Dan Luua developer with a past at Google and Microsoft, has lined up dozens of public predictions of Ed Zitronauthor of the newsletter “Where’s Your Ed At” and probably the most cited commentator among those who argue that the artificial intelligence industry is a bubble destined to burst. The result, published on his personal blog, is a list of dated and verifiable statements which, when compared with real data, are almost all wrong.
The most cited case concerns an intervention in 2024, when Zitron has described Meta, Google and Microsoft as “dying” companiesunable to find new avenues of growth and therefore forced to “stick AI everywhere” out of desperation. Luu compared the statement with the financial statements: Meta closed 2023 with 135 billion dollars of revenues and a profit of 47 billion, which rose respectively to 201 billion and 83 billion in 2025. Alphabet went from 307 to 403 billion dollars in turnover in the same period, Microsoft from 228 to 305 billionboth with double-digit growth profit margins.
The AI skeptic’s goals missed, one by one
The list continues with other specific cases. In February 2025, Zitron had branded Google’s goal of bringing Gemini to as unrealistic 500 million users by the end of the year, claiming that someone at the company should have been fired for staring at him: the actual goal was 750 million. In October 2024 he defined OpenAI’s forecast of exceeding $11.6 billion in revenues in 2025 as “almost a financial crime”, a goal that was then achieved and exceeded.
Same fate for the deadline for the collapse of the sector: when asked in October 2025 about when the AI bubble would burst, Zitron replied “no later than the second quarter of 2026”, a deadline now outdated without anything of the kind having happened. In July 2025 he predicted the end of Cursor, the AI-based code editor, which instead was acquired at a valuation of $60 billion. Also CoreWeavegiven in March 2025 as unable to survive beyond six months without new capital, is now listed on the stock exchange at more than double its IPO price.
| Agency | Period | Revenues | Useful | ||
|---|---|---|---|---|---|
| Amount | % | Amount | % | ||
| Half | 2023 | $135 billion | 16% | $47 billion | 62% |
| 2024 | $165 billion | 22% | $69 billion | 48% | |
| 2025 | $201 billion | 22% | $83 billion | 20% | |
| 1st semester 2026 | $117 billion | 30% | $42 billion | 10% | |
| Alphabet (Google) | 2023 | $307 billion | 9% | $84 billion | 13% |
| 2024 | $350 billion | 14% | $112 billion | 33% | |
| 2025 | $403 billion | 15% | $129 billion | 15% | |
| 1st semester 2026 | $230 billion | 23% | $80 billion | 30% | |
| Microsoft | 2023 | $228 billion | 12% | $101 billion | 21% |
| 2024 | $262 billion | 15% | $118 billion | 17% | |
| 2025 | $305 billion | 17% | $143 billion | 21% | |
| 1st semester 2026 | $173 billion | 18% | $79 billion | 19% | |
According to Luu, the problem lies not only in the missed results but in the method: several of Zitron’s theses are based on second-hand sources, such as Similarweb traffic estimates used instead of official data on active Facebook users, or on causal links that the numbers cited do not really support. Other industry observers are also cited in the analysis, according to which Zitron’s use of data would be ssystematically misleading compared to the primary sources to which it also refers.
Luu, who describes himself as having no direct financial interest in the AI sector, recalls having conducted a similar exercise in 2022 on a group of historical futurologists, finding them in turn mostly wrong in their long-term estimates. The difference, he writes, is in the degree of certainty with which Zitron makes his predictions: statements presented as almost certain and promptly denied by subsequent facts.
