Microsoft Cracks Down on AI Token Spend Amid Rising Costs
Microsoft's top AI executive has sent an internal email to his team warning about token spend on GitHub Copilot. Jay Parikh wrote that 'tokenmaxxing is not what we are optimizing for' and that employees should use AI more carefully. The company has implemented a new policy to monitor and manage token spend, with every division having a budget target.
The email comes after Satya Nadella admitted to being a 'tokenmaxxer' in June, stating that it's addictive but you have to step back when the novelty wears off to say, 'What is it that I'm trying to create?' Microsoft has been trimming its AI usage, cancelling most Claude Code licences in its Experiences and Devices group in May.
The company's internal Copilot guidelines suggest that many engineers spend anywhere from a few hundred dollars to a few thousand dollars a month on tokens. No hard cap figure has been shared yet, but further restrictions may follow as spending is monitored.
This move follows other companies like Amazon, Adobe, Atlassian, and Citi in introducing AI throttling or spend visibility measures.