Microsoft has set AI token budgets for every division and made OpenAI's cheaper GPT-5.6 its default internal model, according to an internal email from executive vice president Jay Parikh to his Core AI teams, first reported by 404 Media. Parikh wrote that "tokenmaxxing is not what we are optimizing for," and said the company is managing token spend with the same discipline applied to other critical resources. He emphasized he does not want to slow the company's march toward being "AI-first," framing the change as a value exercise rather than a cost-cutting one. As of July 2026, every Microsoft division has an AI token budget target, and employees can see their individual AI spending on an internal dashboard. Updated Copilot guidelines note that further restrictions may follow. Tokenmaxxing refers to treating AI consumption as a proxy for productivity, which led to internal leaderboards and frontier-model usage on problems that never needed it. Parikh said the company is not optimising for fewer tokens, but for more impact per token. The guidelines do not publish a target spend figure, but many engineers have been running up hundreds of dollars a month in tokens, with some reaching a few thousand. Microsoft's most recent quarterly results beat Wall Street on revenue, operating income and net income. One employee said the move feels like the ultimate admission that if Microsoft can't afford its own AI products, customers must manage.