Meta warns AI token costs may soon match engineer salaries
Meta's Instagram chief predicts per-engineer AI spending caps within two years, signalling a broader industry reckoning over unsustainable computing costs.
Meta might soon limit how much its engineers can spend on artificial intelligence tools. Instagram chief Adam Mosseri said on a recent podcast that he envisions implementing per-employee token budgets within a year or two.
The rationale is purely financial. “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 on Lenny’s Podcast.
This is not an isolated Silicon Valley problem, but a warning for the wider tech economy. Uber exhausted its entire 2026 AI coding budget by April. Microsoft recently cancelled Claude Code licences to curb soaring token expenses, consolidating its workforce around its own Copilot CLI tool instead.
Meta itself has already felt the heat of unchecked experimentation. The company was on track to spend billions of dollars on AI tokens in 2026, prompting it to dismantle an internal leaderboard that tracked employee AI usage. Mosseri dismissed such tools as a “token incinerator” that failed to generate value.
For investors and European enterprises tracking AI adoption, this signals the end of the grace period. If the world's largest tech firms are struggling to absorb the day-to-day costs of processing AI prompts, businesses elsewhere will face similar margin pressures. AI spending is rapidly transitioning from a research luxury to a standard operational expense that demands strict corporate budgeting.
Mosseri framed the impending shift as basic resource management. “I think of it like…any other resource. 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. 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.” Any future caps at Meta would be tied to an engineer's proven ability to deliver a return on investment.
Relief may eventually come through market competition. Mosseri noted that AI model providers will likely enter a pricing war to attract users, eventually driving token costs down. Until then, tech firms are being forced to treat AI computing not as a limitless utility, but as a finite, expensive resource.
