Microsoft caps AI token use as costs triple despite 98% price cut

Microsoft executives have warned that corporate AI spending has surged even though the price of AI tokens fell about 98% since 2022. Average AI bills for enterprises are now three times higher because employees use far more tokens when AI automates complex tasks.

In an internal email disclosed by 404 Media, Executive Vice President Jay Parikh announced that, starting July 2026, the company will impose department‑level token budgets and switch the default internal model to a cheaper OpenAI offering. Employees can track their personal token consumption on a company dashboard, and many engineers are already spending several hundred to a few thousand dollars each month on AI services. Parikh described the unrestricted practice of “tokenmaxxing” as something the company will no longer optimize.

The move reflects Microsoft’s effort to control rapidly rising AI costs despite the market‑wide drop in token prices, and underscores the broader challenge of managing AI expenses in large enterprises.