© 01net/Canva
At Meta, an internal ranking displayed the 250 employees who consumed the most AI tokens. 60 trillions in 30 days. The table was removed after it leaked. The phenomenon, however, is only just beginning.
The term appeared in the columns of Hacker News at the beginning of 2026, before exploding in a New York Times article on March 20. The 'tokenmaxxing' refers to a practice born in Silicon Valley: consuming a maximum of AI tokens (Claude, ChatGPT, Gemini, Codex) to display productivity and climb internal rankings. Incidentally, it also justifies the billions invested in AI by the executives. A token is roughly three-quarters of an English word.
The Meta case: 60 trillion tokens and an honor board
In April 2026, The Information revealed the existence of "Claudeonomics," a dashboard created by a Meta employee on the intranet. The idea: track the consumption of Claude tokens by the company's roughly 85,000 employees and display the top 250 consumers. Gamified titles like « Token Legend » or « Session Immortal » rewarded the most voracious users. The champion: 281 billion tokens in 30 days (enough to fill Wikipedia 33 times, for scale). The dashboard was closed after leaks, but the phenomenon goes far beyond Meta.
At Amazon, 80% of developers must use the internal AI agent every week. Some launch unnecessary agents just to inflate the counter. At Salesforce, a Mac widget displays the personal AI spending in real time every 15 minutes, with a minimum target. Jensen Huang (Nvidia) even proposes integrating a token budget into salary packages, just like a hiring bonus. His formula: an engineer earning 500,000 dollars who doesn't consume 250,000 dollars of tokens should « deeply alarm him ».
Produce faster, but at what cost
The problem is that burning tokens doesn't mean producing better. A Faros AI study conducted on 22,000 developers shows that the rate of code deleted after being merged («code churn») has increased by 861% in teams with high AI adoption. In other words, we produce code much faster than before, and then throw it away much faster as well.
In France, the phenomenon remains so far discreet. No company has publicly deployed a similar leaderboard. But Mistral has just signed 40,000 licenses with the Caisse des Dépôts, and AI subscriptions are becoming widespread in French IT departments. Its co-founder, who was interviewed by the National Assembly, actually stated that «engineers at Mistral no longer write lines of code». The question will eventually arise here as well: how many tokens are burned, and for what result? The 1980s had the metric of lines of code. The 2020s have tokens. History is repeating itself, just with more zeros.
👉🏻 Follow real-time tech news: add 01net to your Google sources, and subscribe to our WhatsApp channel.
- Google Play Store
- 4K Video Downloader Plus
- VLC Media Player
- Adobe Reader
- uTorrent
- WinRAR
- Xender
- Google Chrome
- CCleaner
- Mozilla Firefox
- Free Avast
- ChatGPT
- Microsoft 365
- Opera One
- Avast One
- AirPods Pro 3
- Google Pixel 9a
- Google Pixel 10
- Google Pixel 10 Pro XL
- iPhone 17
- iPhone 17 Pro
- iPhone 16e
- Samsung Galaxy S25
- Samsung Galaxy S25 Ultra
- Samsung Galaxy A56
- Samsung Galaxy A26
- Samsung Galaxy A17
- Starlink
- Xiaomi Redmi Note 14 4G
- Xiaomi Redmi Note 14 Pro
- Xiaomi 15T Pro










- Google Play Store
- 4K Video Downloader Plus
- VLC Media Player
- Adobe Reader
- uTorrent
- WinRAR
- Xender
- Google Chrome
- CCleaner
- Mozilla Firefox
- Avast Gratuit
- ChatGPT
- Microsoft 365
- Opera One
- Avast One


