Meta may cap engineer AI token budgets as costs rise

Meta currently has no token caps in place for any employee, according to Instagram head Adam Mosseri, but he said on Lenny’s Podcast that he can see a time — perhaps only a year or two away — when limits on Meta employees’ AI token spend, the cost of processing AI prompts and responses, will probably become necessary. “I think that you can imagine, at least in a year or two … that the burn rate of a strong engineer might be the same as their salary, or their cost of employment. And in that world, you’re going to probably need to put in some caps,” Mosseri said.

To rein in those costs, Meta recently shut down an internal AI token spend leaderboard after it put the company on track for billions of dollars in AI costs in 2026. Mosseri described the leaderboard as what he called a token incinerator — an easy-to-build practice that burned through spend without creating much value, prompting Meta to cut it.

Meta is not alone in confronting this reckoning. Uber exhausted its 2026 AI coding budget by April, while Microsoft cancelled its Claude Code licenses — Anthropic’s AI coding tool — and consolidated its engineers around its own Copilot CLI tool amid soaring token costs.

Mosseri argued that AI token spend has to be managed like any other corporate resource. He compared it to how he already allocates limited GPU and CPU capacity, storage, and RAM across teams, along with OpEx — operating expenditure, the day-to-day cost of running a business — for labeling budgets, and payroll for headcount. Token budgets, he said, would work the same way, with per-engineer caps eventually set in proportion to how much a company trusts that engineer to use tokens in an “ROI-positive” way. Looking further ahead, Mosseri said he expects token costs to decline as AI model makers compete on price to win users away from rivals. For now, Meta has trimmed its bill mainly by eliminating what he called the “silly things” it had been doing — like the leaderboard itself.

Why it matters

Soaring AI token costs are forcing major tech companies like Meta, Uber, and Microsoft to rethink how they let employees experiment with AI, treating token spend as a strictly managed corporate resource.