Meta and Tesla move to cap employees' own AI spending as internal bills balloon

Meta has reportedly ordered internal limits on AI token usage for roughly 6,000 employees after they consumed 73.7 trillion tokens in about 30 days. Tesla, separately, will cap individual staff spending on third-party AI services at $200 per week starting July 6, 2026. Both moves surfaced in reporting corroborated across outlets on July 1 and 2.

The throttling is a rare, concrete data point in the debate over whether AI capital spending is paying off, because it shows the cost pressure landing inside the companies building the tools. When the firms most bullish on AI start rationing their own engineers’ usage, the economics of heavy usage come into view.

At Meta, an internal usage leaderboard nicknamed “Claudeonomics” will be dismantled and replaced with a centralized dashboard called AI Gateway that tracks spending in real time and flags spikes; a full token-budget framework is planned for early 2027. Chief Technology Officer Andrew Bosworth pushed back on what staff internally called “tokenmaxxing.”

Tesla’s $200 weekly cap follows some software engineers running up bills in the thousands of dollars on large language model APIs and coding assistants, according to the reporting. The cap excludes beta versions of xAI products, the AI company run by Tesla CEO Elon Musk.

The pattern is not confined to those two. Uber exhausted its annual AI budget in four months and now caps spending at $1,500 per tool per month, one comparable case cited in the reporting.

Neither Meta nor Tesla has published a first-party statement confirming the policies; both accounts rest on outlet reporting, and the 73.7 trillion-token figure could not be verified against a company disclosure. Whether the caps dent productivity or simply trim waste is the test that follows.