Meta's Mosseri: AI Token Budgets Will Soon Be Capped Per Engineer
Instagram head Adam Mosseri thinks tech companies will manage AI spending the same way they manage payroll — with per-person caps.
W
We Read AI
We Read AI
Meta's Adam Mosseri told an interviewer this week that per-engineer AI token budgets are coming to large companies, treated the same way engineering headcount and equipment spend are today. The prediction reflects a growing awareness that unbounded AI tool access has become one of the fastest-growing line items in engineering budgets.
Where this is heading
The default at most tech companies today is unlimited access to Claude, GPT-5, and Cursor for developers. A senior engineer using coding tools aggressively can burn $2,000-5,000 a month in tokens. Multiply by 10,000 engineers and the number gets attention.
The FinOps reality
Per-seat caps are the natural response, but they're operationally messy. Engineers on complex projects legitimately need more; engineers on simple tickets need less. The most likely evolution is a team-level pool with per-project attribution, backstopped by hard ceilings.
The vendor angle
If per-engineer caps become standard, the model providers face an incentive to compete harder on cost per equivalent capability. That may finally break the "everyone charges the same" pattern across Anthropic, OpenAI, and Google.
Source: TechCrunch
#meta#finops#engineering#cost