Engineers at Microsoft now face new limits on AI tool spending. The company is introducing token budgets and updated guidelines. Leadership says the focus is on business impact, not maximizing AI usage.
Microsoft has implemented new internal restrictions on how much its engineers can spend on AI tools, signaling a shift in how the company manages its own use of artificial intelligence. The move introduces token budgets for employees using AI-powered products like GitHub Copilot, with leadership emphasizing that maximizing internal AI consumption is not the company’s primary objective.
This policy change positions Microsoft among the last major tech firms to formally curb employee AI spending, following similar steps by Amazon, Adobe, Atlassian, and Citi. The decision comes as the cost of AI usage has risen, while reported productivity gains have not always matched the investment. According to internal communications, Microsoft executives want teams to focus on outcomes that benefit customers and the business, rather than simply increasing AI usage-a practice some in the industry have called “tokenmaxxing.”
Jay Parikh, executive vice president at Microsoft, informed staff that the company is updating its internal guidance to manage AI token spend with the same discipline as other critical resources. As part of this update, Microsoft is making OpenAI GPT-5.6 the default model for internal use, citing its lower cost compared to other options. The company’s revised Copilot guidelines state that, starting July 2026, each division will have an “AI token budget target,” and employees will be able to track their individual AI spending.
While Microsoft has not disclosed a specific spending cap, internal data shows that many engineers currently spend from several hundred to a few thousand dollars per month on AI tokens. The guidelines note that further restrictions may be introduced as spending is monitored. The company’s leadership maintains that the new policy is not a response to financial pressure; Microsoft’s latest earnings report showed increases in revenue, operating income, and net income, surpassing Wall Street expectations.
Some employees have expressed concern that the new limits reflect the high cost of running AI infrastructure, with one anonymous staff member suggesting it raises questions about the affordability of Microsoft’s AI products for external customers. Microsoft leadership has stated that the company will continue to refine its AI policies as technology and products evolve, reiterating its commitment to becoming an “AI-first” organization. The focus, according to executives, is on achieving greater impact per token spent, not simply reducing token usage.
Microsoft did not immediately respond to a request for comment from 404media.
Microsoft, founded in 1975 and headquartered in Redmond, Washington, reported annual revenue of $211.9 billion for fiscal year 2025. The company employs over 220,000 people worldwide and owns GitHub, the platform behind Copilot. Microsoft has invested heavily in AI, including a multi-billion dollar partnership with OpenAI, and remains one of the world’s most valuable technology companies by market capitalization.
