Why Steve Eisman Isn't Buying Anthropic's IPO

Anthropic

Across the shows: 66 speakers in other episodes, 59 bullish and 51 bearish calls, past 30 days

• Steve Eisman · The Real Eisman PlaybookBearish

  Anthropic's post-token-maxing revenue growth and revenue-to-compute ratio both fell sharply in Q3.

  Eisman rebuilt the quarterly series from the S-1 and reports on Q3 revenue. Sequential revenue growth fell from 147% in Q2 to 45% in Q3. Revenue divided by compute infrastructure expense fell from 2.08x to 1.7x, after rising from 0.69x in Q4 2025 to 1.4x in Q1 and 2.08x in Q2. He reads this as the end of the 'token maxing' surge, when enterprises ordered engineers to use AI without regard to cost. He says the Q3 slowdown shows the peak effect has reversed and that the S-1 is backward-looking, since it only runs through Q2 and omits Q3. He also cites an unverified report that Anthropic cut prices by about 50% in October to win share.

  “Third quarter, token maxing is over, sequential revenue growth goes from 147% to 45%, and the revenue divided by compute goes from 2.08 to 1.7. That tells you something.”at 6:21

• Steve Eisman · The Real Eisman PlaybookNeutral

  Anthropic's lack of shareholder rights is disclosed and does not by itself decide investability.

  Eisman calls the S-1 governance structure ridiculous and says shareholders have no rights. He then says governance does not determine whether the stock is investable. He points to Meta, which has the same kind of criticism but has still performed very well as a stock. He describes the situation as founder worship in tech, which investors can accept or reject. He sees it as a choice for each investor, not a deal-breaker.

  “[Does it say something about whether it's investable?] I don't think so. We could make the same criticism of Meta, and, you know, Meta as a stock generally has done very, very well.”at 13:14

• Ed Elson · Prof G MediaNeutral

  Anthropic's $80B first-half net loss is a red herring from convertible-note fair-value accounting.

  Elson argues that the headline $80B net loss in the first half does not reflect the operating business. The loss came from an increase in the fair value of Anthropic's convertible notes. Investors who lent money later received equity, so accounting rules required a write-up of an expense. The loss mainly reflects how much the company's value rose in a short time. He treats the headline as misleading, not as a sign of weak economics.

  “Because those losses were a product of the increase in the fair value of Anthropic's convertible notes... So the $80,000,000,000 net loss, that's a red herring.”at 32:41

• Ed Elson · Prof G MediaBearish

  Anthropic's revenue growth slowed sharply in Q3, which may signal the AI frenzy is ending.

  Elson argues that the S-1 does not show the full growth slowdown. Growth was about 150% quarter over quarter in Q2 but under 50% in Q3. The filing leaves Q3 out of its main charts, which is a major omission. He says that while this is not necessarily bad, it suggests the AI frenzy may be coming to an end. He treats Q3 as the data point that matters most for the next year, and he agrees with Eisman on the slowdown.

  “revenue growth slowed down significantly in q three... Now that's not necessarily bad, but it does show that maybe the AI frenzy is beginning to come to an end.”at 33:30

• Ed Elson · Prof G MediaBearish

  Anthropic's $800B+ compute commitments will be added to losses, which are about to explode.

  Elson's main point is that the S-1's $800B+ of future compute commitments are not yet in the income statement but will be. The contracts include $110B with Amazon, $115B with Google and $161B with Broadcom. The filing also says the company will significantly increase capex to build its own compute, which the $800B figure leaves out. Elson argues the true commitment is therefore probably much higher and that losses will rise. He says the narrowing losses will not persist, and that the picture will look drastically different next year. He also raises, as a theory, that the IPO may have been timed to avoid showing the losses.

  “that $800,000,000,000 number is actually probably gonna be a lot higher. And as a result, this company's losses are about to explode.”at 34:46

• Ed Elson · Prof G MediaBearish

  Anthropic's reported $2T valuation is not what a proven safe corporate investment looks like.

  Elson's closing argument is that the S-1 shows a company with an incredible product but an unproven business model. Losses, unpriced compute commitments and slowing growth all point to large risk. He says it is still a highly risky venture bet and could go spectacularly wrong. He therefore argues that a valuation near $2T does not reflect a proven, safe corporate investment, and that investors should treat the stock with caution.

  “What it certainly isn't is a proven safe corporate investment, something that its potentially $2,000,000,000,000 valuation might suggest.”at 35:46

CoreWeave

Across the shows: 54 speakers in other episodes, 14 bullish and 50 bearish calls, past 30 days

• Steve Eisman · The Real Eisman PlaybookBearish

  CoreWeave's 15% cost of debt is prohibitive for building AI data centers; the business is at risk.

  Eisman argues that CoreWeave's business model is simple: it raises debt and builds AI data centers. A 15% borrowing cost is, in his view, prohibitive for that model. He says he cannot prove it, but if the cost of debt gets worse, the company is effectively out of business. He notes that NVIDIA could backstop the company, which he sees as possible but not assured. He suggests watching the debt price as the key indicator.

  “At 15%, I think, although I can't prove it, I think it's prohibitive for them to raise debt at 15% and build AI data centers.”at 15:23

Waymo

Across the shows: 4 speakers in other episodes, 3 bullish and 1 bearish calls, past 30 days

• Dan Chiolan · Prof G MediaBullish

  Waymo is far ahead of Tesla in robotaxi, with about seven times Tesla's weekly miles.

  Chiolan says Waymo is clearly the leader in US robotaxis. It does about seven times Tesla's weekly miles. It operates in 15 American cities and completes about 500,000 paid rides a week. Chiolan notes that no Chinese company does as many paid rides, so US commercialization leads globally. He presents these operating metrics as the evidence for Waymo's lead.

  “Currently, Waymo does about seven times the amount of miles that Tesla does weekly. They're in 15 American cities. They're doing 500,000 paid rides weekly.”at 22:18

Tesla

Across the shows: 21 speakers in other episodes, 15 bullish and 19 bearish calls, past 30 days

• Dan Chiolan · Prof G MediaBullish

  Tesla's robotaxi is scaling fast and has a large cost edge, though it still trails Waymo.

  Chiolan says Tesla moved from a service with human monitors to fully driverless commercial service in six cities in a few months. He sees the scaling as real progress, even though Waymo is far ahead on volume. He says Tesla's main edge is cost. The Cybercab costs about $25,000 to produce, while Waymo's vehicles cost more than $100,000 because of their heavier sensor suite. He says Tesla is still significantly behind, but its position has improved quickly.

  “But the biggest advantage that Tesla has over Waymo is cost. So those Tesla Cybercabs cost about $25,000 to produce, whereas Waymo's cost upward of a $100,000.”at 23:53

Hertz Global Holdings

• Dan Chiolan · Prof G MediaBullish

  Car rental companies like Hertz will see a windfall from servicing autonomous vehicles.

  Chiolan says robotaxi companies will have to maintain thousands of vehicles at scale, which means charging, parking and cleaning them. Ride-hail drivers currently do this themselves, so the work is new. He argues the car rental industry is well placed, since it already has facilities around the country. He cites Hertz's announcement that it is standing up a new business line to service autonomous vehicles, and he expects more rental companies to follow.

  “there is an industry here in America that I think is gonna see a huge windfall from this need, which is the car rental industry.”at 30:36

Robotics & autonomous vehicles

• Dan Chiolan · Prof G Media

  Autonomous vehicles can make driving much safer, since human error causes about 94% of crashes.

  Chiolan makes the safety case for AVs. About 40,000 Americans die in car crashes each year, and the global toll is over a million, and car crashes are the leading killer of people aged 5 to 29. He says 94% of crashes are due to human error, so the danger comes from drivers, not cars or roads. He argues AVs can make something Americans do every day far safer.

  “ninety four percent of car crashes are due to human error. So really driving isn't dangerous because cars are dangerous or roads are dangerous. It's because humans are dangerous.”at 20:36

• Dan Chiolan · Prof G Media

  China leads on AV manufacturing cost, but the US leads commercialization; the race plays out abroad.

  Chiolan says China leads in manufacturing, because it can build electric and autonomous vehicles at a much lower cost than any American company. The US leads in commercialization, since no Chinese company does as many paid rides as Waymo. He expects the two countries not to allow each other's AV companies into their home markets. The contest will therefore play out in the Middle East and Europe. There, countries have mostly chosen Chinese companies so far, and American firms have not expanded internationally, which he attributes partly to a focus on the domestic market.

  “I think where China is really winning is in manufacturing because they have the ability to manufacture electric vehicles and autonomous vehicles at a cost that's much lower than than any American company.”at 25:01

• Dan Chiolan · Prof G Media

  Fragmented state-by-state AV rules and souring public opinion will keep the political battle going.

  Chiolan says there is no federal framework for AVs. Each company must lobby every state to test, run driverless service and expand, which creates a messy, drawn-out process. This has produced a large lobbying battle between unions and robotaxi companies, and Waymo has sharply increased its federal lobbying. He also argues AVs are not insulated from the backlash against AI and data centers. Gallup found that the share of Americans who think AVs improve road safety fell by 12 points, from about 52% in 2018, even though AVs have become much safer.

  “it's currently very fragmented, and that's because it's all happening at the state level. There's really been no federal framework or action sort of passed by the federal government.”at 26:53

• Dan Chiolan · Prof G Media

  US lidar makers should win, since about 95% of car lidar is made in China and is a security risk.

  Chiolan describes a picks-and-shovels play. About 95% of passenger-car lidar sensors, the expensive sensors used by Waymo, Zoox and most AV makers, are made in China. He says that is a national-security risk, and Chinese-made sensors are already banned for military use. Congressional bills would also ban them for civilian use. If those bills pass, US companies would have a large opportunity to manufacture the sensors. He considers lidar makers the first opportunity in AVs.

  “I think the first opportunity here are gonna be companies that make LiDAR sensors... currently, 95% of passenger car LiDAR sensors, which are in these autonomous vehicles, are made in China.”at 29:29

• Dan Chiolan · Prof G Media

  Autonomous trucking could be very lucrative, since driver pay is about 44% of trucking's marginal costs.

  Chiolan argues trucking is a large opportunity for autonomy. About 44% of trucking's marginal costs come from driver pay and benefits. If AV technology can automate that cost, the savings would be large. He mentions Aurora, which is testing self-driving trucks on American roads, and Tesla's Semi, which he says will eventually have full self-driving capability but does not yet.

  “the last winner is gonna be the trucking industry. Because if you look at trucking, about 44% of trucking's marginal costs come from driver compensation and benefits.”at 30:56

Open-source AI models

• Steve Eisman · The Real Eisman Playbook

  Chinese open-weight models are taking AI market share as buyers become more cost-conscious.

  Eisman says that after the first half of 2026, customers became much more cost-conscious and the end of token maxing followed. Chinese open-weight models took more market share in that period. He links this to the Q3 revenue slowdown at Anthropic and to the price pressure that might have forced its reported 50% price cut.

  “people got a lot more cost conscious. The Chinese open weight models took more market share.”at 5:04

AI infrastructure & data centers

• Steve Eisman · The Real Eisman Playbook

  Anthropic's roughly $800B of compute commitments will push its expenses up sharply.

  Eisman notes that the S-1 lists future capital commitments of $800B to $900B, depending on what is included. He says the company's expenses will rise once it commits to that spending, though he does not know by how much because he is not the CFO. He cites a report that capex could reach $80B to $100B next year, and he is unsure how it will be accounted for. He calls the commitments a concern, but hedges by saying 'I don't know if the word is concerning'.

  “when you commit to $800,000,000,000 your expenses are going up.”at 8:44

Frontier AI labs & model race

• Steve Eisman · The Real Eisman Playbook

  Anthropic has no moat; it can only win share through price, which is a dangerous franchise.

  Eisman argues that Anthropic's franchise has no moat. If the reported 50% price cut is true, it shows that price is the only lever for gaining share. He says a business that commits to about $800B of compute while its franchise rests entirely on price is dangerous. He hedges that he has not confirmed the price cut, but says that if it is true it reinforces the no-moat thesis. Chinese open-weight models are taking share, which adds to the pricing pressure.

  “it just seems, you know, that there are no moats around this business... if it's the case that they actually did cut prices by 50%, that would reinforce the concept that there are no moats and that the only way you're gonna gain share is price.”at 11:32