Meta’s Adam Mosseri says AI token budgets could soon be capped per engineer

In a recent interview, Instagram head Adam Mosseri said he can see a time in the future, perhaps just a year or two, when putting limits on Meta employees’ AI token spending will become necessary.

“I think you can imagine, at least in a year or two… that the burn rate of a strong engineer could be the same as their salary, or their labor costs. And in that world you’re probably going to have to put some caps in place,” the Meta director said, while speak on Lenny’s Podcast.

AI token spending, a reference to the cost of processing AI prompts and responses, has been a hot topic in recent days. Meta is closed an internal AI token spending leaderboard after AI costs put the company on track billions of dollars in 2026.

Meta isn’t the only one rethinking his approach to AI experiments. Uber also had an AI reckoning after exhausting its 2026 AI coding budget in April. Saw the rising symbolic costs Microsoft cancels Claude Code licenses and consolidates its engineers around its proprietary Copilot CLI tool.

Mosseri’s belief, he explained, is that the cost of AI tokens should be managed like any other resource, which offers an analogy to things like payroll or operating expenses (OpEx), which are the day-to-day costs of running a business.

“I look at it like…any other resource,” Mosseri said. “I have to decide how to deploy capacity for my different teams because I have a limited number of GPUs and CPUs, storage and RAM, etc. I have to decide how to deploy OpEx for budget labeling on my teams. I have to decide how to deploy payroll for headcount on my teams.”

Token budgets will be the same, he added, noting that the per-engineer limit should be proportional to the company’s confidence in their ability to use the budget in an “ROI positive” way.

Meta does not currently have token caps for any employees, Mosseri said, but he believes their use could be healthy in the future. Moving on, he expects token costs to drop as AI model makers start a price war to get people to use their tools versus their competitors.

For now, the company has managed to keep its token costs in check a bit by stopping the “silly things” it was doing, Mosseri noted – like that token spend scoreboard.

“It’s not that difficult to build a token incinerator, and that doesn’t create much value,” he said.