The Enshittifinancial Crisis
We recorded 26 claims from this post. 13 were checked against evidence. 13 can’t be checked. 28 more are held. Read the post at its source.
Claims from this post (26)
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.
Assessment: UnresolvedSaid
“When Oracle announced its $300 billion deal with OpenAI in September — one that Oracle does not have the capacity to fill and OpenAI does not have the money to pay for – analysts heaved and stammered like horny teenagers seeing their first boob:”
What we tested: Oracle lacks the capacity to fulfil its $300 billion deal with OpenAI, and OpenAI lacks the money to pay for it.
Assessment: Can’t be checkedSaid
“It turns out the deal didn’t exist and everybody fell for it! NVIDIA hasn’t sent a dime and likely won’t.”
What we tested: NVIDIA has not sent any money under its letter of intent to invest up to $100 billion in OpenAI, and likely will not.
Assessment: Can’t be checkedSaid
“Similarly, it’s time for everybody — analysts, the media, members of congress, the fucking pope, I don’t care — to start treating these companies with suspicion, and to start demanding timelines.”
What we tested: Everybody, including analysts, the media and members of Congress, should start treating these companies with suspicion and start demanding timelines.
Assessment: Can’t be checkedSaid
“These subtle details suggest that the deal is not going to be for $15 billion, and the lack of activity suggests it might not happen at all.”
What we tested: The NVIDIA/Microsoft investment in Anthropic will likely not total $15 billion and might not happen at all.
Assessment: Can’t be checkedSaid
“These deals are announced with the intention of suggesting there is more revenue and money in generative AI than actually exists.”
What we tested: AI deal announcements are made to suggest more revenue and money exist in generative AI than actually do.
Assessment: UnresolvedSaid
“It allegedly will burn $2.8 billion this year (I believe it burned much, much more) and raised $16.5 billion in funding”
What we tested: Anthropic burned much more than the reported $2.8 billion in 2025.
Assessment: UnresolvedSaid
“Anthropic will make no more than $5 billion this year, and has raised $17.5bn (with a further $2.5bn coming in the form of debt).”
What we tested: Anthropic will make no more than $5 billion in revenue in 2025.
Assessment: UnresolvedSaid
“Anthropic cannot afford to pay Broadcom $10 billion right now!”
What we tested: Anthropic currently cannot afford to pay Broadcom $10 billion for its chip order.
Assessment: ContradictedSaid
“I want to be really, abundantly clear about what’s happening: every single stock you see “growing because of AI” outside of those selling RAM and GPUs is actually growing because of something else.”
What we tested: Apart from RAM and GPU sellers, stocks said to be growing because of AI are actually growing for other reasons.
What happened: Reuters reports an almost 9% early-trading gain after an AI-demand-driven forecast increase and connects the broader rally to AI tools and defense spending. This supports AI as a contributor, while acknowledging a material competing explanation. (Reuters, republished by Yahoo Finance, )
Assessment: UnresolvedSaid
“They are all losing tons of money and have no path to profitability.”
What we tested: Cognition, Perplexity, Cursor, Lovable, Cohere, Replit and Glean are all losing a lot of money and have no path to profitability.
Assessment: Can’t be checkedSaid
“In reality, venture capital shouldn’t have touched LLMs with a fifteen foot pole, because the margins were obviously, blatantly bad from the very beginning.”
What we tested: Venture capital should not have invested in LLMs because their margins were obviously bad from the start.
Assessment: UnresolvedSaid
“I believe that massive valuation drawdowns are a certainty. There are losses coming.”
What we tested: AI startup valuations will certainly suffer massive drawdowns and investors will take losses.
Assessment: UnresolvedSaid
“By the beginning of 2027, Stargate Abilene will be obsolete, as will any and all data centers filled with Blackwell GPUs, as will any and all data centers being built today.”
What we tested: By early 2027, Stargate Abilene and all data centers filled with Blackwell GPUs or being built today will be obsolete.
Assessment: UnresolvedSaid
“Otherwise, when you remove the contracts signed by hyperscalers and OpenAI (which I do not believe has paid anybody other than Microsoft yet), based on my analysis, there was less than a billion dollars of AI compute revenue in 2025, or 0.5831% of the money spent on data centers.”
What we tested: Excluding hyperscaler and OpenAI contracts, AI compute revenue in 2025 was under $1 billion.
Assessment: Can’t be checkedSaid
“In really simple terms, I believe that almost every investment in a data center or AI startup may go to zero.”
What we tested: Almost every investment in a data center or AI startup may go to zero.
Assessment: Can’t be checkedSaid
“The likelihood of the majority of these projects being A) completed within the next year and B) completed on budget is very, very small.”
What we tested: The likelihood is very, very small that most of the data center projects covered by his modeled ~$171.5 billion of data center debt (about 3.2GW) will be completed both within the next year and on budget.
Assessment: ContradictedSaid
“The market for AI compute is not actually growing. In the last two years, no new major consumers of AI compute have emerged.”
What we tested: The AI compute market is not growing; no new major compute consumers have emerged in two years.
What happened: The study identifies Colossus among systems deployed in 2024 and documents its leading scale by March 2025. This establishes a major standalone consumer beyond OpenAI, Anthropic and the hyperscalers within the claimed window. (arXiv; Konstantin F. Pilz and coauthors, )
Assessment: UnresolvedSaid
“Eventually, reality will dawn on one or more of these banks. Projects will get delayed thanks to weather, or budgetary issues, or when customers walk away (as just happened to data center REIT Fermi). Loan payments will start going unpaid.”
What we tested: Data center projects will get delayed and loan payments will eventually start going unpaid.
Assessment: Can’t be checkedSaid
“It is time to wake up. These are the real-life costs of running an AI company. OpenAI and Anthropic are going to be even worse.”
What we tested: OpenAI's and Anthropic's financials will look even worse than the heavy losses of Zhipu and MiniMax.
Assessment: Can’t be checkedSaid
“Generative AI is not a functional industry, and once the money works that out, everything burns.”
What we tested: Generative AI is not a functional industry, and it will collapse once investors realise this.
Assessment: Can’t be checkedSaid
“Before then, NVIDIA’s bubble is likely to burst.”
What we tested: NVIDIA's bubble is likely to burst before debt and venture funding for AI run out.
Assessment: UnresolvedSaid
“The mistake that every investor, commentator, analyst and member of the media makes about NVIDIA is believing that its sales are an expression of demand for AI compute, when it’s really more of a statement about the availability of debt from banks and private credit.”
What we tested: NVIDIA's sales reflect how much debt is available from banks and private credit more than demand for AI compute.
Assessment: UnresolvedSaid
“Eventually, data centers are going to go unbuilt, and data center debt packages will begin to fall apart.”
What we tested: Data centers will eventually go unbuilt and data center debt packages will begin to fall apart.
Assessment: Can’t be checkedSaid
“LLMs are not good enough for people to pay regular software prices at any scale, and the consequences of this will be that every single dollar spent on GPUs has been for exactly one point: manipulating the value of their stocks.”
What we tested: LLMs are not good enough for people to pay regular software prices at any scale, and as a consequence every single dollar spent on GPUs will turn out to have been for exactly one purpose: manipulating the value of their stocks.
Assessment: Can’t be checkedSaid
“AI does not have the business returns and may have negative gross margins.”
What we tested: AI lacks business returns and may have negative gross margins.
Assessment: Can’t be checkedSaid
“The tech industry has run out of hypergrowth ideas, and in its desperation hitched itself to the least-profitable hardware and software in history, then spent three straight years lying about what was possible to the media, analysts and shareholders.”
What we tested: Out of hypergrowth ideas, the tech industry bet on AI, the least-profitable hardware and software in history, and spent three years lying about what it could do.
Held from this post (28)
These records did not pass every check, so nothing about them is published. Held is not a finding.
- The record does not meet the current rules. Our reading, not assessed: LLMs do not create novel concepts and are inconsistent and unreliable.
- The two reviews read the claim differently
- The two reviews read the claim differently
- The two reviews read the claim differently
- A required check could not run. Our reading, not assessed: The AI industry's financial doom-spiral will not be reversed by the needed turnaround in 2026.
- The two reviews read the claim differently
- The two reviews reached different assessments. Our reading, not assessed: If his thesis is right, OpenAI's death will kill most or all other AI startups, including Anthropic, and every AI investor will take massive losses.
- The criteria are ambiguous. Our reading, not assessed: Because no generative AI company has a path to profitability, the AI bubble is really a stress test of the global venture capital, private equity, private credit, institutional and banking system and its willingness to fund all of this forever.
- The criteria are ambiguous. Our reading, not assessed: Apart from hyperscaler GPU purchases from cashflow, the AI industry runs on subsidy, not revenue.
- The two reviews read the claim differently
- A required check could not run. Our reading, not assessed: OpenAI and Anthropic are draining venture capital and redistributing startup capital to a few public companies.
- The two reviews read the claim differently
- The two reviews read the claim differently
- The criteria are ambiguous. Our reading, not assessed: Every AI app is subsidized and priced below its real cost.
- The record does not meet the current rules. Our reading, not assessed: Every AI company is unprofitable with worsening margins as it scales, so none will be able to get acquired or go public.
- The two reviews read the claim differently
- The record does not meet the current rules. Our reading, not assessed: NVIDIA's revenue from customers outside its few largest buyers is collapsing.
- A required check could not run. Our reading, not assessed: Outside of OpenAI and Anthropic, the AI compute market is very small.
- The two reviews read the claim differently
- The criteria are ambiguous. Our reading, not assessed: None of the current AI companies can survive on their own without venture capital, and if even one of them falters and dies, the entire AI narrative will die, making it harder for AI companies to raise money and even harder to sell an AI company.
- The two reviews read the claim differently
- The two reviews reached different assessments. Our reading, not assessed: Other than NVIDIA's revenue from GPU sales and data center construction, investments in AI do not appear to be driving much economic growth at all.
- The criteria are ambiguous. Our reading, not assessed: The 10GW OpenAI-Broadcom chip deal does not appear to actually be happening.
- The record does not meet the current rules. Our reading, not assessed: There is not enough AI demand for the data centers being built, and most data center loans will be underwater within two years.
- The criteria are ambiguous. Our reading, not assessed: Because generative AI is inherently unprofitable, every single AI company depends on continued venture funding (the 'Silicon Valley Welfare System'), and all signs suggest the current pace of funding must continue in perpetuity until everybody gives up or the system cannot sustain the pressure.
- The two reviews read the claim differently
- The criteria are ambiguous. Our reading, not assessed: Every single major bank and financial institution has put hundreds of millions if not billions of dollars into building data centers that take forever to start generating money and then only seem to lose it.
- The criteria are ambiguous. Our reading, not assessed: Cursor, Perplexity, Cognition and Cohere are unprofitable and have no path to acquisition or IPO.