Microsoft Caps Engineers' AI Spend, Defaults to Cheaper GPT-5.6 Model
Updated
Updated · 404 Media · Aug 4
Microsoft Caps Engineers' AI Spend, Defaults to Cheaper GPT-5.6 Model
3 articles · Updated · 404 Media · Aug 4
Summary
Microsoft has imposed new internal limits on engineers’ AI-tool spending and set OpenAI’s cheaper GPT-5.6 as the default model for employee use.
July 2026 guidance gives each division an AI token budget target and lets employees track individual usage, with many engineers currently spending from hundreds to a few thousand dollars a month.
Jay Parikh told staff the goal is not to maximize token use but to get more impact per token, reflecting rising AI costs and uneven productivity gains rather than any cash squeeze.
The move puts Microsoft alongside Amazon, Adobe, Atlassian and Citi in curbing costly employee AI use, even as the company says it still aims to become more AI-first.