Adam Mosseri, head of Instagram, told Lenny’s Podcast that Meta will likely cap AI token spending for its engineers within the next year or two.
The executive noted that the cost for a skilled engineer to generate prompts could soon equal their total salary or employment cost. At that point, hard limits become necessary.
This view follows a period of intense scrutiny over AI spending. Meta removed an internal leaderboard tracking token consumption after costs threatened to reach billions of dollars in 2026.
Other tech firms face similar pressures. Uber exceeded its 2026 AI coding budget by April. Microsoft cancelled licenses for Claude Code and moved its engineers to use its own Copilot CLI tool instead.
Mosseri compares token costs to standard business expenses like payroll or operating expenditure. He treats compute capacity as a finite resource that requires careful allocation across teams.
“I have to decide how to deploy capacity to my different teams because I have a limited number of GPUs and CPUs and storage and RAM etc,” Mosseri said. “I have to decide how to deploy OpEx for labeling budgets across my teams. I have to decide how to deploy payroll for headcount across my teams.”
Future token budgets will depend on management confidence in an employee’s ability to generate a return on investment. Currently, Meta has no such caps for staff.
Mosseri expects pricing to drop eventually as model providers compete for users. For now, the company has reduced waste by stopping initiatives that offered little value, such as the token spend leaderboard.
“It’s not that hard to build a token incinerator, and that doesn’t create a lot of value,” he said.
What it means
Developers and data scientists should expect stricter controls on their AI usage. Projects requiring high volumes of tokens may face delays or require formal approval before they can proceed.
