Meta develops centralized AI spending controls as internal usage costs surge

Meta is building a centralized system to monitor and limit employee AI usage after internal spending rose faster than expected, with full token controls targeted for 2027. A memo sent to roughly 6,000 staff outlines budgets, caps, and automated alerts for unusual spikes, while phasing out internal usage leaderboards tied to “tokenmaxxing.” The company said internal AI adoption is climbing quickly and it could be spending tens of billions on employee AI usage in 2026.