Meta Sets AI Spending Controls as Costs Near Billions

Meta is implementing new spending controls and oversight tools for internal artificial intelligence usage after a surge in employee adoption drove operational costs toward billions of dollars. The company outlined the measures in a memo distributed to approximately 6,000 employees this week, signaling a strategic shift from aggressively promoting AI use to managing its financial impact.

The company now expects its internal AI usage to generate costs running into the billions for 2026. This projection follows a period where Meta actively encouraged a culture of "tokenmaxxing," urging employees to maximize their use of AI-powered tools for coding, research, and daily workflows. The strategy led to a rapid escalation in token consumption, with employees collectively using tens of trillions of tokens in a single month. An internal, employee-built dashboard called "Claudeonomics" had gamified usage with leaderboards and titles like "Token Legend," which reportedly encouraged excessive consumption. That dashboard has since been shut down.

To address the soaring costs, Meta is developing a centralized internal platform called AI Gateway. The system will provide real-time monitoring of...