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

Claims checked against evidence, page 8 of 31

Claims by Gary Marcus and Ed Zitron. Deadlines are judged as of 2026-10-29.

Claims checked against evidence (3100)

Each of these claims was checked against dated evidence.

Page 8 of 31: claims 701 to 800 of 3100, ordered by the date first said, newest first. To find a claim by its words, use search.

  1. Assessment: UnresolvedSaid

    Ed Zitron

    “It is blatantly, alarmingly obvious that Anthropic cannot afford to provide a stable and reliable service to its customers”

  2. Assessment: UnresolvedSaid

    Ed Zitron

    “Even if said numbers were real, the majority of OpenAI, Cursor and Anthropic’s revenues come from subsidized software subscriptions.”

  3. Assessment: UnresolvedSaid

    Ed Zitron

    “NVIDIA Is Selling Years’ Worth of GPUs In Advance And Warehousing Them”

  4. Assessment: UnresolvedSaid

    Ed Zitron

    “And we’re going to find out fairly quickly, because the world of token subsidies is going away.”

  5. Assessment: MixedSaid

    Ed Zitron

    “This does not really make sense given the fact that all support documents and the Claude website reflect that Pro users do not have access to Claude Code.”

    What held: The clarification acknowledges the public documentation problem and subsequently says the changes confused people outside the experiment. This supports the messaging criticism, not a finding that the enrollment percentage was false. (Amol Avasare, Anthropic, )

    What did not: The reporter identifies the Claude Code product page as still advertising Pro access during the episode. Publication was April 22 at 01:16 UTC, still April 21 in US timezones; the observation contradicts uniform website exclusion. (The Register, )

  6. Assessment: SupportedSaid

    Ed Zitron

    “This might sound like a good thing, but it actually means that each request using Opus 4.7 is actually 7.5 of them.”

    What happened: GitHub explicitly announced the 7.5× promotional rate. This establishes the multiplier described in the claim. (GitHub, )

  7. Assessment: ContradictedSaid

    Ed Zitron

    “The standard version of Claude Opus 4.6 has a premium request multiplier of three — meaning that, even with the promotional pricing, Claude Opus 4.7 is around 250% more expensive to use.”

    What happened: The data file records paid-plan multipliers of 3 for standard Opus 4.6 and 7.5 for Opus 4.7. These confirm the baseline but yield a 150% increase, contradicting the claimed 250% increase. (GitHub, )

  8. Assessment: UnresolvedSaid

    Ed Zitron

    “This is a significant move, reflecting the significant cost of running models on any AI product.”

  9. Assessment: UnresolvedSaid

    Ed Zitron

    “The documents reveal that the weekly cost of running Github Copilot has doubled since the start of the year.”

  10. Assessment: UnresolvedSaid

    Gary Marcus

    “And all this from a company thought (or claimed) that GPT-5 was going to be tantamount to AGI (spoiler alert: it wasn’t)”

  11. Assessment: SupportedSaid

    Gary Marcus

    “Today, even after a very strong week, Oracle stands at just $170.”

    What happened: The filing reports the market price of Oracle shares, using Bloomberg historical data. $169.81 rounds to the claimed $170. (BofA Finance LLC; filing copy hosted by Yahoo Finance, )

  12. Assessment: UnresolvedSaid

    Ed Zitron

    “As I’ve said for years, AI companies are inherently unsustainable due to the unreliable and inconsistent outputs of Large Language Models and the incredible costs of providing the services.”

  13. Assessment: UnresolvedSaid

    Ed Zitron

    “On June 1 2026, any article or post about Codex’s efficacy must be rewritten, as rate limits will be halved.”

  14. Assessment: UnresolvedSaid

    Ed Zitron

    “These actions are morally wrong, and are also the direct result of the AI industry’s deceptive and manipulative scare campaign, one promoted by men like Altman and Amodei, as well as doomer fanfiction writers like Yudowsky, and, of course, Daniel Kokotajlo of AI 2027 — both of whom have had their work validated and propagated via the New York Times.”

  15. Assessment: UnresolvedSaid

    Ed Zitron

    “I’m in a research-heavy business, and I have tried the most-powerful LLMs and highest-priced RAG/post-RAG research tools, and every time find them bereft of any unique analysis or suggestions.”

  16. Assessment: UnresolvedSaid

    Ed Zitron

    “It’s also clear, at this point, that Anthropic and OpenAI have both offered subscriptions that were impossible to provide at scale at the price and availability that they were leading up to 2026, and that they did so with the intention of growing their revenue to acquire more customers, equity investment and attention.”

  17. Assessment: ContradictedSaid

    Ed Zitron

    “OpenAI and Anthropic need to show continued revenue growth, which means that they must have capacity available for new customers, which means that old customers will always be the first to be punished.”

    What happened: Anthropic expressly protected existing subscribers while testing a restriction on new signups. The accompanying April 22 reporting establishes the date and identifies the affected service. (Amol Avasare, Anthropic, via X, )

  18. Assessment: UnresolvedSaid

    Ed Zitron

    “And if it doesn’t make that much revenue — and also doesn’t get funding or debt to support it — OpenAI will run out of money, much as Anthropic will if that capacity gets built and it doesn’t make tens of billions of dollars a month to pay for it.”

  19. Assessment: ContradictedSaid

    Ed Zitron

    “Every “AI agent” story is either about code generation, summarizing some sort of information source, or generating something based on an information source that you may or may not be able to trust.”

    What happened: This published AI-agent story describes operating a shop and executing sales, an application beyond content generation on the stated reading. The event date is the story's publication, because the claim concerns which stories existed. (Anthropic, )

  20. Assessment: UnresolvedSaid

    Ed Zitron

    “Every single LLM product is being sold under false pretenses about what’s actually sustainable and possible long term.”

  21. Assessment: UnresolvedSaid

    Ed Zitron

    “Anyone using a subscription to OpenAI or Anthropic’s services needs to wake up and realize that their way of life is going away — that rate limits will make current workflows impossible, that prices will increase, and that the product they’re selling even today is not one that makes any economic sense.”

  22. Assessment: UnresolvedSaid

    Ed Zitron

    “LLMs will always hallucinate, their outputs are not trustworthy as a result, they cannot be deterministic, and any chance of any mistakes of any kind are unforgivable.”

  23. Assessment: ContradictedSaid

    Ed Zitron

    “LLMs will always hallucinate, their outputs are not trustworthy as a result, they cannot be deterministic, and any chance of any mistakes of any kind are unforgivable.”

    What happened: The authors report identical completions for Qwen3-235B-A22B-Instruct-2507 using temperature zero and batch-invariant kernels. The report explains the implementation and supplies a direct counterexample to impossibility. (Thinking Machines Lab, )

  24. Assessment: UnresolvedSaid

    Ed Zitron

    “The con of the Large Language Model era is that any subscription you pay for is massively subsidized, and that any product you use can and will see its service degraded as these companies desperately try to either ease their capacity issues or lower their burn rate.”

  25. Assessment: UnresolvedSaid

    Ed Zitron

    “The so-called “agentic AI” that is supposedly powerful and autonomous is actually incredibly demanding of its human users — you must set it up in so many different ways and connect it to so many different services and check that every “agent” (different chatbot) is instructed in exactly the right way, and that none of these agents cause any problems (they will) with each other.”

  26. Assessment: UnresolvedSaid

    Ed Zitron

    “These conversations are also nonsensical and misleading when you compare them to what Large Language Models can do, and this rhetoric is a blatant attempt to scare people into paying for software today based on what it absolutely cannot and will not do in the future.”

  27. Assessment: UnresolvedSaid

    Ed Zitron

    “In fact, I think that as capacity limits get hit, both Anthropic and OpenAI will degrade the experience for the user (either through model degradation or rate limit decay) as much as they can.”

  28. Assessment: ContradictedSaid

    Ed Zitron

    “I see no scenario where costs come down, or where rate limits are eased.”

    What happened: The section specifies that its changes are effective that day. Doubling included five-hour allowances on existing plans is a direct easing after the post. (Anthropic, )

  29. Assessment: UnresolvedSaid

    Ed Zitron

    “Not only must Anthropic and OpenAI find a way to make their unsustainable and unprofitable services burn less money, but they must also constantly dance with metering out whatever capacity they have to their customers, because the more extra capacity they buy, the more money they lose.”

  30. Assessment: SupportedSaid

    Gary Marcus

    “On the darker side, it really does arm attackers to a greater degree than Mythos’s predecessors.”

    What happened: On the same 32-step attack simulation, Mythos averaged 22 completed steps, versus 16 for the next-best model, Claude Opus 4.6. Its first end-to-end completions establish stronger offensive capability in this evaluation. (UK AI Security Institute, )

  31. Assessment: UnresolvedSaid

    Gary Marcus

    “On the happy side, in its current form, Myth is nowhere near as scary as Tom Fridman (who worries about schoolchildren accidentally taking down power grids) and others made it out to be.”

  32. Assessment: SupportedSaid

    Gary Marcus

    “Second, they show that a neurosymbolic hybrid model, which integrates both a neural networks (for pattern recognition) and a symbolic planner generalizes far better.”

    What happened: The results report 95% versus 34% on three blocks and 78% versus zero for both VLA configurations on the unseen four-block task. (arXiv; Duggan, Lorang, Lu and Scheutz, )

  33. Assessment: SupportedSaid

    Gary Marcus

    “LLMs have persistent weaknesses around planning, reasoning and generalization.”

    What happened: The survey synthesizes recurring reasoning failures and robustness problems across studies while acknowledging substantial capability gains. It supports persistence rather than an assertion that models never reason successfully. (Transactions on Machine Learning Research; arXiv copy, )

  34. Assessment: UnresolvedSaid

    Gary Marcus

    “LLMs are an efficient way to pattern recognition where perfect results are not required, but an inefficient way to reason a plan.”

  35. Assessment: SupportedSaid

    Gary Marcus

    “Third, they show that the neurosymbolic hybrid was vastly more efficent in terms of energy use - by nearly two orders of magnitude.”

    What happened: Table I gives total training energy of 68.5 and 64.9 MJ for the VLAs versus 0.85 MJ for the hybrid: about 81-fold and 76-fold reductions. Table II shows smaller execution savings; its GPU statement concerns execution only. (arXiv; Duggan, Lorang, Lu and Scheutz, )

  36. Assessment: UnresolvedSaid

    Gary Marcus

    “You can read the full paper by Timothy Duggan, Pierrick Lorang, Hong Lu, and Matthias Scheutz, which very much vindicates the Apple reasoning paper, here.”

  37. Assessment: UnresolvedSaid

    Gary Marcus

    “But anybody who thinks LLMs are a direct route to the sort AGI that could fundamentally transform society for the good is kidding themselves.”

  38. Assessment: SupportedSaid

    Gary Marcus

    “Claude Code (in particular version 4.6), which does seem genuinely impressive to me and a number of friends I trust, still makes plenty of mistakes, and still can’t be trusted.”

    What happened: The study uses Claude Code with Opus 4.6 and finds multiple final submissions that fail its correctness tests. It also reports strong performance on easy tasks and notes that many failures are near-complete solutions, limiting how broadly the failure rate can be generalized. (Zhang et al. / arXiv, )

  39. Assessment: UnresolvedSaid

    Gary Marcus

    “It’s obviously not AGI.”

  40. Assessment: SupportedSaid

    Gary Marcus

    “Still, Claude Code ain’t perfect, or even close.”

    What happened: Anthropic traces recurring problems to a change shipped March 26 that repeatedly discarded reasoning history after an idle-session threshold. This documents a material product limitation during the target month. (Anthropic, )

  41. Assessment: UnresolvedSaid

    Gary Marcus

    “Instead, the way Anthropic built that kernel is straight out of classical symbolic AI. For example, it’s in large part a big IF-THEN conditional, with 486 branch points and 12 levels of nesting — all inside a deterministic, symbolic loop that the real godfathers of AI, people like John McCarthy and Marvin Minsky and Herb Simon, would have instantly recognized.”

  42. Assessment: SupportedSaid

    Gary Marcus

    “Putting things differently, Anthropic, when push came to shove, went exactly where I have said for 25 years that the field needed to go: to Neurosymbolic AI.”

    What happened: The study traces requests from the language model through an explicit permission-rule engine before tool execution. This supports the combination Marcus describes under his broad definition; the neurosymbolic label is the evaluator's classification under that definition. (arXiv; Jiacheng Liu, Xiaohan Zhao, Xinyi Shang, and Zhiqiang Shen, )

  43. Assessment: UnresolvedSaid

    Gary Marcus

    “The source code leak proves it. Tucked away at its center is a 3,167 line kernel called print.ts.”

  44. Assessment: SupportedSaid

    Gary Marcus

    “And the thing is, Claude Code is NOT a pure LLM. And it’s not pure deep learning. Not even close.”

    What happened: The study examines the leaked implementation and traces model calls, permission checks, executable tools, and returned results. These are substantive application operations beyond neural inference. (arXiv; Jiacheng Liu, Xiaohan Zhao, Xinyi Shang, and Zhiqiang Shen, )

  45. Assessment: UnresolvedSaid

    Gary Marcus

    “The first part has already come to pass. In time, other three will, too.”

  46. Assessment: UnresolvedSaid

    Gary Marcus

    “Claude Code isn’t better because of scaling. It’s better because it is neurosymbolic.”

  47. Assessment: UnresolvedSaid

    Gary Marcus

    “The demo was definitely proof of concept that we need to get our regulatory and technical house in order, but not the immediate threat the media and public was lead to believe.”

  48. Assessment: UnresolvedSaid

    Gary Marcus

    “The model itself is incrementally better than previous recent models, but certainly not an off-the-chart breakthrough:”

  49. Assessment: UnresolvedSaid

    Gary Marcus

    “ChatGPT can’t reliably run a timer but it has still been implicated in delusions, suicides, cognitive surrender, mass disinformation, and so much more.”

  50. Assessment: UnresolvedSaid

    Ed Zitron

    “AI progress isn’t hampered by a lack of access, talent, resources, novel approaches, or industry buy-in, but by a single-minded focus on Large Language Models, a technology that has been so obviously-limited from the very beginning that Gary Marcus was able to call it in 2022.”

  51. Assessment: UnresolvedSaid

    Ed Zitron

    “Said data centers make up 90% of NVIDIA’s revenue, which means that 8% or so of the S&P 500’s value comes from a company that makes money selling hardware to people that immediately lose money on installing it.”

  52. Assessment: UnresolvedSaid

    Ed Zitron

    “As I discussed a few weeks ago, despite everybody talking about the hundreds of gigawatts of data centers being built “to power AI,” only 5GW are actually “under construction,” with “under construction” meaning anything from “we’ve got some scaffolding up” to “we’re about to hand over the keys to the customer.””

  53. Assessment: UnresolvedSaid

    Ed Zitron

    “As I wrote a few weeks ago, LLMs are good at writing a lot of code, not good code, and the more people you allow to use them, the more code you’re going to generate, which means the more time you’re either going to need to review that code, or the more vulnerabilities you’re going to create as a result.”

  54. Assessment: UnresolvedSaid

    Ed Zitron

    “That “growth” is only possible because all of the companies in question heavily subsidize their products, spending $3 to $15 for every dollar of revenue.”

  55. Assessment: UnresolvedSaid

    Ed Zitron

    “Back in November 2025, I reported that OpenAI had made — based on its revenue share with Microsoft — $4.329 billion between January and September 2025, despite The Information reporting that it had made $4.3 billion in the first half of the year based on disclosures to shareholders.”

  56. Assessment: SupportedSaid

    Ed Zitron

    “The above quote has been used repeatedly to suggest that Anthropic has 50% gross margins and is “profitable,” which is extremely weird in and of itself as that’s not what Dario Amodei said at all.”

    What happened: Amodei explicitly separates the illustrative numbers from information about Anthropic. The surrounding discussion frames the calculation as hypothetical industry economics rather than a current company financial disclosure. (Dwarkesh Podcast, )

  57. Assessment: ContradictedSaid

    Ed Zitron

    “The revenues are terrible, Large Language Models are yet to provide any meaningful productivity improvements, and the only reason that they’ve been able to get as far as they have is a compliant media and a venture capital environment borne of a lack of anything else to invest in.”

    What happened: The study examines an LLM-based assistant in actual customer-support work, with rollout extending into winter 2021. The estimated gain is material under the ordinary workplace reading, although it is not an economy-wide estimate. (The Quarterly Journal of Economics; author-hosted copy, )

  58. Assessment: SupportedSaid

    Ed Zitron

    “To be clear, LLMs can absolutely write code, and can absolutely create software, but neither of those mean that the code is good, stable or secure, or that the same can be said of the software they create.”

    What happened: Anthropic's generated compiler demonstrates substantive software-generation capability while retaining functional limitations. That directly fits the distinction in Zitron's sentence. (Anthropic, )

  59. Assessment: UnresolvedSaid

    Ed Zitron

    “I have confirmed with a source at that there is no actual metric or tracking of any return on investment involved in token burn at Meta, meaning that TokenMaxxing’s only purpose is to burn more tokens to go higher on a leaderboard, and is already creating bad habits across a company that already has decaying products and leadership.”

  60. Assessment: ContradictedSaid

    Ed Zitron

    “If that’s the case, I truly have no idea how it could’ve possibly accelerated so aggressively, and as I’ve mentioned before, there is no way to reconcile having made $5 billion in lifetime revenue as of March 9, 2026, having $14 billion in annualized revenue on February 12 2026, and having $4.5 billion in revenue for the year 2025.”

    What happened: The sentence immediately preceding the quotation says cumulative revenue exceeded $5 billion; it supplies no upper bound. This permits additional revenue beyond $5 billion and defeats the asserted incompatibility of the reported figures. (U.S. District Court, Northern District of California / CourtListener RECAP, )

  61. Assessment: UnresolvedSaid

    Ed Zitron

    “Coding LLMs are popular only because of their massive subsidies and corporate encouragement, and in the end will be seen as a useful-yet-incremental and way too expensive way to make the easy things easier and the harder things harder, all while filling codebases full of masses of unintentional, bloated code.”

  62. Assessment: UnresolvedSaid

    Ed Zitron

    “In simpler terms, Anthropic is cherry-picking four-week windows of API spend — ones that are pumped by big announcements and new model releases — and annualizing them.”

  63. Assessment: UnresolvedSaid

    Ed Zitron

    “Worse still, LLMs allow shitty software engineers that would otherwise be isolated by their incompetence to feign enough intelligence to get by, leading to them actively lowering the quality of code being shipped.”

  64. Assessment: UnresolvedSaid

    Ed Zitron

    “It’s unclear what the actual economic or productivity effects are, other than an abundance of new code that’s making running companies harder.”

  65. Assessment: UnresolvedSaid

    Ed Zitron

    “Combined with mandates to “use as much AI as possible,” this naturally increases the cost of having software engineers, which — and I say this not wanting anyone to lose their jobs — does the literal opposite of replacing workers with AI.”

  66. Assessment: UnresolvedSaid

    Ed Zitron

    “That “fast-growing” part is only possible because both Anthropic and OpenAI subsidize the compute of their subscribers, allowing them to burn $3 to $15 for every dollar of subscription revenue.”

  67. Assessment: UnresolvedSaid

    Ed Zitron

    “At this point, I estimate Anthropic’s revenue split to be more in the region of 75% API and 25% subscriptions, based on its supposed $2.5 billion in annualized revenue (out of $14 billion, so a little under 18%) in February coming from “Claude Code” (read: subscribers to Claude, there’s no “Claude Code” subscription).”

  68. Assessment: UnresolvedSaid

    Ed Zitron

    “This leads to context bloat that absolutely rinses your token budget.”

  69. Assessment: UnresolvedSaid

    Ed Zitron

    “Furthermore, TokenMaxxing is exactly the kind of revenue that disappears first.”

  70. Assessment: UnresolvedSaid

    Ed Zitron

    “It’s also weird that people discussing agents don’t seem to want to discuss that OpenAI’s Operator Agent does not work, that AI browsers are fundamentally broken, or that agentic AI does not do anything that people discuss.”

  71. Assessment: UnresolvedSaid

    Ed Zitron

    “Stargate was a lie! The whole time it was a lie! That time that Sam Altman and Masayoshi Son and Larry Ellison stood up at the white house and talked about a $500 billion infrastructure project was a lie! They never formed the entity!”

  72. Assessment: UnresolvedSaid

    Ed Zitron

    “And with that incredibly easy access, only 3% of households pay for AI.”

  73. Assessment: SupportedSaid

    Gary Marcus

    “Right now he is in trouble. People are increasingly concerned about the economics of his company,”

    What happened: The report describes growing caution about costs and declining secondary-market demand, drawing on named marketplace operators. This supports a change in investor sentiment, though the secondary market is not representative of all investors. (Los Angeles Times / Bloomberg, )

  74. Assessment: SupportedSaid

    Ed Zitron

    “Anyway, The Information's piece also adds another fun detail - that OpenAI's margins were even worse than expected in 2025:”

    What happened: The report compares a 33 percent actual adjusted gross margin with a 46 percent target for 2025. This directly supports the claimed shortfall on the company-defined investor-reporting measure. (The Decoder, )

  75. Assessment: SupportedSaid

    Gary Marcus

    “First, Microsoft’s Mustafa Suleyman just tried to redefine superintelligence down from AI that is smarter than the smartest humans to something a lot more mundane: AI “models capable of delivering product value for millions of enterprises”.”

    What happened: The original interview immediately attributes to Suleyman an explanation of superintelligence in terms of models supplying enterprise product value. This supports Marcus’s reading as public reframing, although it does not establish a formal technical change. (The Verge, )

  76. Assessment: SupportedSaid

    Gary Marcus

    “Third, there is the well-known tendency of generative AI to be sycophantic; to tell users that their ideas were the greatest ever, even when they are not.”

    What happened: The dated 199-case snapshot tests agreement with opposing narrators while preserving the underlying dispute, finding sycophantic contradictions across all 16 configurations. Rates vary substantially, from 0.5% to 31.2%, so the evidence supports a recurring tendency rather than pervasive agreement in every model. (Lech Mazur, )

  77. Assessment: UnresolvedSaid

    Gary Marcus

    “First, strategy, more than e.g., writing requisition memos, requires a broad, deep understanding of the world, but generative AI has never had robust models of the world.”

  78. Assessment: UnresolvedSaid

    Gary Marcus

    “AI is fine for writing memos; nobody should rely on it to plan a war or to guess its likely outcomes.”

  79. Assessment: MixedSaid

    Gary Marcus

    “Every company is investing in AI, but so far most aren’t getting huge returns.”

    What held: A majority reporting minimal financial gains despite substantial investment supports the ordinary reading of the majority-returns clause. The same article describes a successful minority, so this does not imply that large returns were absent everywhere. (Boston Consulting Group, )

    What did not: Actual investment was far from universal among surveyed firms, including large firms. This directly contradicts the literal every-company clause without relying on adoption as a spending proxy. (Federal Reserve Bank of Atlanta, )

  80. Assessment: UnresolvedSaid

    Ed Zitron

    “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.”

  81. Assessment: UnresolvedSaid

    Ed Zitron

    “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.”

  82. Assessment: UnresolvedSaid

    Ed Zitron

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

  83. Assessment: UnresolvedSaid

    Ed Zitron

    “This puts SoftBank in a very precarious position.”

  84. Assessment: UnresolvedSaid

    Ed Zitron

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

  85. Assessment: UnresolvedSaid

    Ed Zitron

    “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.”

  86. Assessment: UnresolvedSaid

    Ed Zitron

    “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.”

  87. Assessment: UnresolvedSaid

    Ed Zitron

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

  88. Assessment: UnresolvedSaid

    Ed Zitron

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

  89. Assessment: UnresolvedSaid

    Ed Zitron

    “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.”

  90. Assessment: UnresolvedSaid

    Ed Zitron

    “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.”

  91. Assessment: UnresolvedSaid

    Ed Zitron

    “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.”

  92. Assessment: UnresolvedSaid

    Ed Zitron

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

  93. Assessment: UnresolvedSaid

    Ed Zitron

    “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.”

  94. Assessment: UnresolvedSaid

    Ed Zitron

    “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.”

  95. Assessment: UnresolvedSaid

    Ed Zitron

    “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.”

  96. Assessment: UnresolvedSaid

    Ed Zitron

    “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.”

  97. Assessment: UnresolvedSaid

    Ed Zitron

    “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.”

  98. Assessment: SupportedSaid

    Ed Zitron

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

    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, )

  99. Assessment: UnresolvedSaid

    Ed Zitron

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

  100. Assessment: MixedSaid

    Ed Zitron

    “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 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, )