The 'tokenpocalypse': companies limit employees' AI use because of its high cost

🕒 Published on Zendoric: July 4, 2026 · 00:29
This 404 Media article is, for the most part, a paywall teaser: the downloaded content only includes the introduction and the closing of the report, without the full account of the facts, figures and internal quotes it promises.
This 404 Media article is, for the most part, a paywalled teaser: the downloaded content only includes the introduction and closing of the report, without the full development of the facts, figures, and internal quotes it promises. Therefore, what follows is a brief summary based exclusively on what is visible in the text received, without inventing additional details.
According to the article's lead, written by Joseph Cox and Emanuel Maiberg, several companies in sectors such as technology, entertainment, and banking —including Amazon, Adobe, Atlassian, and Citi— are reportedly restricting AI use by their employees because costs have skyrocketed. The authors claim to have obtained leaked Slack chats, screenshots of internal dashboards, and emails showing how these companies are asking their workers to use less powerful AI models to curb spending.
The text mentions, as a concrete data point, that in at least one case AI spending has tripled to over $15 million per month. It also indicates that some companies have reportedly cut off access to certain AI models entirely to avoid continued token consumption, and that companies like Adobe are reportedly eliminating unlimited access to Claude.
The article frames this situation as a direct consequence of the rapid corporate adoption of generative AI, combined with the shift in pricing models by AI providers, who have moved from flat rates to billing schemes based on actual consumption (per-token usage). This has reportedly given rise to what the outlet itself calls, in related content, the "tokenpocalypse": a moment when the cost of sustaining massive generative AI use within organizations becomes unsustainable or, at the very least, forces much stricter control.
Beyond this introduction, the rest of the report —which presumably details specific company examples, screenshots of internal conversations, and additional figures— is locked behind 404 Media's paywall, making it impossible to offer a deeper analysis without speculating on unverified content. It's worth noting that 404 Media is an independent outlet founded by technology journalists, known for its coverage based on leaks and internal documents, which lends some credibility to the seriousness of the investigation, although its full scope cannot be confirmed from the material available here.
🔗 Related on Zendoric


