Amazon kills its AI leaderboard — staff inflated token use to climb
TL;DR
Amazon employees gamed internal AI leaderboards by burning tokens on pointless tasks — «Tokenmaxxing». Meta's «Claudeonomics» board logged 60 trillion tokens in 30 days. Uber blew its annual AI budget in four months.
This isn't just Amazon. Amazon employees, chasing positions on internal AI usage leaderboards, started using AI tools for meaningless trivial tasks, artificially inflating token consumption — coined «Tokenmaxxing». The company has now restricted org-wide visible usage stats, reminding employees not to use AI for the sake of using AI.
Meta's «Claudeonomics» went further. A Meta employee built a leaderboard called «Claudeonomics» tracking ~85,000 employees' token consumption. Within 30 days, total usage exceeded 60 trillion tokens. Notably, CEO Mark Zuckerberg and CTO Andrew Bosworth weren't in the top 250. Pulled within days of media exposure.
Uber direct: blew the year's budget in four months. Uber's CTO disclosed in April that the company had already exhausted its full-year AI coding tools budget in four months — having previously actively encouraged maximizing AI tool use via internal leaderboards.
The pattern. Analysts call this textbook Goodhart's Law: when a metric becomes a target, it stops being a good metric. The four big tech companies' 2026 combined capex is closing on $700B. Each faces pressure to show investors AI adoption rates — token count became the perfect performance prop, until employees started manufacturing the numbers.
Deeper question: if a meaningful share of AI usage is performative, how reliable is the demand forecast that hundreds of billions in infrastructure investment depends on?
via Tom's Hardware
Meta's «Claudeonomics» went further. A Meta employee built a leaderboard called «Claudeonomics» tracking ~85,000 employees' token consumption. Within 30 days, total usage exceeded 60 trillion tokens. Notably, CEO Mark Zuckerberg and CTO Andrew Bosworth weren't in the top 250. Pulled within days of media exposure.
Uber direct: blew the year's budget in four months. Uber's CTO disclosed in April that the company had already exhausted its full-year AI coding tools budget in four months — having previously actively encouraged maximizing AI tool use via internal leaderboards.
The pattern. Analysts call this textbook Goodhart's Law: when a metric becomes a target, it stops being a good metric. The four big tech companies' 2026 combined capex is closing on $700B. Each faces pressure to show investors AI adoption rates — token count became the perfect performance prop, until employees started manufacturing the numbers.
Deeper question: if a meaningful share of AI usage is performative, how reliable is the demand forecast that hundreds of billions in infrastructure investment depends on?
via Tom's Hardware
