Meta shifts from "tokenmaxxing" to token managing as internal AI costs reportedly hit billions
Meta is implementing stricter controls on its internal AI usage, moving from a "tokenmaxxing" approach to "token managing." This shift is driven by projected AI costs exceeding billions of dollars for internal operations alone. Starting in 2027, Meta will utilize budgets, allocations, and a new system called "AI Gateway" to govern token consumption, emphasizing that token usage does not equate to actual progress or impact. AI
IMPACT Meta's shift to strict AI token management signals a growing focus on cost efficiency and ROI for internal AI deployments across the industry.