What is ai-spend-cap-tracker?

This repository is not a piece of software but a public, sourced record tracking companies that are limiting or cutting how much their employees spend on AI coding tools. It documents a shift happening in 2026 where organizations that previously encouraged unlimited AI usage are now capping it, a trend the author calls moving from tokenmaxxing to tokenminimizing. The heart of the project is a table listing specific companies, what action each one took, the size of the cap or cut where known, when it happened, which AI tools were affected, and a link to a named news source for every entry. Companies covered include large firms placing dollar caps on tools like Claude Code and Cursor, others cancelling licenses outright in favor of cheaper alternatives, and some centralizing spend through internal tracking dashboards. The project also keeps a section for counter examples, noting at least one company that reversed an earlier restriction and expanded AI tool access instead. The author is explicit about the inclusion rule: an entry only goes into the table if it is backed by a named, linkable source such as a reputable news outlet or an official company statement. Unconfirmed reports are kept separate and marked for verification rather than being added directly. The README also explains the reasoning behind the project, arguing that blanket spending caps are a blunt approach that can hold back the engineers most likely to benefit from AI tools while ignoring other risks like proprietary code exposure, and it points to a related open source project as an alternative approach to routing AI usage more selectively. The data itself is licensed under CC BY 4.0, meaning it can be reused as long as the source is credited, and the project welcomes outside contributions of new sourced entries.