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

Ed Zitron Where's Your Ed At Published

The Subprime Data Center Crisis

We recorded 20 claims from this post. 11 were checked against evidence. 9 can’t be checked. 32 more are held. Read the post at its source.

Claims from this post (20)

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: Can’t be checkedSaid

    “I realize that the comparison between an AI data center and a CDO might seem a little ridiculous, but they’re actually remarkably similar.”

    What we tested: AI data center financing structures are remarkably similar to CDOs.

  2. Assessment: Can’t be checkedSaid

    “That being said, the number is likely much higher.”

    What we tested: Outstanding AI data center debt is likely much higher than Bloomberg's $500 billion estimate.

  3. Assessment: SupportedSaid

    “The problem with these SPV-based deals is that they allow companies to, at least on a balance sheet basis, hide the scale of their debts.”

    What we tested: SPV-based data center deals allow companies to hide the scale of their debts, at least on a balance sheet basis.

    What happened: The same paragraph expressly concludes that Meta is not the venture's primary beneficiary and does not consolidate it; the quoted sentence describes its continuing involvement. The following paragraph discloses its equity investment and maximum exposure, supporting balance-sheet exclusion without establishing concealment from the notes. (Meta Platforms, )

  4. Assessment: Can’t be checkedSaid

    “I want to be very blunt about something: we do not, at this point, have a firm hand on exactly how much demand there is for AI compute, and evidence suggests that it’s much, much smaller than we’ve been led to believe.”

    What we tested: Evidence suggests demand for AI compute is much smaller than the public has been led to believe.

  5. Assessment: UnresolvedSaid

    “Sidenote: Furthermore, any demand for compute that exists currently is inflated by the effective subsidization of paying subscribers to ChatGPT and Claude, who can burn tens of thousands of dollars worth of tokens while only paying $200-a-month.”

    What we tested: Current compute demand is inflated because ChatGPT and Claude subscribers paying $200 a month can consume tens of thousands of dollars of tokens.

  6. Assessment: Can’t be checkedSaid

    “And I’m not sure that people are taking me seriously when I say that AI compute demand does not exist at the scale that it needs to, will likely never reach that scale, and data center construction is a debt-funded asset bubble with ruinous consequences.”

    What we tested: AI compute demand will likely never reach the scale needed to justify the data center buildout.

  7. Assessment: ContradictedSaid

    “Per my own analysis, NVIDIA’s predicted $1 trillion in Blackwell and Vera Rubin GPU sales (by the end of 2027) represents around 40GW of data center capacity, which will, assuming a PUE of 1.35, result in around 30GW of usable capacity. At a cost of around $12 million a megawatt, that works out to around $435 billion in global annual compute revenue to make these data centers necessary.”

    What we tested: By his own analysis, NVIDIA's predicted $1 trillion in Blackwell and Vera Rubin GPU sales by the end of 2027 represents around 40GW of data center capacity, or around 30GW usable at a PUE of 1.35, which at around $12 million a megawatt works out to around $435 billion in global annual compute revenue needed to make these data centers necessary.

    What happened: The following paragraph supplies 40 GW, PUE 1.35 and $12 million per MW. Independently multiplying these stated operands gives $355.56 billion annually; using the rounded 30 GW gives $360 billion. (Where's Your Ed At, )

  8. Assessment: UnresolvedSaid

    “Putting aside whether data centers are profitable or not (they aren’t),”

    What we tested: AI data centers are not profitable.

  9. Assessment: UnresolvedSaid

    “If we assume — and I’m being nice! — that there’s $120 billion in annual compute demand, and take into account that tens of billions of dollars’ worth of data centers have been announced since, this means that there’s over 15 times more data centers being planned than the demand that actually exists, and 70% to 90% of that demand is from Anthropic and OpenAI’s unprofitable services.”

    What we tested: Assuming a generous $120 billion in annual compute demand and counting the tens of billions of dollars of data centers announced since February's Sightline report, there are over 15 times more data centers being planned than the demand that exists, and 70% to 90% of that demand is from Anthropic and OpenAI's unprofitable services.

  10. Assessment: Can’t be checkedSaid

    “Financial institutions have built entire models based on logic that borders on childish.”

    What we tested: Financial institutions have built entire models of AI data center investment safety on logic that borders on childish.

  11. Assessment: Can’t be checkedSaid

    “The scenario I’m talking about is one where the vast majority of AI data centers go unused, and because the vast majority of data centers are paid out of customer revenues, 80% or more of the funds invested in AI data center debt will be lost.”

    What we tested: In a scenario where the vast majority of AI data centers go unused, 80% or more of the funds invested in AI data center debt will be lost, because the vast majority of data centers are paid out of customer revenues.

  12. Assessment: UnresolvedSaid

    “I need everybody to realize that the overbuild has real consequences, and that there’s no post-bubble economy to replace this one, because it’ll be just as expensive to run these things in 2030 as it is today.”

    What we tested: AI data centers will be just as expensive to run in 2030 as today, so no post-bubble economy will replace the current one.

  13. Assessment: UnresolvedSaid

    “It’s easy to dismiss this entirely as “rich people problems,” but AI data centers are increasingly funded — both directly and through private credit — using pension and insurance funds that rely on these (theoretical) payments for future yield to pay out premiums.”

    What we tested: AI data centers are increasingly funded with pension and insurance money that depends on these payments.

  14. Assessment: Can’t be checkedSaid

    “While subprime mortgage defaults were a kind of slow, ugly boil, it’s much more likely that the collapse of the subprime data center bubble will happen in fits and starts as capacity comes online and, assuming Anthropic and OpenAI don’t swoop in, goes unused.”

    What we tested: It is much more likely that the subprime data center bubble will collapse in fits and starts, rather than as a slow boil like subprime mortgage defaults, as capacity comes online and goes unused, assuming Anthropic and OpenAI don't swoop in.

  15. Assessment: UnresolvedSaid

    “If the market continues to sour on AI data center debt, it will eventually become difficult to impossible for hyperscalers to keep issuing bonds, leaving them with only equity sales (like Google’s $85bn stock sale) that are equal parts limited and desperate.”

    What we tested: If the market keeps souring on AI data center debt, hyperscalers will eventually find it difficult or impossible to issue bonds and be left with limited equity sales.

  16. Assessment: Can’t be checkedSaid

    “In a sane world, Coreweave should not have been able to borrow as much as it did — and the same applies to the countless other debt-laden neoclouds that are, for the most part, cookie-cutter versions of Coreweave.”

    What we tested: CoreWeave and similar neoclouds should not have been able to borrow as much as they did.

  17. Assessment: UnresolvedSaid

    “For me to be wrong, there will have to be dramatic amounts of AI compute demand — hundreds of billions’ worth — within the next 3 years, at a time when there’s little more than $120 billion, with 80% or more of that coming from two companies that can only afford it because they have near-infinite sums of venture capital behind them.”

    What we tested: For his thesis to be wrong, there would have to be dramatic amounts of AI compute demand, hundreds of billions of dollars' worth, within the next 3 years, at a time when there is little more than $120 billion, 80% or more of it from two companies that can only afford it because of near-infinite venture capital.

  18. Assessment: Can’t be checkedSaid

    “The AI bubble is, like the great financial crisis, a product of information asymmetry — companies intentionally obfuscating how much revenue they have, how much data center capacity they have, how much revenue that capacity generates, how much demand they have for their services, and how many real dollars actually flow from the AI industry outside of investments in semiconductors.”

    What we tested: The AI bubble results from information asymmetry created by companies deliberately obscuring revenue, capacity, and demand.

  19. Assessment: UnresolvedSaid

    “This is an inevitability — and something that’s very obvious when you sit down and actually try and work out how much capacity there is versus how much people are actually paying for AI compute.”

    What we tested: The collapse of the vast majority of AI data centers is an inevitability, which is obvious when you compare how much capacity there is with how much people actually pay for AI compute.

  20. Assessment: UnresolvedSaid

    “Every time that the media has accepted a spurious announcement or a questionable run rate or a circular deal or the outright refusal of hyperscalers to disclose their AI revenues, they help inflate the AI bubble and endanger the futures of millions of people, especially those tied up in a stock market increasingly-dominated by NVIDIA and other tech stocks.”

    What we tested: Every time the media has accepted a spurious announcement, a questionable run rate, a circular deal, or hyperscalers' outright refusal to disclose their AI revenues, it has helped inflate the AI bubble and endangered the futures of millions of people, especially those tied up in a stock market increasingly dominated by NVIDIA and other tech stocks.

Held from this post (32)

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