Meta Cracks Down on Internal AI Costs as Bill Heads for Billions

Meta is hitting the brakes on its own employees' AI habits. According to an internal staff memo, the company plans to impose limits on token usage across its workforce after internal forecasts showed AI spending ballooning into the billions by 2026.

The crackdown includes encouraging employees to switch to MetaCode, Meta's in-house coding tool, rather than burning through tokens on external or unrestricted AI services.

It's a telling move. Meta has been publicly pouring tens of billions into AI infrastructure, data centers, and model development. But apparently, the internal consumption side of the ledger is getting out of hand too.

Capping employee usage signals that even for a company spending aggressively on AI, there's a limit to how freely that technology gets used behind closed doors. The bill matters — even at Meta scale.