How accurate have Ed Zitron's AI skeptic predictions been?

Software engineer Dan Luu has published an audit of the prediction record of Ed Zitron, one of the most widely cited critics of the AI industry. The article compares a series of Zitron's public forecasts, most of them written for his "Where's Your Ed At" newsletter, with what actually happened afterwards.

Meta, Google and Microsoft are "dying"

Luu examines a November 2024 talk in which Zitron argued that major technology companies such as Meta and Google are dying and are "thrashing around" on AI because they no longer know how to grow. Meta, he said, is "a dying product, and it's kind of a dying company".

Reported numbers point the other way. Meta's revenue and operating profit rose from $135B and $47B in 2023 to $201B and $83B in 2025, with $117B and $42B in the first half of 2026 alone. Alphabet went from $307B and $84B in 2023 to $403B and $129B in 2025, and Microsoft from $228B and $101B in 2023 to $305B and $143B in 2025. If AI were a desperate grab for growth by companies out of ideas, Luu notes, the financials do not show it.

He also traces part of Zitron's case to weak evidence, such as an alleged decline in Facebook monthly active users taken from third-party tracker Similarweb rather than Meta's own reports, even though most estimates show Facebook usage rising since the company stopped publishing those figures in December 2023.

The prediction list

The audit catalogues dozens of forecasts, nearly all marked wrong. In early 2024 Zitron wrote that generative AI was reaching "the upper limits" of what it can do and asked whether we had reached "Peak AI". Later posts said the industry was running out of training data, that OpenAI's growth was stalling, that models were not getting more powerful or more efficient, and that generative AI was "a dead-end technology that has peaked".

Financial calls are listed too. Zitron called OpenAI's own forecast of $11.6B of revenue in 2025 "absurd"; the figure was exceeded. He dismissed an estimate that Anthropic would make $34.5B in 2027 as "laughable", said Google's Sundar Pichai should have been fired for claiming Gemini could reach 500 million users by the end of 2025 — Gemini passed 750 million — and predicted Cursor would die, though it reached a $60B exit. In October 2025 he guessed the AI bubble would pop "no later than Q2 2026".

Reasoning, not just hits and misses

Luu says he checked the reasoning as well as the results. He cites Timothy B. Lee's review of a spreadsheet Zitron used to project Anthropic's revenue, which miscounted weeks and included a February 30 entry, and comments by Juho Snellman that Zitron's numbers are "sleight of hand".

Luu compares Zitron to the futurists he audited in 2022, including Ray Kurzweil: both use numbers to create an aura of credibility and move failed predictions forward in time. He notes that he holds no strong pro- or anti-AI position, owns AI shares only through index funds, and has criticised Google's search quality himself. Even a major AI crash, he argues, would mainly reshuffle who benefits from capabilities that already exist.