Pundit ClaimsAI claims by Gary Marcus and Ed Zitron, with the evidence

Ed Zitron Where's Your Ed At Published

The Subprime AI Crisis Is Here

We recorded 43 claims from this post. 24 were checked against evidence. 19 can’t be checked. 41 more are held. Read the post at its source.

Claims from this post (43)

In the order they appear in the post. Each quotation is the speaker’s words in this post. A label such as “Supported” or “Can’t be checked” is the claim’s current assessment; its page has the evidence and the reasons.

  1. Assessment: UnresolvedSaid

    “At some point, the incredible, toxic burn-rate of generative AI is going to catch up with them, which in turn will lead to price increases, or companies releasing new products and features with wildly onerous rates — like the egregious $2-a-conversation rate for Salesforce’s “Agentforce” product — that will make even stalwart enterprise customers with budget to burn unable to justify the expense.”

    What we tested: Generative AI's burn rate will eventually force price increases or onerous new pricing that enterprise customers cannot justify.

  2. Assessment: UnresolvedSaid

    “The AI industry has done a great job in obfuscating exactly how brittle its economics really are”

    What we tested: The AI industry has obscured how brittle its economics are.

  3. Assessment: UnresolvedSaid

    “Venture capitalists are funded by limited partners (EG: pension funds, investment banks and wealthy individuals), and the venture capital industry is facing an historic liquidity crisis (IE: they can’t raise money and their investments aren’t selling), which means that it cannot sustain the AI industry forever.”

    What we tested: Because venture capital faces a historic liquidity crisis, it cannot sustain the AI industry indefinitely.

  4. Assessment: UnresolvedSaid

    “All evidence points to GPU compute either being a low or negative-margin business.”

    What we tested: GPU compute rental is a low- or negative-margin business.

  5. Assessment: UnresolvedSaid

    “This means that AI labs are often having to pony up massive amounts of up-front capital on top of their incredibly high ongoing costs.”

    What we tested: AI labs often must pay large up-front sums for future compute on top of very high ongoing costs.

  6. Assessment: MixedSaid

    “A coding model relies heavily on chain-of-thought reasoning, which means that despite the cost of tokens coming down (which does not mean the price of providing them has decreased, it’s a marketing move), models are using far, far more tokens, increasing costs across the board.”

    What we tested: Coding models rely heavily on chain-of-thought reasoning and so use far more tokens, which raises costs across the board even though token prices have fallen; lower token prices do not mean the cost of providing tokens has fallen, and are a marketing move.

    What held: Cursor identifies reasoning stages as important to its coding model. This supports the reliance-on-reasoning component, without establishing industry-wide token growth or costs. (Cursor, )

    What did not: This coding comparison reports better performance with fewer tokens. It contradicts an across-the-board increase in token requirements, although it does not measure aggregate provider spending. (Anthropic, )

  7. Assessment: UnresolvedSaid

    “AI labs are thus forced to raise prices on AI startups, either through tolls (priority processing) or raw cost increases.”

    What we tested: Cash and compute pressure forces AI labs to raise prices on AI startups via priority tolls or direct cost increases.

  8. Assessment: UnresolvedSaid

    “AI labs have, this entire time, been massively subsidizing their own products.”

    What we tested: AI labs have been heavily subsidizing their own products all along.

  9. Assessment: UnresolvedSaid

    “The problem is that a subsidy is always a losing proposition, which means that at some point Anthropic and OpenAI will have to massively reduce the amount of tokens that people use on their accounts.”

    What we tested: Anthropic and OpenAI will eventually have to massively cut how many tokens subscribers can use.

  10. Assessment: UnresolvedSaid

    “At some point, the cost of doing business with Anthropic and OpenAI will kill AI startups, as there is no point at which any of them become sustainable, which will in turn kill the revenue from selling access to their models.”

    What we tested: At some point the cost of doing business with Anthropic and OpenAI will kill AI startups, because none of them ever become sustainable, and this will in turn kill the labs' revenue from selling access to their models.

  11. Assessment: UnresolvedSaid

    “At some point, users will be forced to burn tokens at a rate that actually matches their subscription costs, which will reduce the value of the product, which will in turn reduce the amount of subscribers they will have.”

    What we tested: Users will eventually be limited to token use matching their subscription price, lowering product value and subscriber numbers.

  12. Assessment: UnresolvedSaid

    “As I discussed back in AI Is A Money Trap, there really is no liquidity mechanism for the billions of dollars sunk into most AI companies.”

    What we tested: There is no real exit/liquidity path for the billions invested in most AI companies.

  13. Assessment: UnresolvedSaid

    “In any case, nowhere are the subsidies higher than those of Anthropic and OpenAI, who use their tens of billions of dollars of funding to allow users to burn anywhere from $3 to $13 per every dollar of subscription revenue to outpace their competition.”

    What we tested: Anthropic and OpenAI provide the highest subsidies, letting users burn $3–$13 of compute per subscription dollar.

  14. Assessment: Can’t be checkedSaid

    “Price hikes and product changes are likely to accelerate in the next few months as things get desperate.”

    What we tested: AI price hikes and product changes will likely accelerate over the next few months.

  15. Assessment: UnresolvedSaid

    “No! AI training costs are not going away. They are an inherent part of running these companies, and are not capex. They are operating expenses.”

    What we tested: AI training costs are not going away; they are an inherent part of running these companies and are operating expenses, not capex.

  16. Assessment: SupportedSaid

    “The cost of tokens is not the same as the cost of serving tokens!”

    What we tested: Falling token prices do not mean the cost of serving tokens has fallen.

    What happened: The report defines serving expense separately from token price, which it identifies as the user's payment. This directly supports the distinction in the claim. (NIST Center for AI Standards and Innovation, )

  17. Assessment: Can’t be checkedSaid

    “Anthropic claimed that this was simply the expiration of holiday rate limits, but in reality, this is all part of Anthropic’s continual manipulation of rate limits to con customers into buying Claude subscriptions that decay in value.”

    What we tested: Anthropic continually manipulates rate limits to lure customers into Claude subscriptions that lose value.

  18. Assessment: UnresolvedSaid

    “The media campaign worked, and Anthropic was able to close a $30 billion round on February 12, 2026.”

    What we tested: Anthropic's Claude Code media campaign succeeded and helped it close a $30 billion round on February 12, 2026.

  19. Assessment: Can’t be checkedSaid

    “I think there’re a few ways this goes:”

    What we tested: Anthropic's rate limit situation could play out in a few ways.

  20. Assessment: Can’t be checkedSaid

    “Anthropic will announce that it’s “fixed the bug” (IE: eased rate limits it intentionally set) and apologize to the community, prolonging the inevitable. Rate limits will continue to decay over time, just at a slower pace.”

    What we tested: One possible outcome: Anthropic eases limits, calls it a bug fix, and apologizes, while limits keep decaying more slowly.

  21. Assessment: Can’t be checkedSaid

    “Anthropic keeps the limits where they are, and we hit a new normal that makes everybody really mad.”

    What we tested: Another possible outcome: Anthropic keeps the current limits as a new normal that angers users.

  22. Assessment: UnresolvedSaid

    “I wager that this is just the first of a few major belt-tightening operations from both Anthropic and OpenAI as they desperately shoulder-barge each other to file the world’s worst S-1.”

    What we tested: He wagers that this is just the first of a few major belt-tightening operations from both Anthropic and OpenAI as they race toward filing S-1s.

  23. Assessment: UnresolvedSaid

    “Yet the sheer fury of Anthropic’s customers only proves the fundamental weakness of Anthropic’s business model, and the impossibility of ever finding any kind of profitability.”

    What we tested: Anthropic's business model is fundamentally weak, and it will never become profitable.

  24. Assessment: Can’t be checkedSaid

    “While it’s really easy to make fun of people obsessed with LLMs, I want to be clear that Anthropic and OpenAI are inherently abusive companies that have built businesses on theft, deception and exploitation.”

    What we tested: Anthropic and OpenAI are inherently abusive companies that have built their businesses on theft, deception and exploitation.

  25. Assessment: Can’t be checkedSaid

    “I’d argue that annual subscriptions to these services amount to, if not fraud, a level of consumer deception that deserves legal action and regulatory involvement.”

    What we tested: Annual AI subscriptions are consumer deception that deserves legal action and regulatory involvement.

  26. Assessment: Can’t be checkedSaid

    “To be clear, no AI company should have ever sold a monthly subscription, as there was never a point at which the economics made sense.”

    What we tested: AI companies should never have sold monthly subscriptions because the economics never made sense.

  27. Assessment: Can’t be checkedSaid

    “Yet had these companies actually charged their real costs, nobody would have bothered with AI, because even with these highly-subsidized subscriptions, AI still hasn’t delivered meaningful productivity benefits”

    What we tested: Had AI companies charged their real costs, nobody would have bothered with AI, because even with highly subsidised subscriptions AI still hasn't delivered meaningful productivity benefits.

  28. Assessment: Can’t be checkedSaid

    “Based on the reactions of their users, I don’t really see how the AI labs — or AI startups, for that matter — fix this problem.”

    What we tested: Based on users' reactions, he doesn't really see how AI labs or AI startups can fix the problem of users who won't tolerate rate limits and price increases.

  29. Assessment: UnresolvedSaid

    “The very foundation of every AI startup is economically broken.”

    What we tested: Every AI startup is economically broken at its foundation.

  30. Assessment: Can’t be checkedSaid

    “Even if this week’s rate limits are an overcorrection, one has to imagine they resemble the future of Anthropic’s products, and are indicative of a larger pattern of decay in the value of its subscriptions.”

    What we tested: Anthropic's current rate limits likely preview its future and reflect ongoing decay in subscription value.

  31. Assessment: UnresolvedSaid

    “There is no pricing that makes sense that customers will pay at scale, nor is there a magical technological breakthrough waiting in the wings that will reduce costs.”

    What we tested: There is no sensible pricing that customers will pay at scale, and no magical technological breakthrough is waiting in the wings that will reduce costs.

  32. Assessment: UnresolvedSaid

    “Vera Rubin will not save AI, nor will some sort of “too big to fail” scenario, because “too big to fail” was based on the fact that banks would have stopped providing dollars to people and insurance companies would have stopped issuing insurance.”

    What we tested: Neither NVIDIA's Vera Rubin nor a 'too big to fail' bailout will save the AI industry.

  33. Assessment: UnresolvedSaid

    “The AI data center companies that, thanks to readily-available debt, have started 200GW of projects (and only started building 5GW of them) for AI demand that doesn’t exist, entirely based on the theoretical sense that maybe it will in the future.”

    What we tested: Debt-funded data center companies have started 200GW of projects, only 5GW under construction, for AI demand that does not exist.

  34. Assessment: Can’t be checkedSaid

    “I think everybody in the AI industry has deluded themselves into believing they have the mandate of Heaven.”

    What we tested: He thinks everybody in the AI industry has deluded themselves into believing they have the mandate of Heaven.

  35. Assessment: UnresolvedSaid

    “Any reduction in capex from big tech is a sign that the AI bubble is bursting, as NVIDIA’s continued growth only comes from Microsoft, Google, Amazon, Meta, Oracle and other large companies buying tens of billions of dollars of servers from Taiwanese ODMs like Foxconn and Quanta.”

    What we tested: Any reduction in big tech capex is a sign that the AI bubble is bursting, because NVIDIA's continued growth comes only from Microsoft, Google, Amazon, Meta, Oracle and other large companies buying tens of billions of dollars of servers from Taiwanese ODMs.

  36. Assessment: Can’t be checkedSaid

    “Any further price increases or service degradations from AI startups, such as Cursor, Perplexity, Harvey, Lovable or Replit.”

    What we tested: Further price increases or service cuts at AI startups would be a warning sign of the AI bubble's collapse.

  37. Assessment: Can’t be checkedSaid

    “Any discussion of layoffs at AI companies.”

    What we tested: Discussion of layoffs at AI companies would be a warning sign of the AI bubble's collapse.

  38. Assessment: Can’t be checkedSaid

    “The collapse of a data center deal that has yet to commence construction.”

    What we tested: The collapse of a data center deal before construction would be a warning sign of the AI bubble's collapse.

  39. Assessment: Can’t be checkedSaid

    “The collapse of a data center already in construction, but before it’s finished.”

    What we tested: The collapse of a data center under construction would be a warning sign of the AI bubble's collapse.

  40. Assessment: Can’t be checkedSaid

    “The collapse of an already-constructed data center.”

    What we tested: The collapse of a completed data center would be a warning sign of the AI bubble's collapse.

  41. Assessment: Can’t be checkedSaid

    “The Further Collapse of Stargate Abilene: If anything happens to the construction of OpenAI’s flagship data center (being built by Oracle) in Abilene Texas, you know shit is getting bad.”

    What we tested: Any problem with construction of OpenAI's Stargate Abilene data center would signal serious trouble for the AI bubble.

  42. Assessment: UnresolvedSaid

    “This puts SoftBank in a very precarious position.”

    What we tested: SoftBank raising $40 billion in debt payable within a year to fund its share of OpenAI's round, going beyond its promised 25% loan-to-asset ratio, puts it in a very precarious position.

  43. Assessment: Can’t be checkedSaid

    “NVIDIA misses on earnings: This is an obvious one, but I think the markets will crap their pants if NVIDIA misses on earnings estimates.”

    What we tested: If NVIDIA misses earnings estimates, markets will react with panic.

Held from this post (41)

These records did not pass every check, so nothing about them is published. Held is not a finding.