'Tokenmaxxing': the $72K race to use the most AI at work

A new status game is sweeping Silicon Valley — and it's spreading fast. It's called tokenmaxxing: the race to consume more AI tokens than anyone else at your company. And it's no longer just bragging rights. At some companies, your annual review now includes how much AI you actually use.

The term was popularized by Kevin Roose on the New York Times and the Hard Fork podcast. The idea: employees compete on leaderboards to show who's using the most AI — measured in "tokens" (the chunks of text that AI models process every time you send a prompt).

The Numbers Are Staggering

A tool called Tokscale has turned this into a global sport. It tracks AI token usage across Claude Code, Cursor, Codex, Gemini, and 11 other coding assistants — and publishes a public leaderboard. Here's what it reveals:

To put that in perspective, 222.7 billion tokens is enough text to fill Wikipedia 33 times over. And that's just one person.

When Token Counts Show Up in Your Performance Review

This isn't just a developer hobby. Meta became the first major tech company to formally tie employee performance reviews to AI usage in 2026. Your career advancement at Meta now depends, in part, on how effectively you leverage AI tools — and the measurable business outcomes from doing so.

NVIDIA CEO Jensen Huang took it further: he proposed giving engineers an AI token budget on top of their base salary. Tokens to deploy AI agents are becoming, in his words, "one of the recruiting tools in Silicon Valley."

The Human Cost Nobody Talks About

Not everyone is thrilled. A MetLife study of thousands of American workers found:

"Providing clarity around expectations and development is critical," said Shurawl Sibblies, MetLife's Chief Human Resources Officer. The disconnect is real: 91% of employers believe they fairly reward contributions, but only 65% of employees agree.

The Surprising Upside

Here's the twist nobody expected. A Gensler survey of 16,400 workers across 16 countries found that 30% of employees now qualify as "AI Power Users" — and they're actually more connected to their teams, not less.

AI Power Users spend less time working alone (37% of their week vs. 42% for others), more time learning (12% vs. 8%), and more time socializing (11% vs. 9%). "We often assume that more technology means less connection," said Janet Pogue McLaurin, Gensler's Global Director of Workplace Research. "But our data tells a different story."

Track Your Own AI Usage

If you're curious where you stand, Tokscale is free and open source. One command shows your usage across all AI coding tools:

npx tokscale@latest

To submit to the global leaderboard:

npx tokscale@latest submit

It supports Claude Code, Cursor, Codex, Gemini, OpenCode, Kimi, Qwen, and more. The global leaderboard is public — see where you rank among 471 developers worldwide.

Where This Is Heading

Tokenmaxxing raises uncomfortable questions. If AI usage becomes a performance metric, does that mean employees who prefer thinking through problems without AI will be penalized? If token consumption equals productivity, what happens when someone games the metric by running meaningless prompts?

For now, the race is on. Whether it's a leaderboard flex, a performance review checkbox, or a genuine productivity multiplier — AI token usage has become the newest workplace currency. And the people spending $72K on it clearly think it's worth it.

Related Content — Get Started with Easy Claude Code | Free Learning Guides | More AI News