI regulation is really a way to become a protected, utility-like monopoly that no longer has to keep outspending rivals. His point: you either have a normal, regulated business or a moonshot, not both. Privately held, nothing to buy.
38:21There must be some reason why it's not out yet. But it is hilarious. They recently came out and said, "Yeah, we're profitable now if you remove the cost of inference" — the cost of training the model. Yeah. — The cost of training the model. And it's like, well, yeah, because that's your cost of goods sold.
In short: The Future Proof announcement: Claude tools for financial advisors, with Brown's Ritholtz Wealth a design partner. Brown: "a landmark situation for the wealth management industry… we have spent the last 10 years basically cobbling together hundreds, if not thousands of different point solutions… nobody has come up with an intelligence layer that can pull all of those things together… now, for the first time, having Claude in our ecosystem, pulling together all of our systems, reporting to both us and our client in real time… this is, in my opinion, the Holy Grail… I think everything is about to change." "There are 12 logos on there from companies like Schwab and BlackRock" on the provider side. Separately, Rooney: Amodei's essay "kicked all of this off," and OpenAI says it is working with Anthropic and Google on oversight. (Brown's firm is a partner — an interested party.)
Anthropic, the private AI company behind the Claude chatbot, used the Future Proof conference to launch tools for financial advisors. Brown's firm helped design them, so he is not a neutral voice — but his description of the problem is useful.
Advisory firms run dozens of separate software systems for accounting, trading, rebalancing and client reporting, and staff spend hours moving data between them. An AI layer that connects them all and briefs the advisor before each client call would free up time for the human part of the job. If it works, it is a sign that AI's first big payoff in finance is in back-office plumbing rather than in picking stocks.
In short: A survivor: "Google and Anthropic are going to be two of the big survivors," because "Anthropic from day one was really focused on corporate. Why? Because corporations will pay." Now "much bigger than OpenAI… about 65 billion annualized run rate and OpenAI is right around 40," with the ratio reversed from two years ago. The skeptic's note: a frontier leader has an interest in regulation that chokes off open weight — "if you can give yourself a monopoly, wouldn't you?"
Anthropic makes the Claude AI models and focused from day one on selling to businesses, "because corporations will pay." Niles counts it with Google as one of the two likely survivors. The numbers support him: its annualized revenue is now around $65 billion versus about $40 billion for OpenAI — the reverse of a couple of years ago.
He adds a skeptical note on its CEO's safety warning: whichever lab is in the lead benefits if regulation makes life harder for free, open-weight competitors. He doesn't dismiss the safety risk, but says "there can be multiple forces at play at once."
15:57Google is integrating their consumer-based AI into the normal search product that everybody uses. So, they're in a great position on consumer, and Anthropic from day one was really focused on corporate. Why? Because corporations will pay. So, I feel like OpenAI in particular are stuck between those two guys.
In short: With OpenAI, ~70% of the AI market and ~$100B+ of combined revenue by year-end, "out of private financing." Still: "I think Anthropic is going to get their deal done… and I think Musk wants them to get their deal done." Anthropic survives the shake-out, but inside a revenue pool too small for the capex.
Anthropic makes the Claude AI models, which Taylor uses himself (he read a 280-page contract with it in an hour). Together with OpenAI it takes about 70% of AI spending, but he thinks their combined revenue is far too small to pay for the trillions of data centers built for them, and private investors have run out of appetite. Even so, he expects Anthropic's current fund-raise to close, helped by Elon Musk wanting it to. He sees it as a survivor of the shake-out, not a winner of the math.
17:27And I said, the bonds are telling me this doesn't make sense. Are we close? And then when I did it, clutch the pearls. And I was like, sweet Jesus. So what happens now, and that's what matters, I think Anthropic is going to get their deal done. They have to jam that down the streets. I think they're going to get their deal done, and I think Musk wants them to get their deal done.
In short: The source of the day's selloff — and, per Rooney's sources, adjusted-profitable into its IPO. Rooney: "Anthropic CEO Dario Amodei really shocked the tech world with an essay out on Saturday… a three-step plan aimed at tempering how fast some of the most powerful AI models… are going to be improving… third party evaluators… safety standards and then international coordination," citing AI "developing a lot faster than expected" and OpenAI agents hacking Hugging Face. On the listing: some argue "the safety push could actually help Anthropic's IPO, positioning the company now as a responsible player"; "I also confirmed… Anthropic has seen back-to-back quarters of at least adjusted profitability." Terranova suspects OpenAI's pause support is about "putting Anthropic in a position that we don't see them IPO in 2026"; Santoli: "the key will be… Anthropic's S-1… how they're characterizing their actual profitability"; Weiss wants the risk factors (David Sacks flagged product liability for agents).
Anthropic is a private AI lab heading toward a stock-market listing. Its CEO's essay calling for outside evaluators and international coordination on the most powerful AI models is what knocked chip stocks down on the day.
The investing angle is the IPO. Some investors think the safety stance helps (less legal and regulatory risk), and CNBC's reporter confirmed two straight quarters of "adjusted" profitability — a profit figure that excludes some costs. The committee's point is that the real test is the S-1, the public filing a company must publish before listing: it will show how profitable the business really is once everything it pays for is counted, and what risks (like liability for AI agents) it has to disclose.
In short: Wouldn't buy the IPO "not at two trillion dollars" — secondary shares trade nearer $1.3T, so insiders sell six weeks out; Amodei's slow-down essay reads as roadshow damage control; the listing is "a referendum of the entire AI trade."
Anthropic, the maker of Claude, is preparing a stock-market listing at a roughly $2 trillion valuation. Woo wouldn't buy at that price. In private trading its shares were recently priced nearer $1.3 trillion, which means insiders would rather sell now at a big discount than wait six weeks for the listing.
He also sees the CEO's recent essay urging slower AI development as a response to investors on the pre-listing tour who kept asking about China and cyber-security. And he calls the listing a "referendum" on the whole AI boom: if it goes badly, the AI trade could unravel.
35:22By the way, if you look at the secondary market, just so that you know, I haven't looked today but last week for example, there is a secondary market for anthropic shares in the private market. They say they want to raise at $2 trillion. The shares, if I recall, the last time I had looked, were trading more like $1.
In short: Same analogy: "look at the private market value of Anthropic" as an example of how fast value gets recognized — a rerating he expects to spread into the resource sector.
40:39And if there's graphite, graphene, lithium in Canada, those, hydrogen, all of those are going to be important. And you need to be looking at, a lot of them are smaller and speculative. But look what AI did from two years ago to today. Look at Nvidia stock. Look at the private market value of Anthropic.
In short: The leader asking everyone to slow down. His test: if its models could end civilization, "why don't you just slow down?… you could do this easily without an act of Congress" — but Dario wants a pace that "won't sacrifice commercial advantage." The real motive in his reading: ~half a trillion dollars of spending commitments, investors promised profitability (FT: profitable a second straight quarter), an IPO coming — "We need to show healthy margins, good cash flow" — so a coordinated pause lets it cut training spend while cash-rich Meta is forced to stop catching up. "Even if Dario's concerns about AI are sincere… it's completely irrelevant." Also criticised for doomerism and the Manhattan Project comparison.
Anthropic, maker of Claude, is not listed on the stock market, but its CEO Dario Amodei just called for all leading AI companies to slow development together under government rules. Carlson is negative on the motive, not the technology.
His test is simple: if Amodei truly believes his own models are that dangerous, he can slow Anthropic down tomorrow without asking anyone. Instead he wants a slowdown only if everyone else slows too, so Anthropic keeps its lead. Carlson thinks the real reason is money. Anthropic has reportedly committed around half a trillion dollars to computing, has promised investors profits, and is heading toward a stock-market listing (IPO), where it needs to show healthy margins. A rule forcing everyone to spend less would let Anthropic cut costs while stopping cash-rich rivals like Meta from catching up. Using rules to protect your position from competitors is called "regulatory capture," and that's his charge.
26:11It'd be very easy for you to do. Well, you're not doing that. And there's a reason why. See, Anthropic is in an interesting situation. Anthropic already has the lead technologically. They have a big lead over other companies. OpenAI, I would say, is very close. They're neck andneck, but Anthropic and OpenAI lead the rest of the pack.
In short: Eisman: "Anthropic put out like 11 and a half billion in revenue for the June quarter… up over 100% in 3 months" — the strong lab. But with OpenAI it is ~70% of hyperscaler AI revenue (25–35% of total cloud revenue), and Collins: "they're going to let Anthropic come public because they need to… just like SpaceX, it's bad for the market" — new supply. An enterprise friend now sends only "a small percentage of the very very important queries" to Anthropic/OpenAI.
Anthropic, which makes the Claude AI models, is private and growing very fast: about $11.5 billion of sales in a quarter, double the previous quarter. Together with OpenAI it provides roughly 70% of the AI revenue at the big cloud companies, so they matter to the whole AI chain.
The partners worry about two things. Companies are learning to send only their most important work to these expensive models and the rest to cheap free models. And when Anthropic lists on the stock market, it adds a huge amount of new stock for investors to absorb, which historically weighs on markets.
22:50The problem is that the Open AI numbers that have just come out are actually quite poor. — Yes. — Compared to Anthropic, — they had huge market share and they just they lose it every day. — Well, and they keep losing it. And Steve, you're closer to it than we are. The sequential growth rates, are they slowing, accelerating? — So, Anthropic put out like 11 and a half billion in revenue for the June quarter.
In short: The weekend's catalyst, and a valuation he now treats as a ceiling. "Anthropic CEO Dario Amodei publicly called for a formal industry-wide slowdown of the pace of frontier AI model development… the trigger was a security incident involving an autonomous agent swarm… the OpenAI–Hugging Face incident," with Anthropic committing to external auditors like METR and Altman and Musk backing the call. The IPO read: "we think that he's going to delay his IPO till October at the earliest, next year is the most likely," against investors "pushing hard towards an October 26th IPO reported at a close to two trillion valuation… apparently this is a dangerous enough moment that going public is ill-advised, unless, of course, you happen to be a company trying to sell two trillion of equity into it." The letter itself "is not a growth warning… he does not touch the numbers at all… but the fact that he felt the need to explicitly reassure investors… tells you he knows exactly how the market is going to hear it." His mark: "I doubt Anthropic is going to be worth more than 1.7 trillion anytime soon." And the systemic link: "the real risk is that Anthropic or OpenAI run into funding issues and slow down cloud spending, which in turn lowers the backlog growth of Microsoft, Google and Amazon, who then… cut capex… but we won't know that for a while."
Anthropic is a private AI company preparing to list its shares on the stock market, reportedly as soon as late October at a value near $2 trillion. This weekend its chief executive, Dario Amodei, called for the whole industry to slow the pace at which the most powerful AI models are developed, after an incident in which a group of autonomous AI agents behaved in dangerous, unintended ways. Anthropic said it would let outside auditors check its training and releases.
Singh's reading has two halves. On the business, the letter changes nothing: it contains no delayed launches, no spending cuts, no lower revenue targets. On investor psychology, it is awkward timing for a company about to ask the public for an enormous sum — especially as OpenAI's Sam Altman, whose company is not listing this year, called this "an ill-advised moment to go public." Singh now expects the listing to slip to October at the earliest and more likely next year, and doubts Anthropic will be valued above $1.7 trillion any time soon. The larger danger he names is financial: if the AI labs cannot keep raising money, their cloud spending slows, and that ripples back through the biggest technology companies.
Full passage: premium transcript (PDF).
In short: An unverified aside on asymmetric warfare: "I think Anthropic or one of the AI companies was complaining because evidently… somebody was using the AI for targeting of these various facilities… I don't know if that's true or not." A symptom of "garage band warfare," not a view on the company.
6:16Anthropic or one of the AI companies was complaining because evidently either it was the Yemenis, the Houthis, I don't know who, somebody was using the AI for targeting of these various facilities and so they were all up in arms about that. I don't know if that's true or not but this is what I'm talking about, asymmetry okay, garage band warfare if you want to call
In short: Safety alarms ahead of a potential IPO. "Anthropic researcher Jacob Coxon quit, warning that leading AI labs are 'gambling with our lives.' Anthropic alignment lead Evan Hubinger went further, putting the odds of AI killing all humans within the next decade above 10%. Not exactly ideal PR ahead of a potential IPO." Also named with OpenAI as the chatbot-centric strategy Apple is not pursuing. News item; no stance.
In short: Why she isn't worried about two-lab concentration: "Anthropic's not going to invent a vacuum that can go clean your whole house… they're trying to train something that's very broadly applicable" so others can solve real problems cheaply. Critical of its PR: "anthropics entire like media strategy or thing that they tell employees to like talk about the fact that we think that we're all going to die in a few years like this feels… crazy and insane to me."
In short: The company at the centre of the week's AI backlash — and, per Gerstner, its own best safety exhibit. A departing researcher who "spent the last three years doing pre-training research at both OpenAI and Anthropic" posted that "neither company is acting responsibly… they're racing straight to self-improving superintelligence and gambling with our lives." Gerstner (an investor) answered on X: "ridiculous hyperbole from an ex junior employee who worked a total of 6 weeks at Anthropic." His substantive defence is the firm's own disclosure: "they're issuing transparency reports like we saw out of Anthropic yesterday" — 154 pages detailing blocked misuse, including bioweapon attempts and a group in Yemen using Claude to try to develop missiles, plus Alibaba's industrial-scale distillation of the models. Wapner's counter is that the report reads as evidence of the risk, not of its containment. Harrington separately flags the coming S-1 as a market event.
Anthropic is a private AI lab (Altimeter, Gerstner's firm, is an investor). The week's controversy began when a researcher who had worked at both Anthropic and OpenAI resigned publicly, saying neither company is acting responsibly and that people building AI believe it could kill everyone by the end of the decade.
Gerstner's rebuttal is partly about the messenger — six weeks at the firm — and partly substantive: Anthropic had just published a 154-page report on attempts to misuse its systems, including blocked bioweapon research and a group in Yemen trying to use Claude for missile development. He reads that as proof the safety systems work and that the company is unusually transparent; Wapner reads the same document as evidence the danger is concrete rather than hypothetical. The report also disclosed large-scale "distillation" by Alibaba — copying a model's capabilities by training a cheaper model on its outputs.
In short: Half of the dependency he keeps as his caveat — the current AI ecosystem is "almost completely dependent, on the health of Anthropic and OpenAI." After OpenAI's price cut: "I wonder if Anthropic will have to follow with its own price cut." Also: a former Anthropic researcher's extinction warning, on which "count me a skeptic."
14:55And I'll admit this could be correct, but with one major caveat. If Ed Zitron is right, the current, and I do emphasize current, AI ecosystem is dependent, almost completely dependent, on the health of Anthropic and OpenAI. Now, one of them could fail before these new companies are capable of replacing the commitments of Anthropic and OpenAI, and this could be like the dot-com bubble bursting.
In short: Named by the host alongside OpenAI as the AI IPOs that could test the theme; Ciampaglia treats them as competition for capital rather than a threat to nuclear demand, which rests on base load, reshoring and utilities' first load growth in two decades.
2:10But obviously, we're seeing tremendous new sources of demand for uranium in the coming years. One of the new sources is from AI and I want to get your thoughts on this whole AI narrative because this whole ecosystem is really based on two companies, OpenAI and also Anthropic. They're going to go public sometime this year and there's all sorts of speculation out there on how these IPOs will go.
In short: Negative on motives: "Anthropic has been on an outright campaign for regulatory capture for a long period of time" — strict, complex rules that raise the barrier to entry for smaller competitors, plus reluctance on open source. Its staff "view themselves as the most important people in the world saving humanity"; researcher Evan Hubinger's "above a 10% chance" of AI killing all humans within a decade is cited as the mindset. Early funder Jaan Tallinn (a notable AI skeptic, tied to the Survival and Flourishing Fund) invested "to have control over it."
Anthropic is the AI company behind Claude. A former researcher quit with a viral warning that AI could kill everyone within ten years, and politicians quickly used it to push for pausing AI development. Carlson doesn't judge whether that researcher is sincere, but he is skeptical of Anthropic's motives as a company.
His argument is "regulatory capture": when an established company lobbies for complicated, expensive rules, it can afford them and smaller competitors can't, so the rules protect the incumbent. He sees Anthropic pushing for strict regulation, resisting open-source models, and describing its own staff as the only people responsible enough to build powerful AI — a mindset that conveniently justifies limiting everyone else. He thinks a US slowdown would not make AI safer, just hand ground to China.
17:35clear. Anthropic has been on an outright campaign for regulatory capture for a long period of time. In fact, one of the people that funded Anthropic was Yan Talon, a notable skeptic of AI, someone that doesn't even really like AI. And the reason that he originally funded or
In short: Split call on the IPO, not on the company. Le Shrub ("Claudification Overdrive"): the S-1 "has to succeed," so expect weeks of bullish AI narrative, and "it's very obvious to me it's going to get done" — valuation went from "a trillion" to "two trillion" in two months and "no one blinked"; the question is whether it is "the absolute meme top." Paulo: "not obvious to me that this thing is going to get done" — DeepSeek, semis off highs, safety/guardrail noise and a possible Democratic midterm sweep could push it out and bring it back "a lot cheaper"; a missed window would itself mark "a near-term low in the market."
Anthropic, the maker of the Claude AI models, is preparing a stock-market listing at a reported value of around $2 trillion. The two hosts disagree about whether it happens on schedule, not about the company itself.
Le Shrub thinks banks and fund managers earn so much from the deal that they will flood investors with upbeat AI news until it is done — and that it will be done, possibly marking the peak of the AI mania. Paulo is less sure: a cheap new Chinese model, political pressure on AI companies, local opposition to data centers and a possible change of control in Congress in November could delay it and force a lower price. His twist is that a delay could actually help the wider market for a while, because a very large new share sale soaks up money that would otherwise go into existing stocks.
In short: The other half of the pair, treated jointly rather than ranked here — no separation into a stronger and weaker lab of the kind Eisman made himself on Aug-28. Both are inside the ~70% of hyperscaler AI revenue and the 48% of Google Cloud next year figures, and both "remain deeply unprofitable."
Anthropic is the other half of the pair, and in the captured section it is not separated out from OpenAI at all — both are inside the "~70% of hyperscaler AI revenue" estimate, both inside the UBS projection for Google Cloud, and both described as deeply unprofitable.
That joint treatment is itself notable for this archive: in his own interview two weeks earlier Eisman ranked the two explicitly, calling Anthropic the stronger and OpenAI "the weak sister." The guest's framing here does not make that distinction, because the circularity argument does not depend on which lab is healthier — it depends on both being funded by their own customers.
In short: Named alongside OpenAI as a model developer Shopify is making its merchant catalog easily available to, positioning Shopify at the center of agentic commerce.
In short: The private mark that the whole AI credit chain is discounted off, and it is going the wrong way. "Anthropic's valuation has actually come down — estimates from about 1.4 trillion to about 900 billion. They've delayed the IPO again to mid-October. This is as Chinese models gain share." It matters in three places on this call: it is one of the four named customers inside Broadcom's 20-gigawatt, through-2028 commitment; it is the counterparty on a reported $35 billion cloud deal with Lambda, an Nvidia-backed cloud provider; and it is the source of the unrealised marks JPMorgan Asset Management flags as "an unusually high share of S&P 500 earnings due to unrealized valuation gains" — though on his own recomputation, "when you take away the other income or valuation gains, S&P 500 earnings is still very strong at above 30%." (Product news on the same page: Anthropic "launched Claude Fable 5.1 and Mythos 5.1, with Fable 5.1 positioned as the most advanced model yet for coding.")
Anthropic is a private AI laboratory, so it has no share price — but its estimated value has become load-bearing for the public market, because Amazon and Alphabet own stakes in it and book the paper gains through their profits. That is why JPMorgan Asset Management flags "an unusually high share of S&P 500 earnings due to unrealized valuation gains."
This week the mark went the wrong way. Estimates of Anthropic's value have fallen from roughly $1.4 trillion to about $900 billion, and the initial public offering has been pushed back again, to mid-October. Singh attributes it to competition: Chinese open-weight models are taking share, and the price of AI processing has halved in three months, which squeezes the revenue any lab can charge for the same work.
It matters in two other places on the same call. Anthropic is one of four named customers behind Broadcom's commitment to enable more than 20 gigawatts of computing capacity through 2028, and it has reportedly signed a $35 billion cloud contract with Lambda. A great deal of the AI infrastructure story is underwritten by a private company whose own valuation just fell by a third. The reassuring counterweight Singh supplies: strip the unrealised gains out entirely and S&P 500 earnings growth is "still very strong at above 30%."
Full passage: premium transcript (PDF).
In short: The same peer set, and the butt of Nathan's point about model-leaderboard narratives: "the idea on one given day that a stock like Meta is up three and a half percent because some third-party rating is just basically saying that Meta's new model is that much better than Anthropic's Claude 5892 or whatever… it's so dumb." The version-number joke is the argument — leadership that flips week to week is not a moat.
30:48That sort of thing. It's so dumb because at the end point, and you know this, and I know that as far as tech is concerned, you pay attention, you don't really give that much about it, right? I'm just saying the actual tech, but you are fairly certain, as I am fairly certain, that all of these models have a ball people, spend hundreds of billions of dollars training them, go back and forth, this that whatever.
In short: The sustainability datapoint: "what's more impactful is the fact that they actually generated profits. Now, they're saying it's adjusted profit, so we don't know what's in that adjustment, which always drives me crazy, but they say they got to profitability in the June quarter." With OpenAI it leads at the frontier — but "90% of the time people will use a Ford."
Anthropic makes the Claude AI models. The host notes its annual revenue run rate has jumped enormously this year, but what Niles cares about more is that it says it turned a profit in the June quarter. That matters because the biggest doubt about AI is whether anyone can make money selling it rather than just growing fast.
He keeps a healthy suspicion: it's an "adjusted" profit, and "we don't know what's in that adjustment, which always drives me crazy." And he still expects most everyday AI work to go to cheaper, simpler models — you don't take a Ferrari to buy milk — which caps how much the top labs can charge for most tasks.
22:09But even more importantly to me was the profitability of that business actually improved by operating margins expanding by 2% as well. And similarly you cited the Anthropic ARR which is interesting and obviously those are huge numbers but to me what's more impactful is the fact that they actually generated profits. Now, they're saying it's adjusted profit, so we don't know what's in that adjustment, which always drives me crazy, but they say they got to profitability in the June quarter as well.
In short: The better of the two labs, in a layer he is fading. On the company: "I've been pretty negative on OpenAI relative to Anthropic. And now you're seeing it in the revenues… they turned profitable in Q2," it wins "on the enterprise side," it is "going public first" and "should go public at close to two trillion in valuation." On the layer: "the value is going to move from the model providers to the infrastructure providers" — "people aren't going to be using Anthropic to go summarize their emails."
Anthropic is the private AI lab behind the Claude models, sold mainly to businesses. Niles is clearly positive on it relative to OpenAI and clearly cautious about the whole layer it lives in — which is why the stance here is neutral rather than positive.
The favourable half is concrete. It "turned profitable in Q2" while OpenAI appears to have lost more money than the quarter before; it started with enterprise customers, who actually pay, rather than consumers, who have been trained to expect free; it is going public first; and he expects it to list "at close to two trillion in valuation."
The unfavourable half is structural and applies to Anthropic too. Open-source models have cut what anyone can charge per unit of AI output by roughly half since May, and most tasks never needed a frontier model — "people aren't going to be using Anthropic to go summarize their emails." His conclusion is that "the value is going to move from the model providers to the infrastructure providers," i.e. to whoever owns the computers, and then to the businesses built on top of cheap AI. Being the best model vendor is worth less than owning the pipes.
48:24And now you're seeing it in the revenues. And so it's that same process, and profitability, by the way, right? Anthropic is very pro, or not very profitable, but they turned profitable in Q2. And OpenAI, I believe, lost even more money in Q2 relative to Q1. We haven't seen the final figures yet. And so it's kind of the same process that you're going through on the public market side as you're trying to come up with the names that will do well or not.
In short: Paired with OpenAI in the same clause as the profitless corner of an otherwise profitable market — "except for maybe in OpenAI and Anthropic, but there are profits." No company-level view; the mention is doing valuation work, not stock-picking work.
Anthropic is the other large private AI lab named, paired with OpenAI in the same sentence.
Its role is identical: the profitless corner of a market Dillian otherwise credits with real earnings. That distinction matters to his argument because a bubble in profitable companies breaks differently — and usually less catastrophically — than one in companies with no earnings at all.
There is no company-level view here; the mention is doing arithmetic, not stock-picking.
31:31dot-com bubble 25 years ago, 26 years ago but the one thing that's different is there are profits. I mean, except for maybe in OpenAI and Anthropic, but there are profits. So yes, but that revenue growth in Anthropic and OpenAI has been tremendous. So the bear argument was like where's the revenue in the labs, OpenAI, Anthropic. And I just want to say as someone who's kind of calling balls and strikes, the revenue growth has been quite quite good, like among the best ever for history of companies.
In short: Named twice, both times as the frontier leader. Lebenthal, arguing Alphabet needs its new model to land: "about nine months ago, they were on the top of the heap. Since that time, Anthropic has clearly taken over in terms of the most popular, most used large language model. OpenAI is not being left behind either." Weiss names it again as part of the competitive threat to the cybersecurity vendors: "we see every day there are more announcements from OpenAI, from Alphabet, from Anthropic about their cyber tools they're putting out there." Private — no ticker, no position possible; cited as a competitive fact about two public trades.
Anthropic is a private AI lab, so there is no way to own it directly — but it moves two public positions on this show, which is why it is tabled.
First, it is named as the reason Alphabet's stock is down: an owner of Alphabet says Anthropic "has clearly taken over in terms of the most popular, most used large language model," and that Google needs its new release to land. Second, Weiss lists it alongside OpenAI and Alphabet as a source of new cybersecurity tools, which is part of why he thinks the security vendors' valuations are unsafe.
The general lesson is worth carrying: a private company you cannot buy can still be the thing determining what your public holdings are worth. Tracking the frontier-model leaderboard is not a technology hobby if you own Alphabet or a software vendor — it is position monitoring.
In short: Cited with OpenAI: growing "so fast because it's incredibly useful" — companies expand usage after trying it rather than cancelling.
In short: (Private.) Named once and load-bearing: it is the entire reason Terranova wants to be long Zoom after its post-earnings pullback — "take the other side. You want to be long there because of the Anthropic relationship." Anthropic's Claude also appears in Mackenzie Sigalos' Apple reporting as the kind of third-party model Apple might monetize through a Siri AI tier.
In short: The S-1 lands "next week or the week after"; he expects little genuinely new information (the numbers are broadly leaked) but says having the labs public replaces third-hand ARR leaks with metrics and dampens narrative volatility. On the listing itself: "I think it's going to trade at a crazy price. I'm not saying I'll buy it, but there's going to be a lot of enthusiasm for it."
Anthropic is one of the two leading AI labs and is about to file the paperwork to go public. Keller doesn't expect the filing to reveal much — the revenue figures have largely leaked already — but he thinks having a lab publicly listed is genuinely important for everyone else. Right now the entire AI complex trades on third-hand rumours about lab revenue; once there are audited numbers, the whole sector gets a real gauge instead of guesswork, and the narrative swings should calm down.
On the listing itself he's deliberately non-committal: "I think it's going to trade at a crazy price. I'm not saying I'll buy it, but there's going to be a lot of enthusiasm for it." The market wants direct exposure to an AI lab and currently can't get it — you can only buy it wrapped inside Google, or indirectly through chipmakers. He's also relaxed about the money it will absorb: the labs, SpaceX and Google will have raised roughly $500 billion this year and markets took it in stride.
1:00:50I don't know that means it's going to rip, but I think that the market's ready to digest Anthropic, and I think it'll want lab exposure. Frankly, I think it's going to trade at a crazy price. I'm not saying I'll buy it, but I think that there's going to be a lot of enthusiasm for it.
In short: The IPO is now a load-bearing macro variable — and the pitch is one he openly mocks. "Anthropic is preparing for a public listing that could target a $2 trillion valuation and seek up to $100 billion in capital, more than SpaceX, pitching a $30 trillion total addressable market — which is bananas because US GDP is only $32 trillion" (the deck: topping SpaceX's $28.5T, on $11.6B of Q2 revenue, documents expected within weeks for a September/early-October listing). The revenue is real and fast — annualised run rates for Anthropic and OpenAI together went from $29 billion at the start of the year to $105 billion seven months later, most of the growth coming from Anthropic — with Meta alone at up to $10 billion a year and Salesforce building native Claude execution into its product. But the dependency runs the other way too: part of Nvidia's 70% growth guide "is contingent on financing for Anthropic doing an IPO in October and OpenAI securing financing." And its mark-to-market is what inflates GAAP index earnings at Amazon and Alphabet.
Anthropic is preparing to float at a target valuation of $2 trillion, raising up to $100 billion — more than SpaceX raised — and marketing to investors a total addressable market of $30 trillion. Singh's reaction to that last figure is not analytical, it is arithmetical: "which is bananas because US GDP is only $32 trillion."
The revenue underneath is nonetheless extraordinary. Annualised revenue across Anthropic and OpenAI went from $29 billion at the start of the year to $105 billion seven months later, with most of the growth at Anthropic; the company reported $11.6 billion of revenue in a single quarter. Meta alone is reportedly paying up to $10 billion a year, and Salesforce has built the ability to execute its own software's actions from inside Claude.
The reason a private company matters this much to a public-markets call is the dependency running the other way. Part of Nvidia's 70% growth forecast, Singh says explicitly, "is contingent on financing for Anthropic doing an IPO in October." And the paper gains on Amazon's and Alphabet's stakes in Anthropic are what inflate reported American index earnings from 31% growth to 118.5%.
So one unlisted company's flotation now sits underneath the largest listed company's forecast and the index's headline earnings at the same time. That is the concentration risk of this cycle stated in a single sentence.
Full passage: premium transcript (PDF).
In short: Named on both sides of his bear case. As supply: the coming listing (with SpaceX) is part of the $4–5.5T issuance wave funds must sell liquid winners to absorb. As product: "there is very little difference between a frontier model from Anthropic and a frontier model from Qwen or from DeepSeek… in practical composite terms" — so "the game is almost over in terms of pretending that you can justify multi-billion dollar training runs," and "the most successful frontier AI company will be the first one to stop pretending they can train new AI models."
Anthropic appears twice in his bear case, and neither is about the quality of the product.
First, as supply. If Anthropic and SpaceX list, the combined issuance is roughly $4–5.5 trillion — more than every US IPO since World War II put together. Funds don't hold spare cash, so buying into those listings means selling something else first, which pressures whatever is already owned.
Second, as product economics. Kedrosky argues the frontier models have converged: in practical terms he can't tell an Anthropic model from Qwen or DeepSeek, and neither can the people he blind-tests. If the best model isn't meaningfully better than a cheap one, then "the game is almost over in terms of pretending that you can justify multi-billion dollar training runs." His provocation: the first frontier lab to stop training new models gets a windfall, because it stops burning the money.
36:33And that variance has collapsed. There is very little difference between a frontier model from Anthropic and a frontier model from Qwen or from DeepSeek or somewhere else, in practical composite terms. People delude themselves and say there's a huge difference. But I do some testing every once in a while just to piss people off where I'll literally do like a Pepsi-Coke test and say, "Okay, you think you can tell the difference? I'll put them in front of you and see if you can.
In short: The stronger of the two, for now: June-quarter revenue "was like 11 and 1/2 billion… up over 100% versus the March quarter." But the run is explained away by a demand condition that has since ended — "sometime around late June, July, token maxing… ended" — so "the third and fourth quarters are going to be much more interesting… than the first half." The test is the S-1: "I'm hoping to get some clarity on Anthropic when they put out their S-1. Whether they'll give it or not, I don't know."
Anthropic is the other pure model company. On the numbers it is the healthier one: June-quarter revenue of about $11.5bn, more than double the previous quarter.
Eisman's caution is about why that number is so good. Through the first half of the year, he says, customers were "token maxing" — using AI without watching what it cost, the way people spend on a new toy. That behaviour ended around late June or July. So the strong first half measures a spending mood that no longer exists, and the real test is the second half: "the third and fourth quarters are going to be much more interesting for Anthropic and Open AI than the first half of the year."
He is waiting for one specific document. An S-1 is the disclosure a company files before going public, and it would show these revenue and cost figures properly for the first time — "I'm hoping to get some clarity on Anthropic when they put out their S-1. Whether they'll give it or not, I don't know." Prediction markets put Anthropic at 93% to list before OpenAI, so that filing is probably the first real look anyone gets.
6:42And that's because sometime around late June, July, token maxing, which is where people were just spending whatever they had no sensitivity to spending. They just spent money like crazy on tokens. Token maxing ended. And now people are a lot more self-conscious. So, I actually think the third and fourth quarters are going to be much more interesting for Anthropic and Open AI than the first half of the year.
In short: Its private-market mark-up supplied most of Salesforce's EPS beat ($5.06 of $5.90 from equity gains).
In short: The partner that makes Salesforce's "headless" strategy concrete. Claudeforce is "an expanded partnership with Anthropic" under which "Claude can now access Salesforce customer data and execute Salesforce workflows directly from its own interface" — i.e. Anthropic's assistant becomes a front end to Salesforce's data, permissions and business logic without Salesforce's own UI. App Economy frames the arrangement as the resolution of the bear case rather than the confirmation of it: "the concern has been that Salesforce could become the data layer behind someone else's agent. Increasingly, that appears to be the strategy rather than the risk. Salesforce doesn't necessarily need employees clicking through Sales Cloud all day if Claude, ChatGPT, or another agent is still calling Salesforce behind the scenes." Named here as the counterparty, with no stance on the lab itself.
In short: Two opposite readings held at once. The revenue is extraordinary — the run rate "surged ahead of its IPO, rising to more than 65 billion in July of 2026 from 47 billion in May and 9 billion at the end of 2025," with "11.5 billion in preliminary revenue for its latest completed quarter, up from 787 million a year ago, while generating positive operating income and EBITDA," and it is "trying to raise more than SpaceX" with Morgan Stanley, Goldman, JPMorgan and now Citigroup on the listing, IPO expected October. The problem is the mix: per the Financial Times, its flagship frontier model Fable 5 "has plateaued at roughly 11% of total corporate spend on Anthropic tools" as enterprises "actively rout[e] work away from the costliest frontier models towards cheaper, good enough options, including lower tier models like Opus 5, as well as open weight alternatives" — which "challenges the assumption that technical capability leadership automatically translates into revenue." And the $65B ARR itself "was below various data sources that implied higher growth of 75 to 80 billion," which JPMorgan blames for Tuesday's tech-momentum reversal.
Anthropic is one of the two leading American AI labs and is expected to go public in October, with Morgan Stanley, Goldman Sachs, JPMorgan and now Citigroup running the listing. Its growth is remarkable: revenue on an annualised basis went from about $9 billion at the end of 2025 to $47 billion in May and more than $65 billion in July, with the business now producing positive operating profit.
The problem sits inside those numbers. Its most advanced and most expensive model, Fable 5, has stalled at roughly 11% of what corporate customers spend with Anthropic. Companies are deliberately routing routine work to the firm's cheaper models, or to free open-weight models they can run themselves, and saving the expensive one for jobs where a mistake would be costly. The uncomfortable implication, in the report's words, is that "technical capability leadership" does not automatically become revenue — being the best is not the same as being paid for being the best.
There is a second warning in the same week: the $65 billion figure was actually below what other data had implied (75-80 billion), and JPMorgan blames that shortfall for a sharp reversal in technology momentum stocks on Tuesday. Extraordinary growth, an unresolved pricing model, and a valuation about to be set in public.
Full passage: premium transcript (PDF).
In short: (Private, on file to go public.) The favourable half of the day's AI-lab comparison. Kate Rooney: "what is extra tough for OpenAI is that they are now being directly compared to Anthropic, as both companies are now on file to go public with the SEC. Anthropic, from what we have been hearing, just has the better margin profile. I reported earlier this week that Anthropic was profitable, at least on an EBITDA basis. It brought in $11½ billion in the quarter according to sources I've spoken to." On annualised revenue: "OpenAI is around $40 billion; Anthropic just ramped up to $65 billion." Her caveat: "we also don't know if these are apples to apples — they're not public companies, so the accounting can differ slightly. But there's going to be a lot of attention on these S-1s." No committee position.
Anthropic is still private but has filed to go public, and today it came out ahead in the first serious side-by-side comparison with OpenAI. Kate Rooney's reporting: Anthropic was profitable at least on an EBITDA basis (earnings before interest, tax, depreciation and amortisation — a rough measure of whether the core operation makes money before financing and accounting charges), and brought in about $11.5 billion in the quarter.
On annualised revenue, Anthropic has ramped to roughly $65 billion against OpenAI's roughly $40 billion. The judgement that matters to investors is the margin profile: Anthropic appears to be growing faster and losing less.
Two honest caveats: these are private companies, so the accounting may not be directly comparable, and the numbers come from sources rather than filings. Both will have to publish audited figures in their IPO prospectuses, which is where this comparison gets settled.
In short: The asset behind the earnings, named but not investable. Cited as the "AI entit[y]" whose valuation marks flow into Alphabet's reported profits — "companies such as GOOG reporting massive gains on their holdings in AI entities like Anthropic," gains that "represented more than 50% of [GOOG's] profits in the first two quarters of this year." Private, so no ticker and no research links; it appears here purely as the mechanism of the "immense, though unsustainable, earnings booster rocket" that Hay expects to mean-revert.
In short: Now the pace-setter for the whole capex race: "we're going to see Anthropic, which owns Claude, IPO in September-October now… they're looking at 50 billion run rate revenue, and they were on single-digit billions last year." It is also the accounting caveat under the earnings boom — the mark-to-market of Google's stake is why S&P Q2 growth reads 50% rather than ~30%.
20:40I mean, just look at, I think we're going to see Anthropic, which owns Claude, IPO in September-October now. And I think they're looking at 50 billion run rate revenue, and they were on the single-digit billions last year. So, the speed at which these companies are growing is just so tremendous that everyone wants to capture that pie.
In short: Trennert separates the layers exactly as Eisman does: "you could argue the hyperscalers have a real business and they're real moats around it, but it's a lot more capital intensive. The Anthropic's or Open AI's of the world, that's where it's most questionable… they're totally negative cash flow," and "the Chinese are competing with them." Then the systemic number: "something like 70% of the AI hyperscaler revenue is from just those two companies. So if this is the big if, if Open AI and Anthropic ever get in trouble, the ecosystem is in trouble." Eisman: "I think it's undeniable."
Anthropic is one of the two large private AI labs whose spending on cloud computing is the demand underneath the hyperscalers' AI revenue. The striking thing about this segment is that Trennert reaches the same map Eisman built independently three days earlier on the Weekly Wrap.
His version: the hyperscalers have real businesses with real moats, just capital-intensive ones. The labs are "where it's most questionable" — they burn cash, they have Chinese competitors, and roughly 70% of the hyperscalers' AI revenue comes from these two customers. That is a concentration hiding underneath the market's visible concentration: the S&P's top ten are 39% of the index, those companies' AI growth depends on five buyers, and two of those buyers fund themselves by raising capital.
The conclusion is conditional, not predictive: "if Open AI and Anthropic ever get in trouble, the ecosystem is in trouble." Eisman's response — "I think it's undeniable" — is agreement about the structure, not a forecast of the event.
24:15They're totally negative cash flow. Right. Right. Right. The Chinese are competing with them and I read this report by a firm where they basically said that something like 70% of the AI hyperscaler revenue is from just those two companies. So if this is the big if, if Open AI and Anthropic ever get in trouble, the ecosystem is in trouble.
In short: "Broadly understood that Anthropic looks financially a lot healthier than OpenAI. And that's one of the reasons why they're probably going a little bit more aggressively for an IPO now." Also one of the stakes whose mark-ups flow through the hyperscalers' "other income" line, flattering reported net income.
53:13So I only know what my colleagues have reported in the paper. I think it's broadly understood that Anthropic looks financially a lot healthier than OpenAI. And that's one of the reasons why they're probably going a little bit more aggressively for an IPO now. But I'd question with private companies how real sometimes revenue is — not like fending numbers but just, if you just look at the net income of some of the hyperscalers, the public companies now, look at how much is actually classified as other income,
In short: Same problem, stated as a business-model gap: "Anthropic and Open AI are also problematic because they don't have the breadth of revenue streams of Google and Microsoft." Paired with OpenAI as the one thing to monitor — "a key thing to monitor to determine a catalyst for a real sustained sell-off is the health of Anthropic and OpenAI."
Anthropic is the other pure model company — it builds Claude and, like OpenAI, rents its computing power from the hyperscalers rather than owning it.
Eisman's objection is a business-model one rather than a product one: "Anthropic and Open AI are also problematic because they don't have the breadth of revenue streams of Google and Microsoft." If the AI story cools, Google still sells search ads and Microsoft still sells Office; a pure model company has only the model. That single point of failure is what makes the layer fragile even if the technology is good.
Which is exactly why he ends up naming these two as the thing to watch rather than any listed stock: "a key thing to monitor to determine a catalyst for a real sustained sell-off is the health of Anthropic and OpenAI." They are private today, so the health is hard to see — but both are heading toward IPOs, and an S-1 would make the numbers public for the first time.
10:24But on the other hand, the companies he's less bullish on are the LLM providers, specifically the ones that exclusively offer LLMs. So, we're talking about Open AI and Anthropic. — Anthropic and Open AI are also problematic because they don't have the breadth of revenue streams of Google and Microsoft. Google and Microsoft have multiple revenue streams from established businesses which are very unlikely to simply disappear.
In short: Half of the named Achilles' heel. "There just don't seem to be any moats around LLMs. Users switch between models all the time," and cheap Chinese open-weight models are "good enough," so "a price war could break out." With OpenAI it accounts for "70% of hyperscaler AI revenue… and 25 to 35% of total cloud revenue" while it "loses billions and [is] reliant at this point on raising capital for [its] survival." "It feels like a bigger version of Situational Awareness, a huge one-way bet. What's the hedge? There is no hedge for the LLMs." Racing to IPO — which is also when "we will have some real data."
Anthropic is one of the two big private AI labs (with OpenAI) that build the chatbots and models businesses actually use. Eisman's point isn't about how good the models are — it's about how much of the AI economy quietly rests on these two companies' cheques.
Research firms estimate that roughly 70% of the hyperscalers' AI revenue, and 25-35% of their total cloud revenue, comes from Anthropic and OpenAI. Both lose billions a year and stay alive by raising new money. So the giant public companies everyone owns — Microsoft, Amazon, Google — have their fastest-growing business tied to two loss-making private customers.
Worse, he sees no "moat" (nothing that stops customers leaving): people switch between models constantly, and Chinese open-weight models are far cheaper and "good enough," so a price war is possible. He calls the whole setup "a huge one-way bet" with no way to hedge it. But he is careful not to turn that into a sell signal, because there is no data yet to measure the risk — which is why both labs racing to go public is, for him, the single most useful thing that could happen.
7:48The hyperscalers at least on the surface are not the problem. The LLM providers are the potential problem. specifically anthropic and open AI. There just don't seem to be any moes around LLMs. Users switch between models all the time. Perhaps more importantly, the Chinese LLMs are openweight models that are much cheaper than the LLMs provided by Anthropic and OpenAI and enterprises seem to be using the Chinese models more and more.
In short: The source of the mispricing, not a target — Anthropic's launch of Claude Co-work, a financial-analysis plugin, sent S&P Global down more than 25% peak-to-trough in February. "The concerns of Claude are real, but they are heavily overstated to a dramatic degree"; investors "jump out of them as soon as Claude or Anthropic release this one feature without even knowing how it's going to affect the company."
14:16And Bill Ackman watched this happen. He said, "In February of this year, the stock declined more than 25% from peak to trough following Anthropic's launch of Claude Co-work." That's the name of the plugin. Now, as a result, the stock's valuation declined from 25 times to 19 times earnings per share, the lowest valuation in the previous 5 years, and a bargain level for a company that is often cited as one of the world's highest quality businesses.
In short: The other frontier lab the trade depends on. Enterprises shocked by their token bills are moving to open-weight models at a tenth or a hundredth of the cost — which undermines the revenue assumed behind the $1.5T of unfunded commitments the finance industry is relying on.
Anthropic is the other half of the pair the AI trade rests on. The same logic applies: an enormous amount of data-center spending has been committed on the assumption that Anthropic and OpenAI will keep paying for compute at premium prices, yet enterprises shocked by their token bills are actively shopping for cheaper open-weight alternatives.
Hayes thinks the shift to cheaper models eventually improves returns for the companies using AI — but it does so by taking revenue away from the two frontier labs whose success the financing structure assumes. Two companies carrying the credit weight of the entire buildout is the single point of failure he flags.
31:07But a lot of this AI trade is predicated on the two frontier models succeeding, open AI and anthropic. So if there are models that you can get for onetenth or 1/100th of the cost, they've got 1.5 trillion in unfunded commitments related to those two companies. And then you look at other earnings from all the people holding the stock in anthropic and open AAI.
In short: (Private.) Kate Rooney: the IPO "could be as soon as October," the company has been meeting with bankers, and per the Wall Street Journal it has been offering assurances about its growth rate in those meetings while fielding investor questions about its growth and revenue run rate amid cheaper open-source systems coming out of China. The run rate "last topped $47 billion." No comment from Anthropic, though it has publicly defended the value of its more expensive models on a cost-per-task basis. It has filed confidentially; the S1 flips closer to the listing date. Sechan: "we tried to get some in the private markets — this is not one that we were able to participate in," and he cautions on the pattern: demand is enormous, "however the price traction has been you get a big buy-up and then they re-rate. Meta did that, SpaceX did that." No committee position.
Anthropic could go public as soon as October and is already meeting bankers. According to the Wall Street Journal, it has been reassuring prospective investors about its growth rate while fielding pointed questions about one specific risk: cheap open-source AI models coming out of China, which could undercut the prices Anthropic charges. Its revenue run rate last topped $47 billion. Its defence is that more capable — and more expensive — models deliver better value per task completed.
Rob Sechan tried and failed to buy shares privately, and his caution is about the shape of these listings rather than the company: enormous demand produces a big initial mark-up, and then the stock re-rates lower as reality sets in. "Meta did that, SpaceX did that."
In short: (Private.) Cited twice: as the partner behind the Zoom Communications move Terranova says he trusts, and as one of the imminent IPOs — "we are on the cusp of having further IPOs, whether it is from Anthropic, Databricks or OpenAI" — that underpin his refusal to buy the software complex ("significant disintermediation and disruption… short of the software apocalypse"). Firestone separately names Claude as daily-use AI driving chip demand. No committee stance.
In short: One half of the AI duopoly framing that carries the Meta thesis: "there's two companies that are leading the AI race, Anthropic and OpenAI. They're almost like the Apple and Google of AI." Zuckerberg's fear — and Carlson's stated bull case for Meta — is that the pair end up "duopolistic bottleneck gatekeepers to all AI for everyone," the phone-platform outcome repeated. Also named as the flagship benchmark Gemini is measured against. No investment call.
15:09What is possibly even more important than the phone and the App Stores? AI. Now, there's two companies that are leading the AI race, Anthropic and OpenAI. They're almost like the Apple and Google of AI. And Mark Zuckerberg is looking at this unfold. That these two companies, OpenAI and Anthropic, might end up in the exact same situation that Apple and Google ended up with the phone.
In short: Appears in three roles: a frontier lab whose pricing (~$10 per million input tokens) was blowing up enterprise budgets; the accuser that says open-weight models were built by distilling frontier models, citing ~4 million exchanged messages; and Tenable's partner, supplying the frontier model that reads a customer's code and assets for exposure management. The frontier layer is the one layer they think open weights are not bullish for.
2:36And so if you're a cloud provider for instance, enterprises are going to be using more AI given the fact that it's cheaper to use. They don't have to pay $10 per million input tokens for anthropic or chatbt. So demand is going to increase and if you're a GPU provider that's going to translate to higher GPU prices and the economics are going to increase for you.
In short: The mark-to-market vehicle behind the hyperscalers' reported earnings: Amazon's stake went from an $8B investment to "something north of 75 billion," booked as other income. She doesn't opine on the business — only that those are paper gains that reverse into "massive losses" if the AI valuations cool off.
Anthropic is the private AI lab (maker of Claude) that both Amazon and Google have invested in. Pomboy isn't making a judgement on the company itself — she uses it as the mechanism that explains a lot of this year's reported profit growth. Because the stake is private and its valuation keeps being re-set higher in new funding rounds, each re-set flows through its owners' income statements as profit.
The catch is symmetry. Adam's question — what happens to Amazon's and Google's earnings if the AI bubble deflates and those stakes get re-marked down? — gets her driest answer of the interview: "These are the pesky details that we're not supposed to think about or talk about… we're going to see that probably." The same lever works in both directions.
14:53And the latest valuation I saw of Amazon's position there was now something north of 75 billion. So, from 8 to 75 billion. And they have to book those mark-to-market gains as other earnings other income. So, it gets factored into the earnings numbers, even though it's not actually money that they earned.
In short: Named twice, on both sides of the AI infrastructure trade: as one of the cloud-compute customers already buying SpaceX's AI capacity (with Google), and as one of the customers lined up for AMD's Helios racks (with Microsoft and Oracle). A frontier lab renting capacity from a rocket company and committing to a non-NVIDIA rack is the compressed version of the issue's theme. (Referenced; not a stance call.)
In short: Asked directly whether the private valuations are too high, he reframes to structure and lands on commoditization: "if you're starting to split the workloads that way, I think over time, my belief is that the model layer becomes more of a commodity… yes, Anthropic and OpenAI become more commoditized over time, and the value accrues to more of the infrastructure players." Separately it is the actor in the Figma board episode he uses to argue enterprises must keep data proprietary. (Contrast 2026-JUL-31, where he called Anthropic the corporate-side winner — the company view is unchanged; what he is fading here is the layer.)
Anthropic is the private AI lab behind the Claude models, sold mainly to businesses. Asked whether its private valuation — and OpenAI's — is too high, Niles refuses to argue about the price and argues about the position instead.
His case: open-source models cut the cost of producing AI output by roughly 90%, and most of what companies actually ask for is easy. "You don't need a Ferrari to go to the corner store to get milk." Once cloud platforms automatically send the easy work to the cheap model and only the hard work to the frontier, the frontier labs lose the routine volume that would have made them enormously profitable — "Anthropic and OpenAI become more commoditized over time, and the value accrues to more of the infrastructure players."
Read this as a bet against the layer, not against the company. Four days earlier he named Anthropic the corporate-side winner of AI on hard financial evidence — profitability and an unprecedented revenue ramp — and nothing here contradicts that. What he is saying is that being the best model vendor is worth less than owning the platform that decides which model gets used.
Anthropic also appears in his governance warning: it sat on Figma's board, resigned, and shipped a competing product about a week later. His conclusion for corporate customers — keep your proprietary data proprietary — is a caution about handing your data to any model vendor, this one included.
2:33AND IF YOU GO AHEAD AND YOU'RE ON AZURE OR GOOGLE CLOUD PLATFORMS OR, YOU KNOW, WHATEVER, THOSE SERVICES WILL ROUTE WHATEVER YOU'RE TRYING TO DO TO THE BEST MODEL AVAILABLE FOR THE TASK. AND I THINK IN THAT SCENARIO, YES, ANTHROPIC AND OPENAI BECOME MORE COMMODITIZED OVER TIME, AND THE VALUE ACCRUES TO MORE OF THE INFRASTRUCTURE PLAYERS, WHICH INCLUDES THE CLOUD PLATFORMS AND SEMICONDUCTORS.
In short: The source of the quarter's biggest number that isn't operating income: Amazon's net profit included a $53.4B non-operating gain, primarily the valuation markup of its Anthropic stake, following May's $65B Series H at a $965B valuation — up from $380B in February, more than doubling in three months. A reminder that a large slice of reported Big Tech profit is now other people's funding rounds rather than trading performance, and the same strip-the-investment-gain test the newsletter runs on Microsoft and Tesla. (Referenced; not a stance call.)
Anthropic is the private AI lab behind Claude, and Amazon is one of its big backers. In May it raised $65 billion of new money at a valuation of $965 billion — more than double February's $380 billion. Because Amazon's stake gets re-valued whenever a new round prices it, that single event produced a $53.4 billion gain in Amazon's quarterly profit.
Worth understanding for what it is: none of that is money Amazon earned from customers, and none of it arrived in cash. It's an accounting mark on a private shareholding, driven by what other investors were willing to pay. It flatters the headline profit and would reverse just as quickly if a later round priced lower. Referenced in passing, not a stance call.
In short: The corporate-side winner, and the only one of the two private model labs he wants: "Anthropic interests me because I've been saying this for a while. You only have a certain number of winners… in corporate, you have Anthropic, which… got to profitability in Q2 and their revenues are ramping like nothing we've ever seen in history for a company of that size."
Anthropic is the private AI lab behind the Claude models, sold mostly to businesses rather than consumers. Asked whether either of the big private labs would interest him as a public stock, Niles takes this one — "Anthropic interests me because I've been saying this for a while."
What makes it the corporate winner in his framework is evidence rather than narrative, and he cites two specific facts. It reached profitability in the second quarter — meaning it takes in more than it spends, which for a company burning cash on training runs is unusual and hard to fake. And its revenue is "ramping like nothing we've ever seen in history for a company of that size" — growth without precedent at that scale.
The structural point behind the pick: if a category only supports a couple of winners, the split is usually by customer type. Google takes the consumer slot; Anthropic takes the business/enterprise slot. Two winners, two distinct customer bases, no direct collision.
0:47If you think about it, you go, well, who's the winner in search? That's just Google. What about in e-commerce? That's just Amazon. Are you going to have five different guys win in AI? No. I think Google wins in consumer. They have the complete stack. I think they win in AI overall. But then in corporate, you have Anthropic, which as I said earlier, they got to profitability in Q2 and their revenues are ramping like nothing we've ever seen in history for a company of that size.
In short: Now the named catalyst to monitor. As an LLM provider it is on the wrong side of his new split: "here the debate has really shifted because there just don't seem to be any moats or at best the moats are shallow. Enterprises are switching between models and using cheaper open-source Chinese models in order to control costs… Anthropic and OpenAI are also problematic because they don't have the breadth of revenue streams of Google and Microsoft." Hence: "a key thing to monitor to determine a catalyst for a real sustained selloff is the health of Anthropic and OpenAI." Plus the policy charge: the job-destruction story "could be propaganda propagated by Anthropic and OpenAI so that the federal government will step in and regulate AI to the benefit of Anthropic and OpenAI. Demanding regulation based on a false narrative would be a very disturbing way to create moats."
Anthropic makes one of the two leading closed-source AI models. This episode gives it a specific and unusual role: not just a company he's sceptical of, but the thing he would watch to time a market-wide sell-off.
The problem is structural. Closed models can't stop customers leaving, because "enterprises are switching between models and using cheaper open-source Chinese models in order to control costs" — so there is no moat, "or at best the moats are shallow." And unlike Google or Microsoft, Anthropic has no other business to fall back on: "they don't have the breadth of revenue streams."
Why that becomes everyone's problem: the data-centre companies' order books are largely made up of promises from these two firms. "If the lack of moats begins to cause them problems, then the entire AI ecosystem could go through a correction phase."
He also repeats, in his own voice, the accusation Gil Luria made four days earlier — that the widely-promoted story of AI destroying jobs "could be propaganda propagated by Anthropic and OpenAI so that the federal government will step in and regulate AI to the benefit of Anthropic and OpenAI. Demanding regulation based on a false narrative would be a very disturbing way to create moats." His own view is that AI produces "job dislocation, but net job creation."
7:28Google and Microsoft have multiple revenue streams from established businesses which are very unlikely to simply disappear. They also have hyperscaler businesses to balance their vulnerability. But their LLM businesses are also questionable. A key thing to monitor to determine a catalyst for a real sustained selloff is the health of anthropic and open AI.
In short: "The clearest holdout." Anthropic did not sign the open-weight letter and is absent from NVIDIA's alliance, arguing it does not want to ban open models but supports stricter chip controls, action against industrial-scale distillation, and mandatory safety testing for all capable systems — the sharpest counter-position in the fight. It is also a Foundry-listed model provider (Claude wrapped by Microsoft's common APIs and governance), and the source of the $3.2B investment gain Microsoft left inside adjusted EPS while excluding a $480M OpenAI gain — the quarter's quality-of-earnings flag. (Referenced; not a stance call.)
Anthropic (the maker of Claude) is the odd one out in this fight. It refused to sign the industry letter defending downloadable AI models and stayed out of Nvidia's new security alliance. Its position isn't that open models should be banned — it wants tighter controls on advanced chips, enforcement against "industrial-scale distillation" (rivals cheaply cloning a model by training on its answers), and compulsory safety testing for any sufficiently capable system.
It shows up in Microsoft's numbers too. Microsoft owns a stake in Anthropic, and a $3.2 billion paper gain on it was left inside Microsoft's adjusted earnings while a smaller $480 million gain on its OpenAI stake was stripped out — meaning some of Microsoft's headline profit came from a share-price mark, not from selling software. Anthropic's models are also offered on Microsoft's own marketplace, which is the whole "we host everyone" strategy in miniature. Referenced, not a stance call.
In short: "One of the bottlenecks is that Anthropic and OpenAI are responsible for a lot of the spending that people are doing. And Anthropic and OpenAI have models that are now much more expensive than the competition. And I think that is a potential bottleneck." The clip opens mid-argument on why "the closed source models are just much more expensive."
Anthropic is one of the two private labs (with OpenAI) whose model-training and inference budgets drive much of the AI spending everyone else is now underwriting. Eisman's concern is price: their models are "closed source" — you rent access rather than run them yourself — and closed models are "just much more expensive" than the open-weight competition.
That is what he calls "a potential bottleneck." If the two labs generating the demand also charge the most, customers have an obvious incentive to move to cheaper alternatives — and the spending case that supports the whole AI trade weakens from the demand side rather than from any technical failure.
4:43But look, again, I think one of the bottlenecks is that Anthropic and OpenAI are responsible for a lot of the spending that people are doing. And Anthropic and OpenAI have models that are now much more expensive than the competition. And I think that is a potential bottleneck. — A much better cook because of AI. — And you are.
In short: The moat question lands here. Eisman's wrap: "unclear still to me how much of a moat Anthropic or OpenAI have." Luria's two charges: the model-dependency risk — "the government told Anthropic to rein in Fable and they didn't… if you were a business that built your business directly on top of a Fable model, you're out of business"; and the policy play — "why are Sam and Dario scaring us? Because they're pulling the ladder. What they want is friendly regulation. They want to shut out open source, shut out Chinese companies… so those two are the only winners." Credit where due: combined OpenAI+Anthropic run-rate is above $75B, "real economic activity."
Anthropic builds one of the two leading closed-source AI models — closed meaning customers can only use it through Anthropic, which controls the code and the settings. That control is the business model and, on this episode's argument, also the vulnerability.
Three charges. Pricing: Chinese open-source models cost a fraction as much, so a closed provider is charging "five to seven times more" for something claimed to be equivalent — "I'd be petrified," says Eisman. Dependency: when regulators ordered a change to one of its models, any customer that had built directly on top of it was left stranded overnight, which is now an argument for never depending on a single model. Politics: Luria accuses Sam Altman and Dario Amodei of "pulling the ladder" — pushing a scary job-loss narrative in order to win regulation that keeps out open source and Chinese rivals, "so those two are the only winners."
The counterweight, which Luria supplies himself: Anthropic and OpenAI together are already running above $75 billion of revenue from nothing two years ago. The demand is real; the durability of the margin is what's in doubt. Eisman's own conclusion is simply "unclear still to me how much of a moat Anthropic or OpenAI have."
48:58That's how it's worked forever and that's how it's going to work with AI. Now, why are Sam and Dario scaring us? Because they're pulling the ladder. What they want is friendly regulation. They want to shut out open source. They want to shut out Chinese companies. They want to shut out everybody else. So they've developed this narrative that oh the jobs are going to get lost and this is so dangerous that you have to be careful and they've developed this narrative because they want the government to put in
In short: Cited as the mechanism inside Alphabet's earnings and backlog rather than as a view on the company: the ~34% earnings contribution is "largely coming from Google's stake" in Anthropic (or OpenAI), and on the hyperscaler backlog chart "anthropic is the big part of their there" — so "there's some questions about the health of those backlogs."
40:20But Fred said, well, look, a lot of these earnings are one-time gains that are coming from the AI bubble, right? So that 34% is largely coming from Google's stake in the I can't remember whether it's anthropic or open AAI. it's probably invested for all I know. — but also he talked about how these earnings are also sort of artificially rosy because there's a huge schedule of depreciation costs coming from these investments that has really yet to hit the P&L.
In short: Two roles this week: one of the two stakes (with SpaceX) whose markup produced 87% of Alphabet's record net income, and the enterprise-AI winner squeezing OpenAI — Niles is negative on OpenAI precisely because it is "caught between Google in consumer AI and Anthropic in enterprise." Also signed a deal with AMD for tens of billions of dollars of AI servers over the weekend.
Full passage: premium transcript (PDF).
In short: Reference, not a stance: one of the private AI labs whose stake Google marks up to book "Other Income" — the same "Great Circularity" markup mechanism Paulo has detailed, now the driver of 87% of Google's 2Q "income."
In short: Named next in the same IPO pipeline ("here comes Anthropic") that keeps Canadian bank capital-markets divisions "very lucrative." Reference only.
19:18Can it get any better than you get to take a multi-t trillion dollar IPO in SpaceX? Can it get better than that? Well, we'll see. Here comes Anthropic and maybe Ben and I. We'll see. — This week, David Rosenberg said he always bubble spotting said Canadian banks are in a bubble. Time to take profits.
In short: Grouped with SpaceX/OpenAI at 20–70× revenue — a pass on the same valuation logic; Oxbow doesn't short and doesn't ride momentum, so it simply waits.
Anthropic (the Claude AI models) is the third mega private AI name. Same verdict as OpenAI: at 20–70 times revenue, the best case is fully priced, leaving little upside.
His approach to these extreme valuations is neither to chase them nor to bet against them, but to stay away and wait for a price that offers a real margin of safety.
22:12And right now valuations like what we see in the private marketplace for these three big companies that have everyone's attention, if they're trading at 20 to 70 times revenue when by comparison Google went public at eight and a half times revenue. So that was a more reasonable, great IPO to try to invest in during the first year that it became public.
In short: Counterparty on one of xAI's short-dated "spot" compute deals (a billion-ish, ~3-month outs); part of the researcher brain-drain churn between the labs.
30:48But so on the eve of the IPO, we all know that xAI was a money pit, right? And talk about overbuilding for their own needs and then they become essentially renting out the compute. — were bought for SpaceX stock valued at 250 billion in February. — Yeah. And so now they do these deals, one with Anthropic and one with Google and — two spot deals with tremendous outs.
In short: Grouped with SpaceX/OpenAI at 20–70× revenue — a pass. Also noted hyperscalers are shifting toward cheaper Chinese models that do ~90% of tasks "at a fraction of what Anthropic or OpenAI might charge."
Anthropic (maker of the Claude AI models) is the third of the mega private AI names. Same verdict as OpenAI and SpaceX: at 20–70 times revenue, it's a pass for a value-focused buyer.
He adds an interesting demand risk: even big customers like Microsoft are starting to route work to cheaper Chinese AI models that do about 90% of the job "at a fraction of what Anthropic or OpenAI might charge" — pressure on the frontier labs' economics that the valuations don't reflect.
22:09So that's boosting their earnings artificially, right? Those aren't necessarily operating earnings. But then also you've got all this CapEx spending that's going on right now. And the depreciation cost of all this CapEx is coming in the future. It's not right now, right? So when you really factor in what the ongoing depreciation costs are going to be, yeah, profit margins are actually going to be coming down a fair amount because of that.
In short: (Private.) News alert (Kate Rooney): Anthropic is starting to line up investor meetings ahead of its IPO in the next couple of weeks, with a listing possibly "as soon as October" — a potential trillion-dollar IPO and "the first of these pure-play AI labs" to price (it filed confidentially with the SEC ~six weeks ago). Also flagged by Sechan: Anthropic and OpenAI "at some point are not going to be able to raise the capital" privately — "they need the IPO market." No committee stance.
In short: Like OpenAI: burning cash, wants an IPO, being forced to slash prices and offer subsidies as Chinese open-source models undercut it. Half the hyperscalers' backlogs sit with these two frontier labs — if they hit trouble it reflects straight back into the hyperscalers.
Anthropic (maker of Claude) is in the same spot as OpenAI: burning cash, wanting an IPO to raise more, and being forced to slash prices and hand out subsidies as cheaper Chinese open-source models eat its lunch. Because these two frontier labs sit behind roughly half of the hyperscalers' backlog orders, their trouble is the hyperscalers' trouble — a single point of failure the market isn't pricing.
19:07they require lots of cash, yet the hyperscalers' backlogs, they represent 50% of the hyperscalers' backlogs, if they end up with a lot of trouble, if they go bankrupt. And I believe some of these companies will end up going bankrupt. I mean, Oracle has, their debt is two and a half times their sales from last year.
In short: Placed with OpenAI on the "premium and defensible" end of the framework — both want the model layer highly differentiated so they can charge directly for APIs/agents/subscriptions ("the model IS the monetization layer"). Carlson believes this premium motivation loses to the commoditizers over time as Meta/Google throw "endless compute" at closing the gap.
17:00When we get to OpenAI and Anthropic, their motivations are entirely different. They both want their model layer to be premium and defensible to be highly unique. They want AI costs to go down, but they also want them to be highly differentiated so that they can continue to charge directly for their APIs so that people have a reason to pick OpenAI over Anthropic.
In short: Same grouping — named in the IPO-pipeline list and in the monthly-loss sentence. No analysis of the business.
In short: The article's center of gravity. Ellenbogen: the LLM leader since Claude Opus 4.5 (late Nov) — revenue went from ~$9B to >$50B in six months ("incredible"); has filed confidentially for an IPO that will let investors own the LLM layer directly and make the market "much more discerning" about AI cash-flow duration. Giroux: proof the world's best LLM runs on low-cost Google/Amazon ASICs — the end of Nvidia's monopoly. Jain's caution: with OpenAI, its spending underpins a big share of AI capex; if closed-source revenue growth falters vs open-source (now ~2/3 of tokens), the capital-raising chain is at risk.
The private AI company is the article's gravitational center. Ellenbogen: since Claude Opus 4.5 shipped in late November, Anthropic's revenue exploded from ~$9 billion to over $50 billion in six months, and it has confidentially filed for an IPO — which would finally let investors own the AI-model layer directly instead of through proxies, and force the market to judge which AI companies have durable cash flows. Giroux adds a twist: Anthropic built the world's best model on cheap Google and Amazon chips, proving Nvidia is optional. Jain supplies the caution: Anthropic and OpenAI's spending props up much of the AI buildout, and open-source models now serve two-thirds of all usage — if the money engines sputter, the whole capex chain feels it.
In short: The no-moats exhibit — "one day Anthropic is on top and the next day it's Gemini and the next day it's someone else." Spending trillions on a business "that has no moats is a recipe for a price war."
12:07As to moes, one day Anthropic is on top and the next day it's Gemini and the next day it's someone else. Spending trillions on a business that has no moes is a recipe for a price war, not for high levels of returns on massive investments on capex. And it looks like companies are starting to experiment with Chinese AI because it's much cheaper.
In short: (Private.) Kate Rooney breaks the news: former Fed Chair Ben Bernanke appointed to Anthropic's independent Long-Term Benefit Trust (which can appoint board members / advise on risk), to help the public-benefit corp think through AI's effect on jobs, markets and the economy.
In short: Part of the Meta bull case — SemiAnalysis reports Meta "in final talks with Anthropic to get access to a private instance of Claude," and Carlson expects "a $10 billion Anthropic deal to kick off the flywheel" for Meta's NeoCloud. Separately, CEO Dario Amodei is walking back last year's "half of entry-level jobs" AI-doom warning.
14:11So Meta believes that they can improve their ad recommendation system by over 10x. Third, they say that we believe Meta is in final talks with Anthropic to get access to a private instance of Claude. This would be akin to Bedrock, Foundry or Vertex from the hyperscalers. There are multiple use cases for Meta ranging from internal usage to building the premier sales and marketing SaaS powered by Frontier AI agents.
In short: The most telling SCA example — a memory/storage supply agreement paired with Micron's equity investment in the AI lab's Series H round (a memory maker buying equity in the very demand it's locking down).
Anthropic (the AI lab behind Claude) shows up here as the clearest example of how strange and tight the memory market has become. Micron didn't just sign Anthropic to a long-term contract to buy memory — it also invested in Anthropic, taking an equity stake in the lab's latest funding round (its "Series H").
The article's point: a chip maker buying a piece of one of its own biggest future customers is a memory supplier putting money into the very demand it's locking down — a sign of how strategic, and how scarce, AI memory has become. It's the most vivid illustration of Micron turning panic over supply into durable, contracted relationships.
In short: (Private.) Filed confidentially; unprofitable — losing ~$2.50 for every $1 earned (PitchBook). Named as the AI vendor Microsoft's Copilot pivoted to (then away from), and as "arrogant" alongside OpenAI in Weiss's pricing-pressure point.
In short: Passing reference — Claude Code named as a leading rival in the brutally competitive AI-coding field Cursor must defend against.
In short: Claude 4.8 "priced into a corner": JPMorgan blocked Claude in Hong Kong on cost/compliance and Microsoft is swapping to cheaper/DeepSeek models. With DB's "Mythos" model circulating and trillion-dollar IPOs being prepped confidentially, the long-term enterprise moat looks "far less secure than originally hyped."
Full passage: premium transcript (PDF).
In short: Watch-list, not impulse-buy — filed confidentially; may become a defining franchise, but owning it on day one at peak-euphoria valuations before a single public quarter is a weak case. Let it report, let insiders sell, let expectations move from story to numbers.
Anthropic (maker of the Claude AI models) has filed confidentially to go public and could be one of the most valuable IPOs in history. The article's advice is the same discipline it applies to every hot new issue: put it on a watch list rather than buying on day one.
The reason is statistical. Over 40 years, IPOs as a group underperform — and the worst odds belong to companies that are large and not yet profitable, which describes Anthropic exactly. Day-one buyers pay a price set at peak excitement, before the company has reported a single public quarter and before insiders are allowed to sell. Waiting lets the hype drain, lets the lock-ups expire, and lets the story be replaced by actual numbers — a much safer entry if it really is a future giant.
In short: The US told Anthropic its latest innovation can't be exported — software export controls that "basically shut it down." Raises his AI-nationalization / security-war risk: if a DeepSeek-style moment recurs (China distilling Western models into open source), it would be "a massive credit crisis" for the $4T-capex names.
19:08told Anthropic that their latest innovation, you can't export it, right? So they're not controlling Claude. So they're not controlling products anymore. They're saying, I'm controlling your software. There's export controls on it. And they basically had to shut it down. And I wonder if there's a risk that AI becomes so powerful, like mythos, for example, they're not letting it out. It's very select.
In short: A US export-control directive suspended all access to its top models — even to its own foreign-national employees — after a jailbreak method surfaced (Amazon, an OpenAI investor, reportedly told the government). "Very heavy-handed… hurts the future of Anthropic." The lesson: "it's not a great business strategy to go to war with the United States government." (IPO due in the fall.)
Anthropic is a leading AI lab (maker of advanced models), and Eisman's story is a cautionary tale about picking fights with Washington. After Anthropic tried to restrict the Defense Department's use of its models, the government classified it as a "supply-chain risk," and then — once a method to "jailbreak" (bypass the safety limits on) its top models surfaced — issued an export-control directive cutting off access to those models, even for Anthropic's own foreign-national staff. That effectively shuts the flagship products down.
His blunt lesson: "it's not a great business strategy to go to war with the United States government." He also notes the tip reportedly came from Amazon — which invests in rival OpenAI — calling it "a sixth-grade cafeteria level of snitching." For an AI lab heading toward an IPO this fall, having its best models frozen by Washington is a serious blow.
4:17There was some really bad news for Anthropic over the weekend. If you will recall a few months ago, Anthropic tried to limit the usage of its models by the Department of Defense. The DoD was not amused and classified Anthropic as a supply chain risk.
In short: Named alongside OpenAI / Gemini as an AI-model vendor Meta won't depend on.
22:59Zuckerberg describes it as a company that builds new applications, new experiences, new features. To be able to build all of this stuff, you need to have the best tools possible. If Meta is reliant on renting those tools from Gemini, renting them from OpenAI or Anthropic, then now they're no longer full stack. They're beholden to someone else.
In short: The cautionary tale — private valuation $965B (past OpenAI's), the most valuable standalone AI lab on paper, yet a single export-control directive forced it to disable Fable 5 and Mythos 5 worldwide (couldn't screen users by nationality in real time). For a pure-play lab the frontier model IS the business — no cushion. Has filed confidentially to IPO.
Anthropic is a private AI lab (maker of the Claude models) whose paper valuation just rocketed to $965 billion — on paper the most valuable standalone AI company in the world. Then the article's whole point landed: the US government issued an export-control order restricting who could use its two most powerful models. Because Anthropic can't reliably check every user's nationality in real time, it had to switch those models off for everyone, everywhere.
That's the danger of being a "pure-play" lab: the AI model basically is the entire business, so when it gets switched off there's no Search engine or app store or cloud division still earning money to cushion the blow. The same models, ironically, are valuable enough that elite partners (Apple, Google, Microsoft, CrowdStrike) test them to find security flaws — genuinely dual-use technology. The takeaway for anyone eyeing Anthropic's coming IPO: its biggest asset can be turned off by a phone call from Washington.
In short: Named with OpenAI as the next IPOs that "are going to need dollars to be found elsewhere" — part of the equity-supply wall the melt-up market must absorb after a 20-year shrinking-float era.
7:24We've never seen so much equity ownership. Feels like kind of like a housing crisis again when your barber's talking to you about Nvidia call options. There's a problem generally. So, but it's not a hollow market. Earnings are real. And I think we can bring it all back to this whole AI thing. Is it real? And it is.
In short: Next in the queue behind SpaceX ("you can't have 45% of the index that's already in technology and then bring in SpaceX, Anthropic, OpenAI"). The host adds that his maxed-out subscription is subsidized ~20:1 — an unresolved unit-economics problem before any listing.
Anthropic is next in the IPO queue behind SpaceX, and that's the problem: "you can't have 45% of the index that's already in technology and then bring in SpaceX, Anthropic, OpenAI." Every one of these arrives fully valued after years of private-market gains, and passive funds must buy it regardless. The host adds a business-model caveat from the user's side — a maxed-out subscription is being subsidized at roughly 20 to 1, so the economics have to change before the public gets the bill.
11:32How does all this play out, do you think? — Well, the good news is there's a lot of millionaires that have been created and I love the stories of the welder that's now a millionaire at SpaceX. I mean, there's wonderful stories. But it's similar to Facebook when it came public.
In short: The borrower at the center of "Big Sky": a $35B Apollo/Blackstone private-credit SPV (A1 $6B at +1% Treasuries; A2 $24B at 5.75%; junior $4.5B at 8.5%) — and lenders were given no access to its financials ahead of the IPO. Peak-euphoria circular financing.
Full passage: premium transcript (PDF).
In short: Named (its Claude models) as one of DocuSign's new IAM integrations powering the Agent Studio / agentic-review features.
In short: "Your grandma using Claude is just going to rack up a Duolingo competitor is totally insane" — generic LLMs need fine-tuning on proprietary data, so a general chatbot can't replicate Duolingo's data moat.
17:44Proprietary data is the most important moat ever. Maybe the most important moat today. And having that proprietary data allows you to fine-tune a specific application or procedure to a specific purpose. You can only do that with proprietary data. So, the idea that your grandma using Claude is just going to rack up a Duolingo competitor is totally insane.
In short: More overvalued new supply "coming to market" after SpaceX — "all overvalued relative to historical norm." He doesn't buy overvalued assets.
14:32Again, just more of the strip mining and stealing and taking advantage of the average person. I'm totally against it, but nobody cares what I think. So, hey, if you did well, Allah be with you. Take the money and run, I guess. But this isn't investing, make no mistake. So, this is going to continue because we have other things coming to market too. Anthropic at some point, OpenAI, other things, these things are all overvalued relative to historical norm. And again, I don't
In short: Grouped with OpenAI as a private mega-cap that doesn't show its books — opacity is "always the case when you have private markets… they attract people who like to obfuscate." Part of the IPO-wave evidence and the private-credit over-concentration concern.
Anthropic is grouped with OpenAI as a private AI giant that won't open its books. Gundlach's point is structural: private markets "attract people who like to obfuscate," because being private means nobody can really see what's happening. He ties this to a brewing private-credit problem — lenders quietly over-concentrated in AI/software they've mislabeled — and to the IPO wave that may be draining liquidity out of the broader stock market.
42:55So they're obfuscating which is always the case when you have private markets. Private markets they attract people who like to obfuscate because by being private people can't really look at what's happening. Two companies OpenAI and Anthropic they do not show their books to anybody. One of them wanted to get a loan recently, a private placement loan to get some capital and a potentially willing lender said, "Let me take a look at your numbers.
In short: Filed a confidential S1 (~$100B expected). Cited as evidence AI is commoditized — "one week Gemini is on top and the next it's Anthropic," no differentiation, no moats. Part of the ~$360B equity-supply wave.
13:20But later the same day, Thursday, he cancelled the bombing and stated that a deal is close at hand. We shall see. But the market rallied on that news. Moving on. Last week, Anthropic filed a confidential S1 for an IPO. I'm guessing the size of that offering will be around 100 billion. This week, OpenAI filed its own confidential S1.
In short: Next up in the IPO wave — an estimated ~$60B raise at close to a $1T valuation (91% odds it announces this year per prediction markets); part of the ~$280B issuance competing for the same investor cash as SpaceX.
52:06at that time it had very liquid you it had relatively liquid options, had a market cap of well over 7 billion against net assets near 650 million. So it was basically trading at 11 times NAV. And we did an adjusted NAV assuming, SpaceX would IPO at 1.75 trillion., Anthropic would IPO at close to a trillion.
In short: Next in the IPO queue after SpaceX — part of the ~$200–250B of immediate raises (and ~$3T of eventual unlocks) the market must absorb by selling everything else.
Anthropic (the maker of Claude) is next in the IPO queue. The issue isn't the company — it's the arithmetic: with SpaceX, Anthropic and OpenAI together, roughly $200–250 billion must be raised immediately, and about $3 trillion of insider shares unlock within 6–12 months after. Every dollar that buys these IPOs is a dollar sold out of something else in the market.
8:08The actual raise which is the money that we need to come up with someplace to pay for the shares being offered is 80 billion. Okay, add that 80 billion. Google's — and add to that Anthropic is coming up. We're going to have OpenAI coming up. There's about 200, 250 billion of immediate raise. But here's the thing that I'm actually focusing more on, Larry, is 6 to 12 months after that, all the insiders and the VCs and the early investors in those companies, it's not 200 billion.
In short: Part of the planned supply wall but may never get its chance if SPCX shuts the ECM window; lumped into the Great Circularity — the AI complex's "cashflow" is VC-sourced cash relaundered via capital raises.
Anthropic is the AI lab behind Claude, expected to IPO later this year. Paulo sees it stuck in line behind SpaceX: if the SpaceX deal exhausts investors' appetite for new shares, Anthropic may never get its window. He also folds it into what he calls the Great Circularity — the observation that much of the AI industry's apparent "cash flow" is really venture-capital money raised by the AI labs, spent on hardware, and recycled back around, rather than profits earned from outside customers. That looks fine while fresh money keeps arriving; it breaks when the venture investors stop buying and start cashing out.
In short: Cited to illustrate that "no one has enough capital": "Everyone needs help funding, which is why they're giving letter of credit to Anthropic" — the AI product will be used far more, but at far lower prices.
3:20And I think that's the real pricing model issue that this is having is no one has enough capital. Google's proving that. Everyone needs help funding, which is why they're giving letter of credit to Anthropic. And we're going to use a lot more of this product. It's just going to be at far lower prices, and the chips are probably going to have to do that, too.
In short: His highest-conviction position (invested Aug 2025 at the ~$180B round). Coding is "the true unlock" — ~20M coders × ~$20-30k/yr ≈ a ~$0.5T market from coding alone; critical IP, an enterprise brand (CIOs "say Claude first"), escape velocity/scale and recursive self-improvement.
Anthropic is the private company behind the Claude AI models — Sacerdote's single biggest, highest-conviction bet, made in August 2025 at a roughly $180 billion valuation. A "foundational model" is the core AI engine that everything else is built on; companies pay per "token" (the chunks of text the model reads and writes), so heavy users rack up big bills.
His thesis: coding is the "true unlock." Engineers using Claude to write software were burning ~$100/day in tokens — about $20-30k a year each — and with ~20 million coders worldwide that's a ~$500 billion market from coding alone. Anthropic has stayed ahead in coding, has a "moat" (a durable edge) from critical know-how, and an enterprise brand so strong that CIOs "say Claude first." It has hit "escape velocity" — enough scale and fundraising muscle to keep pulling away — and is even feeding its own coding tool back into improving its models, so progress is accelerating.
6:31And mind you, that was on 7 8 9 month old technology. We could see just on the coding market alone that Anthropic had a tremendous opportunity ahead of it. So I think at the time, this is pretty funny, we wrote in our letter, you know, we made the investment um at the 180 valuation. And we said, and I think they were hoping to get to a nine billion — one to nine. Yeah.
In short: Private — named in the same IPO-pipeline breath (auto-transcript "Tropic"); cited only as a venue/trading point for a rules-based index, not on fundamentals.
59:49Now this is a competitor in theory to the S&P 500. But instead of there being this committee that governs the S&P 500, which is shown to be fairly arbitrary at times, Bloomberg is fully rules-based. — Mhm. — And one of the inclusion criteria is that it allows these IPOs to be admitted much more quickly. And so investment merits of investing in SpaceX and Tropic, um, OpenAI, any of this pipeline of IPOs aside, exchanges are trading venues.
In short: Named with SpaceX/OpenAI as a coming mega-IPO sucking liquidity from other names — which he hopes pressures resource/financial markets so he can buy cheaper.
37:32I wonder what these IPOs — SpaceX, Anthropic, OpenAI — will do to liquidity, sucking capital from other names. The beginning of the end? I certainly hope so. There are a lot of names in conventional financial services and natural resources I'd like to own much more of — I'd be delighted to see Exxon Mobil, Agnico Eagle or Franco-Nevada fall by half. The only way I've found to become profoundly materially richer is to buy undervalued assets and wait until they return to value. So my hope is that higher oil prices and IPO-driven illiquidity lead to materially worse equity markets, particularly in financial services and natural resources — the markets I know best.
In short: SBF illegally invested $500M for 7.84% of Anthropic — now likely worth ~$70–100B, "one of the greatest tech investments in history." Right bet, wrong money, wrong time (pre crypto-crash).
23:28And usually that ends up with jail time. But I also consider Sam Bankman Freed to be one of the unluckiest people in prison today. First of all, one of the investments that Sam Bankman Freed made with that money that he was illegally trading, he invested $500 million into Anthropic, buying $7.84% of the company.
In short: His monetization proof-point: annualized revenue run-rate has "gone vertical" — ~$1B a year ago, $9B in December, $14B in January, $30B, then $44B a few weeks ago. Agentic coding (Claude) is the real use case that's "reached takeoff velocity."
Anthropic (the private company behind the Claude AI models) is Rasgon's favorite evidence that AI spending is producing real revenue, not just hype. Its annualized revenue run-rate has "gone vertical": roughly $1 billion a year ago, $9 billion in December, $14 billion in January, $30 billion, then $44 billion a few weeks before this episode.
The driver is "agentic" coding — AI that writes software, spinning up dozens of sub-agents that each handle a piece of the work. Rasgon calls it the first AI use case to reach "takeoff velocity": real demand, and a real willingness to pay. It's also why he dismisses the "no return on AI" doomsday — companies are demonstrably paying for tokens, in some cases more than they pay for the staff they replaced.
53:14And maybe you argue, well, they're using too much, but I think they are monetizing. I mean, just as one point example, you can look at Enthropic. Enthropic does periodically release their annual like annualized revenue run rate. — They've gone vertical, right? So, I mean, the last number they gave, which was a few weeks ago, they were 40$44 billion annualized revenue.
In short: Cited as a genuine AI "player" leaning on the compute build-out — referenced (not rated) in arguing SpaceX's AI bet faces serious competition.
1:34IT WAS JUST IN STARLINK AND SPACEX. — THAT'S SO HARD. IT'S THE AI. — IT'S THE AI THAT THAT IS INCREDIBLY CAPITAL INTENSIVE AND GROK, WITH ALL DUE RESPECT TO ELON MUSK, IS NOT A WORLD CLASS AI COMPANY. AND, YOU KNOW, I DON'T THINK ANYBODY SPEAKS OF GROK AS AS AT THE LEADING EDGE. — NO. BUT THE THE COMPUTE POWER THAT THEY ARE BUILDING HAS EVEN PLAYERS LIKE ANTHROPIC.
In short: A $35B Apollo/Blackstone debt financing funds Anthropic's purchase of Google TPU chips Broadcom helped develop (Broadcom backstopping the largest portions) — part of the AI XPU Platform targeting 20+ GW of LLM compute through 2028. Also runs Rubrik's Project Glasswing (early access to Mythos research) and powers Veeva/GitLab AI workflows.
In short: The catalyst — its "Mythos" model, judged too dangerous to release widely, sent enterprises scrambling to reassess defenses, igniting the AI-security demand wave both PANW and CRWD reported into. Private; referenced.
In short: Named with SpaceX as the mega-IPOs that will "suck in trillions" this summer.
4:42If you look at the whole history, we're going to go down that rabbit hole a little bit. The whole history of the US stock market going all the way back to 1792, there's only been 1.5 trillion of IPOs cumulatively. And we're going to do 4.6 six actually I think is the number this summer again it's not all going to hit the market at once because as you're aware there's lockups so the insiders can't sell all at once although they're making it easier for them to they're going to be able to sell it quicker than they normally can or their lockups will
In short: Frontier-model capabilities "plateauing"; competition (Cursor/Copilot/Codex) closing in on Claude Code; Microsoft reportedly canceled most Claude Code licenses — Q1/Q2 hyper-growth may not be sustainable, the biggest problem heading into its IPO.
Anthropic is the private AI company behind Claude, heading toward a stock-market debut (IPO) later this year. Woo sees two problems. First, the genuinely useful, widely accessible abilities of top AI models are "plateauing" — improving in small steps rather than leaps — which makes it harder to keep charging more.
Second, competition is catching up to its Claude Code product, and Microsoft reportedly dropped most of its licenses. Together with customers reining in AI budgets, that suggests Anthropic's recent breakneck growth may not hold — an awkward backdrop for an IPO.
9:06I see this as being the biggest problem facing Anthropic and ChatGPT ahead of their IPO later this year. For Anthropic, there's another problem. That is competition is closing in. Cursor, GitHub Copilot, OpenAI Codex are all quickly catching up to Claude Code. This is one reason why Microsoft reportedly has canceled most of its Claude Code licenses.
In short: Part of the ~$300B IPO wave hitting the market at "very expensive levels" and "being really picked off by China." (Danny's Kalshi pick: an Anthropic-IPO contract at 72¢.)
Anthropic is a private U.S. AI company (maker of Claude), part of the wave of roughly $300 billion in AI listings hitting the market "at very expensive levels." His double worry: the prices are stretched, and he says these models are "being really picked off by China" — copied cheaply, which dilutes the return on all that investment.
33:02And so the AI thing and the space war and the anthropic and open AI, these companies are are just coming into the market at very expensive levels and they're being really picked off by China. — Any other places like Brazil or anything that's catching your eye as far as resource driven economies? I realize there's a lot of you know you know elections coming up um you know in some of these places so it's hard to really game that but what are you looking at there as well kind of resourceoriented emerging markets — well we we did this with Argentina so
In short: A Grok competitor that is also xAI's biggest customer — a $1.25B/month compute deal (May 2026–May 2029, ~$45B contracted, not in backlog) that helps cover the AI segment's burn. App Economy notes Anthropic crossed ~$30B annualized in Apr 2026, ahead of OpenAI, as a valuation peg for SpaceX's AI segment.
In short: Paired with OpenAI in the "take the over on $200B combined revenue" call; Claude Code cited as giving better answers than the regular model even for investing questions.
12:20Yeah. It's fascinating. As a monopolist provider, they're rate limiting supply at some fundamental level. They dismissed Sam Altman as a podcast bro after they met with him. — [laughter] — You and we were talking about this backstage, which is you would take the over on OpenAI and Anthropic combined at 200 billion of revenue maybe 12 months, 18 months.
In short: The marks that make the myth: valuation ran $183bn (Sep-25) → $380bn (1Q26), and the WSJ reported a mooted $30bn round at a $900bn valuation — capital raised to fund $200bn (Google) / $100bn (Amazon) compute commitments. Alphabet is its largest investor, Amazon among its largest backers; their stake mark-ups are booked as "Other Income." The purest expression of the Great Circularity.
Anthropic is the engine of the whole scheme in Paulo's telling. Its "value" leapt from $183 billion to $380 billion in a couple of quarters and is reportedly being shopped at $900 billion — and it uses the cash it raises to promise hundreds of billions in spending to its own investors, Google and Amazon. As long as each new round is priced higher, Google and Amazon can keep marking up their stakes and calling it profit. The danger: none of these prices is set by a real public market. The moment Anthropic (or OpenAI) actually lists on a stock exchange, its share price becomes the "one true mark" nobody can fudge — and if it ever falls, the hyperscalers' paper profits go into reverse.
In short: Owned foundational-model leader — revenue "going from 100 million to a billion to 9 billion, and then it's already at 45 billion" run-rate, maybe $100B; with OpenAI ≈ $200B combined by year-end at "staggering," locked-in-compute margins — "you could be looking at something that's like 18 times earnings."
Anthropic is the private company behind the Claude AI models — one of Whale Rock's owned "foundational model" bets (the core AI engines everything else is built on). Customers pay per "token" (the chunks of text a model reads and writes), so usage translates straight into revenue.
Sacerdote's point at Sohn is the sheer slope: revenue went from $100M to $1B to $9B and is "already at 45 billion" on a run-rate basis, maybe heading to $100B. Even better is the profit picture — because these companies locked up scarce computing capacity early at fixed cost, every extra dollar of token revenue is very high margin, so profitability could look "staggering," on the order of 18× earnings. That, he argues, kills the worry that all the AI spending has no payoff.
6:48And I said, "It doesn't even need to grow for this stock to be a buy." This one's moving faster. And the revenue growth that we're seeing at Anthropic — going from 100 million to a billion to 9 billion, and then it's already at 45 billion. It's going to be maybe 100 billion.
In short: Racing OpenAI to be the first LLM IPO (~$1T+) — billionaires "dumping into passive investors" through fast-tracked index inclusion.
Anthropic is an AI company racing OpenAI to be the first large-language-model firm to go public, at a valuation he pegs above $1 trillion.
Same complaint as SpaceX: he sees billionaires "dumping" richly valued shares into ordinary index-fund investors by getting fast-tracked into the indices — pushing pricey, unproven stock onto passive buyers who don't even choose to own it.
5:14for each of those companies. Do these historically massive IPOs plus the AI hyperscaler spending create a scenario that in effect maybe insulates the markets potentially from a collapse until we at least get through these issues? — one thing about liquidity and the Trump team, and I'll get into the whole everything you laid out is a very well laid out.
In short: Reference — named in ACN's AI-platform partner ecosystem (alongside Google Cloud, Microsoft, Amazon, SAP, Salesforce, ServiceNow). Also: "the well-publicized misfires even by the best AI apps, such as Claude's Opus 4.6, underscore… the critical importance of having expert supervision" — the human-oversight argument behind the consulting demand.
In short: Its model Claude is being "distilled" / stolen via China's Kimi (≈25k fake accounts raiding it daily) and pushed to open source — diluting US AI return-on-capital.
Anthropic is a private U.S. AI company (maker of the Claude assistant). He raises it as a threat to U.S. AI profits: he alleges Chinese platforms are "distilling" Claude — using roughly 25,000 fake accounts to copy its outputs and rebuild its capabilities cheaply, then pushing them into free open-source models. If U.S. firms spend billions and rivals copy the result for almost nothing, the return on that spending shrinks.
41:57They're taking the code, they're putting it into the this what's called distillation. And it's basically taking it into this Kimmy device, the platform, and then they're putting it out to open source. So, imagine a $2 trillion cap backs with Larry Ellison flexing his muscles Zuckerberg flexing his muscles $2 trillion of cap ex everyone's trying to outspend each other, right? And this is classic malinvestment.
In short: Named alongside SpaceX and ByteDance as one of the private businesses a Scottish Mortgage holder gets exposure to. Not in the disclosed top five, and no stance is offered.
In short: Catalyst reference, not a stance: "last week's release of Anthropic's Claude Cowork set off a demolition in US software on fears it could substitute for or commoditize many existing SaaS and application workflows" — the trigger behind the software ROS basket (IGV and its constituents).
In short: Reference: cited (via The Information) as poised to out-earn OpenAI this year on API/coding revenue — evidence OpenAI's technological lead has narrowed, part of the AI-narrative turn.
Nothing matches this filter.
Verbatim excerpts from the public transcripts (auto-pulled at each mention's timestamp, lightly cleaned). Timestamps link into the video; "source page" opens that commentator's full analysis at this row. Click a mention's header line to expand it (one open at a time). For personal study — not investment advice.