Claims checked against evidence by Ed Zitron, page 8 of 11
Deadlines are judged as of 2026-10-29. About Ed Zitron and the sources covered.
Claims checked against evidence by Ed Zitron (1044)
Each of these claims was checked against dated evidence.
Page 8 of 11: claims 701 to 800 of 1044, ordered by the date first said, newest first. To find a claim by its words, use search.
Assessment: UnresolvedSaid
Observation
“They can’t, because it isn’t happening at scale, because generative AI does not have the capabilities that people like Dario Amodei and Sam Altman repeatedly act like they do”
Our reading: Businesses replacing workers with generative AI is not happening at scale.
No horizon Evidence as of
Limitation: The target month is an evaluator interpretation.
Source: Sincerity Wins The War
Assessment: UnresolvedSaid
Causal claim
“generative AI does not have the capabilities that people like Dario Amodei and Sam Altman repeatedly act like they do”
Our reading: Generative AI lacks the capabilities Dario Amodei and Sam Altman claim, which is why it is not replacing workers at scale.
No horizon Evidence as of
Limitation: The earlier benchmark does not establish the capabilities of systems available in May 2025 or connect its failures to staffing decisions [e5].
Source: Sincerity Wins The War
Assessment: UnresolvedSaid
Causal claim
“Generative AI chatbots are driving people insane by providing them an endlessly-configurable pseudo-conversation too, though that’s less of a “use case” and more of a “text-based video game launched at scale without anybody thinking about what might happen.””
Our reading: Generative AI chatbots are driving people insane by giving them an endlessly configurable pseudo-conversation.
No horizon Evidence as of
Limitation: Insane is interpreted as serious psychiatric deterioration, not ordinary distress or unusual beliefs.
Source: Sincerity Wins The War
Assessment: UnresolvedSaid
Observation
“and then failed because there is no proof that this is actually happening other than that which the media drums up.”
Our reading: There is no proof, other than what the media drums up, that what Roose's piece describes about AI and entry-level work is actually happening.
No horizon Evidence as of
Limitation: Hiring declines alone do not establish premature AI substitution or reductions in entry-level development caused by expectations of AI displacement [e4].
Source: Sincerity Wins The War
Assessment: SupportedSaid
Observation
“Software engineers using LLM tools — as I talked about on a recent episode of Better Offline — are finding some advantages, but LLMs are far from a panacea.”
Our reading: Software engineers get some advantages from LLM tools, but LLMs are far from a panacea.
No horizon Evidence as of
Limitation: The decisive evidence measures reported experience rather than independently timed productivity.
Source: Sincerity Wins The War
Assessment: UnresolvedSaid
Observation
“One that you can just ask to do some shit and know it’ll do it, and it’d also be very obvious what it could actually do, which is not the case right now. A better ChatGPT would quite literally be a different product.”
Our reading: ChatGPT currently cannot be relied on to do what it's asked and its capabilities are unclear; a better ChatGPT would be a different product.
No horizon Evidence as of
Limitation: Evidence about one feature cannot establish the reliability of every ChatGPT model and use.
Source: Never Forget What They've Done
Assessment: UnresolvedSaid
Observation
“LLMs are famous not for their efficacy, but their inconsistency, with even ardent AI cultists warning people not to trust their output.”
Our reading: LLMs are famous for their inconsistency rather than their efficacy, and even ardent AI supporters warn people not to trust their output.
No horizon Evidence as of
Limitation: The speaker gives no population or reputation measure.
Source: Never Forget What They've Done
Assessment: ContradictedSaid
Observation
“At this point, no real information exists.”
Our reading: As of May 2025, no real information exists about the AI device OpenAI and Jony Ive are developing.
No horizon Evidence as of
Limitation: The accessible evidence is secondary reporting; the original Journal article and underlying recording were not inspected directly.
Assessment: UnresolvedSaid
Observation
“OpenAI’s operating costs are insane, and the company has already committed to several grand projects, while also pushing deeper and deeper into the red.”
Our reading: OpenAI's operating costs are insane and its losses keep growing even as it commits to grand projects.
No horizon Evidence as of
Limitation: No comparable series of realized losses and operating expenses through March 31, 2025 is provided.
Assessment: ContradictedSaid
Observation
“Despite all that media and investor attention — despite effectively the entirety of the tech industry focusing on this one specific thing — we're still yet to get any real consumer product.”
Our reading: Despite the media and investor attention and effectively the entire tech industry focusing on large language models, there is still no real consumer product from them other than ChatGPT.
No horizon Evidence as of
Limitation: Both sources are Google disclosures; the usage figure is not independently verified in this packet.
Assessment: ContradictedSaid
Observation
“We're three joyless years into this crap, and there isn't a single consumer product of note other than ChatGPT, a product that gained its momentum through a hype campaign driven by press and markets that barely understood what they were hyping.”
Our reading: After three years of generative AI, there is no notable consumer LLM product other than ChatGPT, whose momentum came from hype by press and markets that barely understood it.
No horizon Evidence as of
Limitation: The result uses adoption as the ordinary meaning of of note; the speaker supplied no threshold.
Assessment: UnresolvedSaid
Observation
“On top of that, it requires more money than has ever been invested in a startup.”
Our reading: OpenAI needs more money than has ever been invested in a startup.
No horizon Evidence as of
Limitation: A record individual funding round does not establish a record for cumulative startup investment.
Assessment: UnresolvedSaid
Observation
“It has no moat, its models are hitting the point of diminishing returns and have been for some time, and as popular as ChatGPT may be, it isn't a business and constantly loses money.”
Our reading: OpenAI has no moat.
No horizon Evidence as of
Limitation: The evidence is contextual rather than a direct measurement of OpenAI's durable protection in the target period.
Assessment: UnresolvedSaid
Observation
“It has no moat, its models are hitting the point of diminishing returns and have been for some time, and as popular as ChatGPT may be, it isn't a business and constantly loses money.”
Our reading: OpenAI's models are hitting diminishing returns and have been for some time.
No horizon Evidence as of
Limitation: Company-reported cost comparisons are not independent measurements of total development resources.
Assessment: UnresolvedSaid
Observation
“It has no moat, its models are hitting the point of diminishing returns and have been for some time, and as popular as ChatGPT may be, it isn't a business and constantly loses money.”
Our reading: ChatGPT, however popular, isn't a business and constantly loses money.
No horizon Evidence as of
Limitation: The evidence concerns a different company scope, product subset, or reporting period.
Assessment: UnresolvedSaid
Causal claim
“Every single tech bubble followed the same path, and that path was paved with flawed, deferential and specious journalism, from small blogs to the biggest mastheads.”
Our reading: Every tech bubble, including AI, was enabled by flawed and deferential journalism.
No horizon Evidence as of
Limitation: The search did not produce comprehensive causal evidence for all named episodes.
Source: The Era Of The Business Idiot
Assessment: SupportedSaid
Observation
“None of these tasks are things that require you to use AI.”
Our reading: Tasks like summarizing messages, discussing podcasts, meeting prep and research do not require AI.
No horizon Evidence as of
Limitation: The examples do not reproduce Nadella's workload or commute.
Source: The Era Of The Business Idiot
Assessment: UnresolvedSaid
Causal claim
“And I'd argue that ChatGPT filled that void by doing the thing that people wanted Google Search to do: answer a question, even if the user isn't really sure how to ask it.”
Our reading: Zitron argues that ChatGPT filled the void left by Google Search by doing what people wanted Google Search to do: answer a question even when the user isn't sure how to ask it.
No horizon Evidence as of
Limitation: The experiment used a small sample and replicated tool interfaces.
Source: Reality Check
Assessment: ContradictedSaid
Forecast
“It also, at this point, is pretty obvious that generative AI isn't going to do much more than it does today.”
Our reading: Generative AI will not become much more capable than it is today.
Settled before its deadline Evidence as of
Limitation: The results do not establish gains across all practical applications or demonstrate general intelligence.
Source: Reality Check
Assessment: ContradictedSaid
Observation
“The expectation that generative AI will transform into something much, much more powerful requires you to first ignore the existing limitations, believing it to be more capable than it is, and also ignore the fact that these models have yet to show meaningful improvement over the past few years.”
Our reading: Generative AI models have not shown meaningful improvement over the past few years.
No horizon Evidence as of
Limitation: The result depends on interpreting 'meaningful improvement' as substantial capability gains.
Source: Reality Check
Assessment: UnresolvedSaid
Observation
“ChatGPT is, if I'm honest, better at processing search strings than Google Search, which is not so much a sign that ChatGPT is good at something as it is that Google has stopped innovating in any meaningful way.”
Our reading: ChatGPT is better at processing search strings than Google Search, which is less a sign that ChatGPT is good than a sign that Google has stopped innovating in any meaningful way.
No horizon Evidence as of
Limitation: Older product comparisons cannot determine contemporaneous performance.
Source: Reality Check
Assessment: UnresolvedSaid
Observation
“As I wrote before, OpenAI effectively is the generative AI industry, and nothing about the rest of the generative AI industry suggests that the revenue exists to sustain these ridiculous, obscene and fantastical projections.”
Our reading: OpenAI effectively is the generative AI industry, and nothing about the rest of the generative AI industry suggests that the revenue exists to sustain OpenAI's revenue projections.
No horizon Evidence as of
Limitation: Query volume is not user count or revenue.
Source: Reality Check
Assessment: UnresolvedSaid
Observation
“Agents do not exist as a product that can be sold at that scale.”
Our reading: AI agents do not exist as a product that can be sold at that scale, meaning the $3 billion, mass-market sales level just discussed.
No horizon Evidence as of
Limitation: The statement combines product readiness, sales scale, and mass-market demand without defining their relationship.
Source: Reality Check
Assessment: UnresolvedSaid
Observation
“Yes, I fundamentally believe that 500 million people a week could be using ChatGPT as some sort of search replacement, and no, I do not believe that's a functional business model, in part because if it was, ChatGPT would've been a functional business.”
Our reading: Using ChatGPT as a search replacement is not a functional business model.
No horizon Evidence as of
Limitation: The evidence does not measure search-replacement usage or its attributable revenue and operating costs.
Source: Reality Check
Assessment: UnresolvedSaid
Conditional forecast
“SearchGPT will be significantly more expensive to run at Google's scale than ChatGPT — both infrastructurally and in the cost of revenue, with OpenAI forced to create a massive advertising arm that currently doesn't exist at the company.”
Our reading: If run at Google's scale, SearchGPT will be significantly more expensive to run than ChatGPT, both infrastructurally and in cost of revenue, and OpenAI will be forced to create a massive advertising arm that it does not currently have.
No deadline stated Evidence as of
Only if: if SearchGPT runs at Google's scale
Source: Reality Check
Assessment: UnresolvedSaid
Observation
“There seems to be little concern — or belief — that there is any kind of risk at the heart of OpenAI, a company that spent $9 billion in 2024 to lose $5 billion.”
Our reading: OpenAI spent $9 billion in 2024 and lost $5 billion.
No horizon Evidence as of
Limitation: The evidence does not establish completed-year spending on the basis underlying the $9 billion estimate.
Source: Reality Check
Assessment: UnresolvedSaid
Forecast
“OpenAI requires more compute resources than anyone has ever needed, and will continue to do so in perpetuity.”
Our reading: OpenAI requires more compute than anyone has ever needed and will keep doing so indefinitely.
No deadline stated Evidence as of
Limitation: The ranking depends on the meaning of requirements and the organizational boundary.
Assessment: UnresolvedSaid
Conditional forecast
“In the event that OpenAI cannot fulfil its obligations, CoreWeave will collapse. It is that simple.”
Our reading: If OpenAI cannot meet its obligations, CoreWeave will collapse.
No deadline stated Evidence as of
Only if: if OpenAI cannot fulfil its obligations
Assessment: UnresolvedSaid
Observation
“It's also important to note that absolutely nobody other than NVIDIA is making any money from generative AI. CoreWeave loses billions of dollars, OpenAI loses billions of dollars, Anthropic loses billions of dollars, and I can't find a single company providing generative AI-powered software that's making a profit.”
Our reading: No one except NVIDIA makes money from generative AI; OpenAI, Anthropic and CoreWeave lose billions.
No horizon Evidence as of
Limitation: This is not a finding that NVIDIA was the only profitable provider.
Assessment: ContradictedSaid
Observation
“OpenAI, as I've written before, is effectively the entire generative AI industry, with its nearest competitor being less than five percent of its 500 million weekly active users.”
Our reading: OpenAI is effectively the entire generative AI industry; its nearest competitor has under 5% of its weekly active users.
No horizon Evidence as of
Limitation: Uses reputable reporting of company evidence presented in court because the primary exhibit was not obtained.
Assessment: UnresolvedSaid
Forecast
“I expect the latter cost to balloon with OpenAI's investment in physical infrastructure for Stargate.”
Our reading: OpenAI's owned-infrastructure costs will balloon due to its Stargate investment.
No deadline stated Evidence as of
Limitation: The magnitude required by 'balloon' is unspecified.
Assessment: UnresolvedSaid
Conditional forecast
“If OpenAI fails, then CoreWeave fails, which then hurts NVIDIA.”
Our reading: If OpenAI fails, CoreWeave will fail, which will hurt NVIDIA.
No deadline stated Evidence as of
Only if: if OpenAI fails
Assessment: SupportedSaid
Causal claim
“SoftBank's exposure to OpenAI is materially harming the company.”
Our reading: SoftBank's exposure to OpenAI is materially harming SoftBank.
No horizon Evidence as of
Limitation: JCR affirmed the ratings; the adverse action concerned the outlook [e1].
Assessment: UnresolvedSaid
Observation
“As a reminder: OpenAI loses money on every single user.”
Our reading: OpenAI loses money on every single user.
No horizon Evidence as of
Limitation: News reporting substitutes for unavailable underlying financial records.
Assessment: ContradictedSaid
Observation
“OpenAI has become a load-bearing company for the tech industry, both as a narrative — as previously discussed, ChatGPT is the only Large Language Model company with any meaningful userbase — and as a financial entity.”
Our reading: OpenAI is a load-bearing company for the tech industry, narratively and financially, and ChatGPT is the only LLM product with a meaningful userbase.
No horizon Evidence as of
Limitation: Meaningful is interpreted as substantial recurring active use.
Assessment: UnresolvedSaid
Causal claim
“OpenAI has never been forced to stand on its own two feet or focus on efficiency, and I believe the constant enabling of its ugly, nonsensical burnrate has doomed this company.”
Our reading: Constant enabling of OpenAI's burn rate has doomed the company.
No horizon Evidence as of
Limitation: Public financial information is incomplete.
Assessment: UnresolvedSaid
Conditional forecast
“OpenAI has, up until this point, had fairly impressive uptime. Still, if it is running up against a wall, this streak will end.”
Our reading: If OpenAI is hitting capacity limits, its streak of good uptime will end with service degradation.
No deadline stated Evidence as of
Only if: if it is running up against capacity limits
Assessment: UnresolvedSaid
Conditional forecast
“In any case, assuming the deal goes as planned, CoreWeave will still burn $15 billion in 2025.”
Our reading: Even if the OpenAI deal goes as planned, CoreWeave will burn $15 billion in 2025.
Deadline passed Evidence as of
Only if: assuming the OpenAI deal goes as planned
Assessment: UnresolvedSaid
Conditional forecast
“For this situation to work even for the next year, these companies will have to raise tens of billions of dollars just to maintain the status quo.”
Our reading: For the situation to work even for the next year, CoreWeave, OpenAI and SoftBank will have to raise tens of billions of dollars just to maintain the status quo.
Deadline passed Evidence as of
Only if: for this situation to work even for the next year
Assessment: SupportedSaid
Observation
“That’s no discussion of the massive, world-changing consequences of a (again, totally theoretical, no proof this exists) artificial intelligence that's as smart and capable (again, how is it capable?) as a human being.”
Our reading: Human-level AGI is totally theoretical and there is no proof it exists.
No horizon Evidence as of
Limitation: There is no universally accepted AGI test.
Assessment: UnresolvedSaid
Causal claim
“which means that CoreWeave's continued ability to pay its bills is both contingent upon and absorbed by its revenue from OpenAI.”
Our reading: CoreWeave's ability to keep paying its bills depends on, and is consumed by, its revenue from OpenAI.
No horizon Evidence as of
Limitation: Later financing cannot by itself establish what alternatives were reasonably available when the claim was made.
Assessment: UnresolvedSaid
Conditional forecast
“SoftBank is now both CoreWeave and OpenAI, and if SoftBank buckles, both CoreWeave and OpenAI are dead.”
Our reading: If SoftBank buckles, both CoreWeave and OpenAI will collapse.
No deadline stated Evidence as of
Only if: if SoftBank buckles
Assessment: ContradictedSaid
Observation
“I can find no way to explain this other than that Microsoft has decided it doesn't want to be OpenAI's compute provider any further, and that any further expansion that OpenAI needs will be its responsibility to acquire.”
Our reading: Zitron can find no explanation other than that Microsoft has decided it does not want to be OpenAI's compute provider any further and that any further expansion OpenAI needs will be OpenAI's responsibility to acquire.
No horizon Evidence as of
Limitation: This grades the asserted supply decision, not the speaker's private reasoning.
Assessment: UnresolvedSaid
Conditional forecast
“The ramifications of the end of hyper-growth mean a massive reckoning for the valuations of tech companies, which will lead to tens of thousands of layoffs and a prolonged depression in Silicon Valley, the likes of which we've never seen.”
Our reading: The end of tech hyper-growth will cause a massive valuation reckoning, tens of thousands of layoffs, and an unprecedented prolonged depression in Silicon Valley.
No deadline stated Evidence as of
Only if: the end of hyper-growth
Assessment: ContradictedSaid
Forecast
“I estimate OpenAI will lose as much as $14 billion in 2025.”
Our reading: OpenAI will lose as much as $14 billion in 2025.
Deadline passed Evidence as of
Limitation: The speaker did not define loss or specify accounting adjustments.
Assessment: UnresolvedSaid
Observation
“I’m wary of putting too much stock into them, because they’re easily gamed, and quite often, they’re not an effective way of showing actual progress.”
Our reading: AI benchmarks are easily gamed and often fail to show actual progress.
No horizon Evidence as of
Limitation: The earlier contamination studies are context, not period-matched outcome evidence.
Assessment: UnresolvedSaid
Forecast
“CoreWeave is set to lose $15 billion this year. Its projected revenue, according to The Information, is only $4.6 billion.”
Our reading: CoreWeave is set to lose $15 billion in 2025 on projected revenue of only $4.6 billion.
Deadline passed Evidence as of
Limitation: This is an unresolved measurement comparison, not an unelapsed forecast.
Assessment: UnresolvedSaid
Causal claim
“OpenAI's ability to pay CoreWeave is contingent on its ability to continue raising money, as it is set to lose over $14 billion in 2025.”
Our reading: OpenAI's ability to pay CoreWeave depends on its ability to keep raising money, because it is set to lose over $14 billion in 2025.
No horizon Evidence as of
Limitation: The usable financial evidence is news reporting, not directly inspected OpenAI financial statements.
Assessment: UnresolvedSaid
Observation
“It does not anticipate being profitable until 2030, and does not have any explanation as to how they will become so other than “we will build Stargate.””
Our reading: OpenAI does not expect profit until 2030 and has no explanation of how it will get there beyond building Stargate.
No horizon Evidence as of
Limitation: The absence of an explanation in a short article does not establish that no explanation existed.
Assessment: SupportedSaid
Observation
“In effect, CoreWeave spends $1.43 to make $1.”
Our reading: In 2024 CoreWeave spent about $1.43 for every $1 of revenue.
No horizon Evidence as of
Limitation: The exact arithmetic gives $1.45, not $1.43.
Source: CoreWeave Is A Time Bomb
Assessment: UnresolvedSaid
Observation
“Cloud GPUs are incredibly commoditized, and the core factors of differentiation between the various players are price, availability, and the exact hardware available.”
Our reading: Cloud GPU compute is highly commoditized, differentiated mainly by price, availability and hardware.
No horizon Evidence as of
Limitation: The cited evidence consists of provider disclosures.
Source: CoreWeave Is A Time Bomb
Assessment: MixedSaid
Observation
“Converting cryptocurrency mining data centers to HPC data centers is effectively starting from scratch. There is no easy or logical way to repurpose a bitcoin miner for AI.”
Our reading: Bitcoin mining infrastructure cannot easily be repurposed for AI; conversion is effectively starting from scratch.
No horizon Evidence as of
Limitation: This is a split assessment of hardware replacement and facility reuse.
Source: CoreWeave Is A Time Bomb
Assessment: SupportedSaid
Observation
“Worse, despite making $1.9 billion in revenue during the 2024 financial year, the company lost $863 million in 2024, with its entire future riding on "explosive growth" that may not materialize, and even if it does, would require CoreWeave to spend unfathomable amounts of money on the necessary capex investments.”
Our reading: CoreWeave lost $863 million in 2024 despite $1.9 billion in revenue, and its entire future rides on "explosive growth" that may not materialize and that, even if it does, would require CoreWeave to spend unfathomable amounts on capex.
No horizon Evidence as of
Limitation: Support is limited to the historical figures; the evidence does not independently establish the unquantified judgments about growth dependence or required capital spending.
Source: CoreWeave Is A Time Bomb
Assessment: UnresolvedSaid
Observation
“While it may sell by-the-hour compute to regular companies, it's clear that that market just doesn't exist at a meaningful revenue point.”
Our reading: The market for by-the-hour AI compute sold to regular companies does not exist at meaningful revenue.
No horizon Evidence as of
Limitation: A small percentage of one provider's revenue can still be substantial in dollars.
Source: CoreWeave Is A Time Bomb
Assessment: UnresolvedSaid
Observation
“CoreWeave's future is one that requires an industry that has yet to show any meaningful product-market fit to grow so significantly that compute companies turn into oil companies”
Our reading: The generative AI industry has yet to show meaningful product-market fit, yet CoreWeave's future requires it to grow enormously.
No horizon Evidence as of
Limitation: The retention evidence predates the target quarter.
Source: CoreWeave Is A Time Bomb
Assessment: SupportedSaid
Observation
“Regardless of the reason, the company selling the literal fuel of the supposed AI revolution is losing hundreds of millions of dollars doing so.”
Our reading: CoreWeave is losing hundreds of millions of dollars selling AI compute.
No horizon Evidence as of
Limitation: The decisive quotation is reputable news reporting of the filing because the contemporaneous primary documents could not be read successfully.
Source: CoreWeave Is A Time Bomb
Assessment: UnresolvedSaid
Observation
“In short, CoreWeave's position is one that at the very least competes with the hyperscalers, and is both a fascinating and disturbing window into the actual money that these companies do (or don't) make, and the answer is "not very much at all."”
Our reading: Companies providing AI compute, including hyperscalers, make not very much money at all from it.
No horizon Evidence as of
Limitation: The evidence does not isolate AI-compute profitability across the relevant providers for October–December 2024 [e1, e2].
Source: CoreWeave Is A Time Bomb
Assessment: UnresolvedSaid
Forecast
“The problem is that, yes, GPUs are depreciating assets, a fact that will eventually become problematic for these companies.”
Our reading: GPU depreciation will eventually become a problem for compute sellers whose loans are collateralized by GPUs.
No deadline stated Evidence as of
Limitation: Accounting impairment is only one possible manifestation of the predicted problem.
Source: CoreWeave Is A Time Bomb
Assessment: SupportedSaid
Observation
“CoreWeave’s S-1 tells the tale of a company that appears to be built for collapse, with over 60% of its revenue dependent on one customer, Microsoft.”
Our reading: CoreWeave appears built for collapse, with over 60% of its revenue coming from Microsoft.
No horizon Evidence as of
Limitation: The readable filing is a March 20 amendment reporting the earlier financial year.
Source: CoreWeave Is A Time Bomb
Assessment: UnresolvedSaid
Forecast
“OpenAI has no path to profitability, guaranteeing it will need more cash, and right now — at the time when it needs the most it’s ever needed — SoftBank, the only company willing to provide it, has proven that it will have to go to great lengths to do so.”
Our reading: OpenAI has no path to profitability, which guarantees it will need more cash.
No deadline stated Evidence as of
Limitation: The 2025 figures come from leaked statements as reported in the press, not from a public filing.
Source: Power Cut
Assessment: UnresolvedSaid
Observation
“Worse still, thanks to this agreement, OpenAI’s future — both in its ability to expand its infrastructure (which appears to be entirely contingent on the construction of Stargate) and its ability to raise funding — is entirely dependent on SoftBank, which in turn must borrow money to fund it.”
Our reading: OpenAI's infrastructure expansion and fundraising are entirely dependent on SoftBank, which must borrow to fund it.
No horizon Evidence as of
Limitation: The announcements and reporting do not disclose financing conditions or fallback arrangements sufficient to establish dependence or independence [e1, e2].
Source: Power Cut
Assessment: UnresolvedSaid
Observation
“It is time for investors and the general public to begin demanding tangible and direct numbers on the revenue and profits related to generative AI, as it is becoming increasingly obvious that the revenues are small and the profits are non-existent.”
Our reading: Generative AI revenues are small and profits are non-existent.
No horizon Evidence as of
Limitation: Small is relative and has no stated threshold.
Source: Power Cut
Assessment: UnresolvedSaid
Observation
“SoftBank has committed to over $40 billion of costs that it cannot currently afford, taking on as much as $24 billion in debt in the next year to help sustain one more funding round and the construction of data centers for OpenAI, a company that loses money on literally every single customer.”
Our reading: SoftBank has committed to over $40 billion of costs it cannot currently afford, taking on as much as $24 billion in debt in the next year to help sustain another OpenAI funding round and the construction of data centers for OpenAI, which loses money on every customer.
No horizon Evidence as of
Limitation: The borrowing clause has elapsed, but the principal claim concerns the financial position at publication.
Source: Power Cut
Assessment: UnresolvedSaid
Forecast
“I must repeat that Microsoft will have spent over $200 billion in capital expenditures by the end of 2025.”
Our reading: By the end of 2025, Microsoft will have spent over $200 billion in capital expenditures since 2023.
Deadline passed Evidence as of
Limitation: This record does not assess Microsoft's capital spending in any particular period.
Source: There Is No AI Revolution
Assessment: SupportedSaid
Observation
“OpenAI burned more than $5 billion last year.”
Our reading: OpenAI burned more than $5 billion in 2024.
No horizon Evidence as of
Limitation: The result depends on reading burn as net loss, as suggested by the surrounding discussion.
Source: The Generative AI Con
Assessment: UnresolvedSaid
Observation
“In short, most of the coverage you read on artificial intelligence is led by companies that benefit financially from you thinking artificial intelligence is important and by default all of this coverage mentions OpenAI or ChatGPT.”
Our reading: Most of the AI coverage you read is led by companies that benefit financially from you thinking AI is important, and by default all of this coverage mentions OpenAI or ChatGPT.
No horizon Evidence as of
Limitation: Industry sourcing, industry news pegs, and company control of framing are different measures.
Source: The Generative AI Con
Assessment: UnresolvedSaid
Causal claim
“Furthermore, nothing about this product moves OpenAI toward profitability. In fact, I think they're doing the opposite.”
Our reading: Nothing about Deep Research moves OpenAI toward profitability, and he thinks it moves OpenAI in the opposite direction.
No horizon Evidence as of
Limitation: Compute intensity alone does not establish net financial contribution.
Source: The Generative AI Con
Assessment: UnresolvedSaid
Causal claim
“Anthropic is currently raising $2 billion at a $60 billion valuation primarily based off of this trumped-up marketing nonsense.”
Our reading: Anthropic's $60bn valuation fundraise rests primarily on inflated marketing claims.
No horizon Evidence as of
Limitation: Public investor announcements can themselves be promotional.
Source: The Generative AI Con
Assessment: UnresolvedSaid
Observation
“To be clear, OpenAI burned approximately $5 billion on compute alone.”
Our reading: OpenAI spent about $5 billion on compute alone in 2024.
No horizon Evidence as of
Limitation: The packet contains no verified full-year reconciliation of training, research and inference compute costs.
Source: The Generative AI Con
Assessment: MixedSaid
Observation
“The one "big thing" that they've been able to do is to use "reasoning" to make the Large Language Models "think" (they do not have consciousness, they are not "thinking," this just means using more tokens to answer a particular question and having multiple models check the work), which mostly results in them being a bit more accurate when generating an answer, but at the expense of speed and cost.”
Our reading: The one big thing AI model companies have been able to do is 'reasoning', which means using more tokens and multiple models checking the work rather than actual thinking, and mostly results in models being a bit more accurate at the expense of speed and cost.
No horizon Evidence as of
Limitation: The performance results are developer-reported and concentrated on reasoning tasks.
Source: The Generative AI Con
Assessment: UnresolvedSaid
Forecast
“The tech industry — and part of our economy — is accelerating at speed into a brick wall, driven by people like Sam Altman, Dario Amodei, Marc Benioff, and Larry Ellison, all men that are incentivized to have you value their companies based on something other than what their businesses actually sell.”
Our reading: The tech industry and part of the economy are heading for a crash driven by AI leaders' hype.
No deadline stated Evidence as of
Limitation: "Brick wall" and "part of our economy" have no defined threshold or scope.
Source: The Generative AI Con
Assessment: SupportedSaid
Observation
“User numbers alone tell you nothing about the sustainability or profitability of a business, or how those people use the product.”
Our reading: User numbers alone tell you nothing about a business's sustainability or profitability, or about how its users use the product.
No horizon Evidence as of
Limitation: The result depends on reading 'nothing' as insufficient to establish the conclusions, rather than literally no information.
Source: The Generative AI Con
Assessment: UnresolvedSaid
Observation
“Deep Research uses OpenAI's o3 model which can cost as much as $1,000 a query, and while I imagine these prompts aren't that expensive, they are still significantly more so than a regular query from ChatGPT.”
Our reading: Deep Research queries cost significantly more than regular ChatGPT queries.
No horizon Evidence as of
Limitation: This is an evidentiary gap, not evidence that the relative-cost claim is false.
Source: The Generative AI Con
Assessment: UnresolvedSaid
Observation
“There is no "artificial intelligence" part of Microsoft's revenue or earnings. This is literally Microsoft taking anything with "AI" on it and saying "we made money!"”
Our reading: Microsoft's $13bn AI run-rate figure is not a distinct reported segment but a relabelling of anything with AI.
No horizon Evidence as of
Limitation: The absence of an AI reporting segment does not establish how the supplemental metric was assembled.
Source: The Generative AI Con
Assessment: UnresolvedSaid
Forecast
“And when this all falls apart — and I believe it will — there will be a very public reckoning for the tech industry.”
Our reading: The generative AI boom will fall apart, leading to a very public reckoning for the tech industry.
No deadline stated Evidence as of
Limitation: Public reckoning permits several reasonable forms of accountability.
Source: The Generative AI Con
Assessment: UnresolvedSaid
Forecast
“The damage will be tens of thousands of people fired, long-term environmental and infrastructural chaos, and a profound depression in Silicon Valley that I believe will dwarf the dot-com bust.”
Our reading: The damage will be tens of thousands of people fired, long-term environmental and infrastructural chaos, and a profound depression in Silicon Valley that he believes will dwarf the dot-com bust.
No deadline stated Evidence as of
Limitation: AI-related layoffs can reflect adoption or spending shifts rather than a bubble collapse.
Source: The Generative AI Con
Assessment: UnresolvedSaid
Observation
“There isn't a single god damn startup in the history of anything — other than perhaps Facebook — that has had this level of coverage at such an early stage.”
Our reading: Apart from perhaps Facebook, no startup has had as much media coverage so early as ChatGPT/OpenAI.
No horizon Evidence as of
Limitation: The study does not cover the full target period, November 30, 2022 through February 17, 2025 [e1].
Source: The Generative AI Con
Assessment: SupportedSaid
Observation
“DeepSeek's models don't require massive new data centers (they run on the GPUs currently used to run services like ChatGPT, and can even work on more austere hardware), nor do they require an endless supply of bigger, faster NVIDIA GPUs every year to progress.”
Our reading: DeepSeek's models do not require massive new data centers or ever-bigger NVIDIA GPUs to progress.
No horizon Evidence as of
Limitation: V3's training hardware is company-reported; the report does not establish the facility's construction history [e8].
Source: Deep Impact
Assessment: UnresolvedSaid
Observation
“Right now, I do not believe that there are companies that are stymied by the pricing that OpenAI and their ilk offer, nor do I think there are many companies or use cases that don’t exist because Large Language Models are too expensive.”
Our reading: Right now he does not believe any companies are stymied by the pricing of OpenAI and similar firms, nor that there are many companies or use cases that don't exist because LLMs are too expensive.
No horizon Evidence as of
Limitation: Both underlying observations predate the target period [e1, e2].
Source: Deep Impact
Assessment: UnresolvedSaid
Observation
“He ended with “look forward to bringing you all AGI and beyond” — something which, I add, has always been close on the horizon in Altman’s world, although curiously has yet to materialize, or even come close to materializing.”
Our reading: AGI, which Altman has always portrayed as close on the horizon, has yet to materialize or even come close to materializing.
No horizon Evidence as of
Limitation: The strongest direct evaluation predates the January target period and is context only.
Source: Deep Impact
Assessment: UnresolvedSaid
Observation
“OpenAI only makes 27% of its revenue from selling access to its models — around a billion dollars in annual recurring revenue — with the rest ($2.7 billion or so) coming from subscriptions to ChatGPT.”
Our reading: OpenAI makes only 27% of its revenue, around $1 billion in annual recurring revenue, from selling access to its models, with the rest (about $2.7 billion) coming from ChatGPT subscriptions.
No horizon Evidence as of
Limitation: The original financial documents were unavailable.
Source: Deep Impact
Assessment: UnresolvedSaid
Observation
“There really isn’t anything special about any of these companies anymore — they have no moat, their infrastructural advantage is moot, and their hordes of talent irrelevant.”
Our reading: LLM companies like OpenAI have no moat; their infrastructure advantage and talent no longer matter.
No horizon Evidence as of
Limitation: The capability comparison is developer-reported.
Source: Deep Impact
Assessment: UnresolvedSaid
Observation
“We also know that OpenAI’s training and inference costs in 2024 were around $7 billion, meaning that either refining current models or building new ones is quite costly.”
Our reading: OpenAI's training and inference costs in 2024 were around $7 billion.
No horizon Evidence as of
Limitation: Primary financial records isolating the relevant expenses were not obtained.
Source: Deep Impact
Assessment: UnresolvedSaid
Conditional forecast
“Once these models are hosted elsewhere, and once DeepSeek's methods (which I'll get to shortly) are recreated (which won't take long), I believe we'll see that these prices are indicative of how cheap these models are to run.”
Our reading: Once DeepSeek's models are hosted elsewhere and its methods recreated (soon), its low prices will prove to reflect genuinely low running costs.
No deadline stated Evidence as of
Only if: once models are hosted elsewhere
Source: Deep Impact
Assessment: UnresolvedSaid
Observation
“OpenAI and Anthropic no longer have moats to raise capital with.”
Our reading: OpenAI and Anthropic no longer have moats with which to raise capital.
No horizon Evidence as of
Limitation: Moat is a qualitative investment concept.
Source: Deep Impact
Assessment: UnresolvedSaid
Conditional forecast
“This isn’t a particularly exciting pitch to investors or the public markets, because there’s no product, innovation or business model to point to, and if they’d actually try and productize it and turn it into a business, it’s quite obvious at this point that there really isn’t a multi-trillion dollar industry for generative AI.”
Our reading: It is quite obvious at this point that if LLM companies actually tried to productize LLMs and turn them into a business, there really isn't a multi-trillion-dollar industry for generative AI.
No deadline stated Evidence as of
Only if: if they actually tried to productize it and turn it into a business
Source: Deep Impact
Assessment: SupportedSaid
Forecast
“One thing we do know is that OpenAI and Anthropic will now have to either drop the price of accessing their models, and potentially even the cost of their subscriptions.”
Our reading: OpenAI and Anthropic will have to cut API prices, and possibly subscription prices.
Settled before its deadline Evidence as of
Limitation: Support depends on including lower-priced successors within a model family.
Source: Deep Impact
Assessment: UnresolvedSaid
Forecast
“As a result, I don’t really see anything changing, beyond the eventual collapse of the API market for companies like Anthropic and OpenAI.”
Our reading: The API market for OpenAI and Anthropic will eventually collapse, with little else changing.
No deadline stated Evidence as of
Limitation: No deadline or numerical definition of collapse was supplied.
Source: Deep Impact
Assessment: UnresolvedSaid
Observation
“After all, OpenAI loses money on every transaction — even its $200-a-month "ChatGPT Pro" subscription.”
Our reading: OpenAI loses money on every transaction, including the $200/month ChatGPT Pro subscription.
No horizon Evidence as of
Limitation: The narrower Pro statement is substantiated, but is not equivalent to the full claim.
Source: Deep Impact
Assessment: UnresolvedSaid
Forecast
“While NVIDIA will find other ways to make money — Jensen Huang always does — it's going to be a hard sell for any hyperscaler to justify spending billions more on GPUs to markets that now know that near-identical models can be built for a fraction of the cost with older hardware.”
Our reading: Hyperscalers will struggle to justify spending billions more on GPUs to markets now aware near-identical models can be built cheaply on older hardware.
No deadline stated Evidence as of
Limitation: The market-reaction evidence is reputable reporting rather than a controlled attribution of investor behavior.
Source: Deep Impact
Assessment: UnresolvedSaid
Observation
“As I've written about many, many, many times, the Large Language Models run by companies like OpenAI, Anthropic, Google and Meta are unprofitable and unsustainable, and the transformer-based architecture they run on has peaked.”
Our reading: LLMs run by OpenAI, Anthropic, Google and Meta are unprofitable and unsustainable.
No horizon Evidence as of
Limitation: CNBC had not seen OpenAI's accounts, and the reported loss estimate excluded stock compensation and some other expenses [e1].
Source: Deep Impact
Assessment: UnresolvedSaid
Observation
“As I've written about many, many, many times, the Large Language Models run by companies like OpenAI, Anthropic, Google and Meta are unprofitable and unsustainable, and the transformer-based architecture they run on has peaked.”
Our reading: The transformer-based architecture underlying LLMs has peaked.
No horizon Evidence as of
Limitation: The result depends materially on treating the sentence as a historical observation.
Source: Deep Impact
Assessment: UnresolvedSaid
Causal claim
“In any case, the massive expense of running generative models hasn’t been the limiter on their deployment or success — you can blame that on the fact that they, as a piece of technology, are neither artificial intelligence nor capable of providing the kind of meaningful outcomes that would make them the next Smartphone.”
Our reading: The massive expense of running generative models hasn't been what limits their deployment or success; the limit is that they are neither artificial intelligence nor capable of the meaningful outcomes that would make them the next smartphone.
No horizon Evidence as of
Limitation: The deployment survey was commissioned by an AI vendor and measures reported experiences.
Source: Deep Impact
Assessment: UnresolvedSaid
Observation
“And, as I noted at the time, it was not particularly reliable, failing to accurately count the number of times the letter ‘r’ appeared in the word “strawberry.””
Our reading: OpenAI's o1 was not particularly reliable at launch, failing to count the r's in 'strawberry'.
No horizon Evidence as of
Limitation: The original failure may survive in an image or conversation that read_page did not expose.
Source: Deep Impact
Assessment: UnresolvedSaid
Forecast
““AGI” isn’t coming.”
Our reading: AGI is not coming.
No deadline stated Evidence as of
Limitation: AGI has no agreed empirical test.
Source: Deep Impact
Assessment: MixedSaid
Observation
“The launch of its "Operator" "agent" was a joke — a barely-functional product that is allegedly meant to control your computer and take distinct actions, but doesn't seem to work.”
Our reading: OpenAI's Operator agent is barely functional and does not seem to work.
No horizon Evidence as of
Limitation: Successful simple tasks do not establish dependable completion of harder tasks.
Source: Deep Impact
Assessment: UnresolvedSaid
Causal claim
“I do not believe that the companies making foundation models (such as OpenAI and Anthropic) have been incentivized to do more with less, and because their chummy relationships with hyperscalers were focused almost entirely on "make the biggest, most hugest models possible, using the biggest, most hugest chips," and because the absence of profitability didn’t stop them from raising more money, efficiency was never a major problem for them.”
Our reading: He hypothesises that foundation model companies such as OpenAI and Anthropic were not incentivised to do more with less, and that because their hyperscaler relationships focused almost entirely on the biggest models and chips, and because being unprofitable didn't stop them raising money, efficiency was never a major problem for them.
No horizon Evidence as of
Limitation: The announcements do not establish financing terms, internal budget constraints, or the relative priority of scale and efficiency. [e1, e2, e3]
Source: Deep Impact
Assessment: ContradictedSaid
Observation
“Large Language Models (and reasoning models) are niche.”
Our reading: LLMs and reasoning models are niche products.
No horizon Evidence as of
Limitation: The result depends on reading niche as narrow market reach rather than limited usefulness.
Source: Deep Impact
Assessment: UnresolvedSaid
Conditional forecast
“And if OpenAI cuts its prices to compete with DeepSeek, its losses will only deepen.”
Our reading: If OpenAI cuts prices to compete with DeepSeek, its losses will deepen.
No deadline stated Evidence as of
Only if: if OpenAI cuts prices to compete with DeepSeek
Source: Deep Impact
Assessment: UnresolvedSaid
Observation
“After all, OpenAI lost $5 billion (after $3.7 billion in revenue, too!), and Anthropic just under $3 billion in 2024.”
Our reading: In 2024 OpenAI lost $5 billion on $3.7 billion revenue and Anthropic lost just under $3 billion.
No horizon Evidence as of
Limitation: This does not mean that no 2024 financial results are available.
Source: Deep Impact