According to leaked internal memos monitored by Beating, Meta Platforms plans to restrict employee AI token usage to control soaring internal costs. The company is building a central AI Gateway to monitor real-time employee AI consumption, set budgets, and cap token usage. Meta projects internal AI use alone will generate billions of dollars in costs during 2026.
To further reduce expenses, Meta is shifting focus toward in-house AI tools. The leaked memos show Meta is pushing employees to phase out third-party programming tools like Anthropic's Claude in favor of MetaCode, its internally developed coding assistant. While external model access remains available, Meta plans to implement stricter budget and quota approval mechanisms through its self-built gateway.