The Trouble with ‘Tokenmaxxing’: Amazon Tells Engineers to Stop Chasing AI Usage Metrics
Reported by YourStory.com, Tom's Hardware, The Register and 4 more outlets7 sources ↓
Amazon told engineering teams to judge AI use by results for customers and the business, not token consumption. The company warned that treating AI usage metrics as targets could raise costs and make software more complex.
Sources
- YourStory.comRead ↗Amazon tells devs to stop tokenmaxxing and write real code
- Tom's HardwareRead ↗Amazon tells engineers to stop ‘tokenmaxxing’ to look productive
- The RegisterRead ↗Amazon tells tokenmaxxing devs to slow their AI roll
- BigGo FinanceRead ↗Amazon Tells Engineers to Quit 'Tokenmaxxing' as AI Usage Metrics Backfire
- FortuneRead ↗Amazon warns against tokenmaxxing and tells engineers to use their judgment when using AI
Details
The Register reported that Amazon made the guidance public in a note to engineering teams on Tuesday. The memo, written by Amazon vice president and distinguished engineer Terk Terkowitz, said developers’ roles and output should not be defined by their use of AI.
It warned that using them as targets could increase costs and software complexity, and said engineers should use judgment about where current AI tools help their work.
An Amazon spokesperson confirmed to The Register that the memo was intended to discourage tokenmaxxing. The company also urged teams to stay connected to their customers and business needs; the memo said that finding AI does not help a task is not a failure.
Written from coverage as of
Stocks in this story
Change since open: 20 Nov to 21 Nov 2026
Last refresh