# Ed Zitron argues in Where’s Your Ed At that AI finance runs on bad numbers [Canonical reading](https://escapemeridian.com/readings/ed-zitron-argues-in-where-s-your-ed-at-that-ai-finance-runs-on-768e4603) Ed Zitron’s piece was published on October 13, 2026. He argues that annualized run rates—short-period revenue extrapolated to a year—can mislead when the period and calculation are undisclosed, especially for consumption-based sales. He says these figures invite comparisons that may not be like-for-like and can make company revenue seem more meaningful than the underlying data warrants. His wider argument is that opaque revenues, costs, compute capacity and data-center economics make it difficult to assess the AI industry’s financial condition. Zitron argues that venture-funded AI startups’ spending on model providers supports OpenAI and Anthropic’s revenues. In his account, startups can lose money when users consume more tokens than subscriptions cover; if funding dries up, the model providers could lose revenue too. He extends his critique to GPU rental prices and data-center returns, arguing that available figures rely on incomplete or mismatched comparisons. Zitron’s conclusion is that investors and journalists should not treat projections and annualized run rates as substitutes for actual financial data. He warns that distinguishing weak financial information from meaningful figures matters to investors’ risk of losing money. The piece does not establish the full underlying finances of the companies it discusses; its account of their economic dependencies is Zitron’s.”, Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims. ## Original source Ed Zitron · Article Original source date: Oct 13, 2026 (2026-10-13T16:03:00.000Z). Date basis: Date supplied by the publisher’s feed. Added to Escape Meridian: 2026-11-04T23:48:24.652Z. First observed: 2026-10-23T18:13:48.830Z. Retrieved: 2026-10-23T18:15:40.096Z. - [Annualized Enron Rate](https://www.wheresyoured.at/annualized-enron-rate/) ## Limits Text as retrieved on October 23, 2026; originally published October 13, 2026. The piece presents Zitron’s analysis and interpretations of reported financial figures; the retrieved text does not provide a complete, independently verifiable account of the companies’ underlying finances. Zitron’s assertions about how startup token spending supports model-provider revenues are arguments in the piece, not demonstrated outcomes. ## Open question What are the companies’ actual revenues, costs and compute capacity, and how would those figures change Zitron’s account of the AI industry’s financial dependencies? ## Source-stated Zitron argues that annualized run rates lack value when the period and calculation are not specified. > Without an actual period of time — and I mean an exact one, such as “July 1 to July 31, 2026” — and the exact way it was calculated, this number effectively means nothing - [Supporting source](https://www.wheresyoured.at/annualized-enron-rate/) ## Source-stated Zitron argues that AI startups’ spending on OpenAI and Anthropic models depends on venture funding, and that this spending supports those labs’ revenues. > The vast majority of hyperscaler AI revenues come from Anthropic and OpenAI’s compute spend that is dependent on near-perpetual funding, much like the vast majority of Anthropic and OpenAI’s revenues come from AI startups that depend on near-perpetual funding to continue their spend. - [Supporting source](https://www.wheresyoured.at/annualized-enron-rate/) ## Source-stated Zitron says the available information does not establish the financial condition of most AI companies or the economics of AI data centers. > We do not know the revenues or much about the economics of Perplexity, Harvey, Instinct, Reflection, Cognition, Replit, Lovable, and Factory - [Supporting source](https://www.wheresyoured.at/annualized-enron-rate/) ## Source-stated Zitron argues that startup customers’ token use can exceed their subscription revenue, leaving startups dependent on continuing investment. > Customers of AI startups are repulsed by the actual cost of tokens, meaning that for every dollar of subscription revenue, AI startups must allow customers to burn anywhere from 2 times to 10 times their revenue in tokens. - [Supporting source](https://www.wheresyoured.at/annualized-enron-rate/) ## Source and contributor context Ed Zitron · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media retrieval status: unchecked. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## AI compute The computing resources, meaning chips, data centers and electricity, used to train or run AI models. - [Definition reference](https://developers.google.com/machine-learning/glossary#compute) ## Graphics processing unit A chip that performs many calculations at once; the main hardware used to train and run AI models. - [Definition reference](https://en.wikipedia.org/wiki/Graphics_processing_unit) ## OpenAI OpenAI develops AI models and products and publishes research, release announcements and evaluations. - [Explore this context](/entities/openai) - [Reference](https://openai.com/news/) ## Anthropic Anthropic develops Claude and publishes research about AI capability, safety and how people use its systems. - [Explore this context](/entities/anthropic) - [Reference](https://www.anthropic.com/news) - [Reference](https://www.anthropic.com/research/team/economics) ## Editorial provenance Version 1. Publication ID: 7ed8b5dd-b222-2b6e-9088-35f7ef948f2a. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-04T23:43:31.094Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model. Observation: 36f3acf6f99841c28b57e74b21be6ac8990dcc685ad88e6e36acc6c796c9c5fe. Source text hash: 01a160b5b69a61fdad7d0ca2f8a38ec5dcd81a5330c45b362d0cbe1eb5633117. ## Revision and source-check history ### Version 1 — 2026-11-04T23:48:24.652Z Ed Zitron’s piece was published on October 13, 2026. He argues that annualized run rates—short-period revenue extrapolated to a year—can mislead when the period and calculation are undisclosed, especially for consumption-based sales. He says these figures invite comparisons that may not be like-for-like and can make company revenue seem more meaningful than the underlying data warrants. His wider argument is that opaque revenues, costs, compute capacity and data-center economics make it difficult to assess the AI industry’s financial condition. Zitron argues that venture-funded AI startups’ spending on model providers supports OpenAI and Anthropic’s revenues. In his account, startups can lose money when users consume more tokens than subscriptions cover; if funding dries up, the model providers could lose revenue too. He extends his critique to GPU rental prices and data-center returns, arguing that available figures rely on incomplete or mismatched comparisons. Zitron’s conclusion is that investors and journalists should not treat projections and annualized run rates as substitutes for actual financial data. He warns that distinguishing weak financial information from meaningful figures matters to investors’ risk of losing money. The piece does not establish the full underlying finances of the companies it discusses; its account of their economic dependencies is Zitron’s.”, ## Original source Ed Zitron · Article Original source date: Oct 13, 2026 (2026-10-13T16:03:00.000Z). Date basis: Date supplied by the publisher’s feed. Added to Escape Meridian: 2026-11-04T23:48:24.652Z. First observed: 2026-10-23T18:13:48.830Z. Retrieved: 2026-10-23T18:15:40.096Z. - [Annualized Enron Rate](https://www.wheresyoured.at/annualized-enron-rate/) ## Limits Text as retrieved on October 23, 2026; originally published October 13, 2026. The piece presents Zitron’s analysis and interpretations of reported financial figures; the retrieved text does not provide a complete, independently verifiable account of the companies’ underlying finances. Zitron’s assertions about how startup token spending supports model-provider revenues are arguments in the piece, not demonstrated outcomes. ## Open question What are the companies’ actual revenues, costs and compute capacity, and how would those figures change Zitron’s account of the AI industry’s financial dependencies? ## Source-stated Zitron argues that annualized run rates lack value when the period and calculation are not specified. > Without an actual period of time — and I mean an exact one, such as “July 1 to July 31, 2026” — and the exact way it was calculated, this number effectively means nothing - [Supporting source](https://www.wheresyoured.at/annualized-enron-rate/) ## Source-stated Zitron argues that AI startups’ spending on OpenAI and Anthropic models depends on venture funding, and that this spending supports those labs’ revenues. > The vast majority of hyperscaler AI revenues come from Anthropic and OpenAI’s compute spend that is dependent on near-perpetual funding, much like the vast majority of Anthropic and OpenAI’s revenues come from AI startups that depend on near-perpetual funding to continue their spend. - [Supporting source](https://www.wheresyoured.at/annualized-enron-rate/) ## Source-stated Zitron says the available information does not establish the financial condition of most AI companies or the economics of AI data centers. > We do not know the revenues or much about the economics of Perplexity, Harvey, Instinct, Reflection, Cognition, Replit, Lovable, and Factory - [Supporting source](https://www.wheresyoured.at/annualized-enron-rate/) ## Source-stated Zitron argues that startup customers’ token use can exceed their subscription revenue, leaving startups dependent on continuing investment. > Customers of AI startups are repulsed by the actual cost of tokens, meaning that for every dollar of subscription revenue, AI startups must allow customers to burn anywhere from 2 times to 10 times their revenue in tokens. - [Supporting source](https://www.wheresyoured.at/annualized-enron-rate/) ## Source and contributor context Ed Zitron · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media retrieval status: unchecked. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## AI compute The computing resources, meaning chips, data centers and electricity, used to train or run AI models. - [Definition reference](https://developers.google.com/machine-learning/glossary#compute) ## Graphics processing unit A chip that performs many calculations at once; the main hardware used to train and run AI models. - [Definition reference](https://en.wikipedia.org/wiki/Graphics_processing_unit) ## OpenAI OpenAI develops AI models and products and publishes research, release announcements and evaluations. - [Explore this context](/entities/openai) - [Reference](https://openai.com/news/) ## Anthropic Anthropic develops Claude and publishes research about AI capability, safety and how people use its systems. - [Explore this context](/entities/anthropic) - [Reference](https://www.anthropic.com/news) - [Reference](https://www.anthropic.com/research/team/economics) ## Editorial provenance Version 1. Publication ID: 7ed8b5dd-b222-2b6e-9088-35f7ef948f2a. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-04T23:43:31.094Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model. Observation: 36f3acf6f99841c28b57e74b21be6ac8990dcc685ad88e6e36acc6c796c9c5fe. Source text hash: 01a160b5b69a61fdad7d0ca2f8a38ec5dcd81a5330c45b362d0cbe1eb5633117. ### uncertain — 2026-11-04T23:43:31.094Z — anthropic/claude-opus-5.5 Publication: 7ed8b5dd-b222-2b6e-9088-35f7ef948f2a attribution — pass: The piece is a single essay by Ed Zitron on his site wheresyoured.at (Where’s Your Ed At). The reading attributes the arguments to him throughout and presents them as his claims, not as established facts. > The entire AI bubble is inflated based on stupid, made up numbers that only exist to pump the bags of investors summary — uncertain: The core argument is captured accurately, but two issues affect the reading. First, “tokens” is central to the startup-economics argument and is neither explained nor in the definitions. Suggested fix: add a term-library entry for tokens. Second, the hedged wording (“can mislead”, “may not be like-for-like”) flattens Zitron’s force. He calls run rates a marketing exercise that “tells you nothing”. A smaller repair would restore that conviction while keeping it attributed. The summary also ends with stray characters (”,). > Reporting “run rates” is a marketing exercise, and if you disagree, please answer me the following date — pass: evidence.originalPublished is 2026-10-13T16:03Z, whose UTC calendar day is October 13, 2026. The reading states that day, which matches the evidence. The evidence kind is publisher-page, so no archive disclosure is needed. The reading presents nothing as happening after that date and uses no hindsight framing. Executive projections, such as Cognition’s, are not presented as outcomes. support — pass: Each claim’s excerpt directly supports it as Zitron’s stated argument. The run-rate critique, the startup and venture-funding dependency, the unknown startup revenues and the 2x–10x token burn are all grounded in the text. The investor-risk warning is grounded in his closing line. The question follows from his stated data gaps. The page records a later modification date, but editedAfterPublication is not flagged, and the cited passages are integral to the argument. > Your ability to distinguish between which of them you’re talking to is critical to not losing your shirt. media — pass: No media was supplied or selected (mediaId null), and no candidates exist. There is nothing to misidentify. ## Recorded references Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter. No public references recorded here yet. [Reference data (JSON)](/readings/768e4603a6a931d8a6060a0d7426fb9b0161a27883cde11080bbbc2f9d26e157/references.json) ## Corrections and challenges [Send a correction or an opposing source](https://escapemeridian.com/corrections/) [JSON record](https://escapemeridian.com/readings/ed-zitron-argues-in-where-s-your-ed-at-that-ai-finance-runs-on-768e4603.json)