Microsoft curbs 'tokenmaxxing' among its own engineers
Parikh said he wants everyone focused on maximising outcomes that move the needle for customers and the business.
Microsoft wants its engineers to stop treating AI token usage as a badge of honour. An internal memo warns staff that tokenmaxxing, burning through as many tokens as possible, is not the goal and introduces new limits to bring spending under control.
Jay Parikh, an executive vice president at Microsoft, told employees in the email that they "need to be aware of how" they consume tokens on platforms such as GitHub. "Tokenmaxxing is not what we are optimising for," he wrote. "I want all of us focused on maximising outcomes that move the needle for customers and our business."
Why is Microsoft telling engineers to stop token maxxing?
Parikh said the company is "updating our internal guidance and managing token spend with the same discipline we apply to every other critical resource". As part of the shift, OpenAI's cheaper GPT-5.6 model will now be the default option for engineers, replacing pricier alternatives for everyday coding tasks.
Under updated Copilot guidelines cited in the memo, Microsoft divisions now operate against AI token budget targets, with employees given tools to track their own usage. No specific spending caps have been shared publicly, but the guidelines note that many engineers currently spend anywhere from hundreds of dollars a month to several thousand.
The new measures are in light of the introduction of GitHub's usage-based pricing for Copilot in April. The reason cited for the decision then was that there were increased computing costs associated with pricey AI agents.
Nevertheless, Parikh was careful to highlight the fact that Microsoft is still "AI-first" and that it is not backing down on its own AI endeavours. Nonetheless, the company has now joined many others who have started putting a curb on their skyrocketing AI spending.
According to Bloomberg, Uber had exhausted its entire annual budget allocated for AI spending within a few months before setting the upper limit at $1,500 per month per employee.
