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

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

“I think you can imagine, at least in a year or two… that the burnout rate of a strong engineer might be the same as their salary, or the cost of hiring them. And in that world, you’re probably going to need to set some caps,” the Meta CEO said, while… to talk On Lenny’s podcast.

Spending AI tokens, referring to the cost of processing AI claims and responses, was a reference to the cost of processing AI claims and responses A topic that arouses a lot of interest In recent days. Dead closed Internal leaderboard for AI token spending After the costs of artificial intelligence put the company on the right path Billions Dollar in 2026.

Meta is not alone in rethinking its approach to AI experience. Uber too It had an artificial intelligence account Having exceeded its 2026 AI development budget by April. It experienced high nominal costs Microsoft cancels Claude Code licensed it, consolidating its engineers around its Copilot CLI tool instead.

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

“I think it’s just like… any other source,” Mosseri said. “I have to decide how to deploy capacity to my different teams because I have a limited number of GPUs, CPUs, storage, RAM, etc. I have to decide how to deploy OpEx to name budgets across my teams. I have to decide how to deploy payroll to headcount across my teams.”

He added that token budgets will be the same, noting that each engineer’s cap should be proportional to the company’s confidence in its ability to use the budget in a “positive ROI” way.

Mosseri said Meta does not currently have a nominal cap for any employee, but he believes its use could be healthy in the future. In the future, token costs are expected to decline as AI model makers engage in a pricing war to entice people to use their tools over their competitors.

For now, the company has been able to rein in token costs a bit by stopping the “silly things” it was doing, Mosseri noted — like the token spending leaderboard.

“It’s not difficult to build a symbolic crematorium, and it doesn’t create a lot of value,” he said.