Amazon employees 'tokenmaxxing' to show AI tool usage
Amazon employees are using the company's internal AI tool, MeshClaw, to automate non-essential tasks and inflate their token consumption, a practice dubbed 'tokenmaxxing.' This behavior stems from pressure to adopt AI after Amazon set targets for over 80% of developers to use AI weekly and began tracking token usage on internal leaderboards. Despite Amazon's assurance that token statistics won't affect performance evaluations, employees believe managers monitor the data, creating perverse incentives. The Seattle-based company has widely deployed MeshClaw in recent weeks, allowing AI agents to connect to workplace software. This push reflects Silicon Valley's drive to demonstrate returns on massive AI investments; Amazon is expected to spend $200 billion in capital expenditure this year, mostly on AI and data center infrastructure.
Key facts
- Amazon employees are 'tokenmaxxing' by using MeshClaw to automate unnecessary AI activity.
- MeshClaw is an internal AI tool deployed widely in recent weeks.
- Amazon set targets for over 80% of developers to use AI each week.
- Amazon tracks AI token consumption on internal leaderboards.
- Amazon says token statistics won't be used in performance evaluations.
- Employees believe managers monitor the data.
- Amazon is expected to spend $200 billion in capital expenditure this year.
- The majority of capital expenditure goes toward AI and data center infrastructure.
Entities
Institutions
- Amazon
- Financial Times
Locations
- Seattle
- United States