Why Meta Employees Are Burning 60 Trillion AI Tokens to Climb KPI Rankings

In Silicon Valley a new form of "showy consumption" called tokenmaxxing has emerged, where the amount of AI tokens used becomes a productivity benchmark and a status metric for "AI‑native" talent.

At Meta, an employee‑built leaderboard named Claudeonomics aggregates AI usage data from more than 85,000 staff members and ranks the top 250 token consumers. The leaderboard mimics a competitive game, offering bronze, silver, gold, platinum and jade badges, as well as titles such as "Model Connoisseur" and "Token Legend". In the last 30 days the total token usage on the board exceeded 60 trillion tokens (about 45 trillion English words). The top individual burned an average of 281 billion tokens , which could translate to millions of dollars depending on model pricing.

Meta also provides an official token‑usage dashboard for software engineers, who are the biggest consumers, allowing other roles to monitor their own consumption. Internal anecdotes reveal engineers running AI agents for hours solely to increase their token counts.

Similar token‑driven cultures have appeared in Chinese tech giants. On March 16, Alibaba launched the Alibaba Token Hub (ATH) business unit focused on creating, distributing and applying tokens. The same month Tencent upgraded its MaaS platform to a TokenHub , integrating WeChat, Mini‑Programs, QQ and other services, and reportedly began ranking departments by token usage, with employees devising work‑flows that generate tens of thousands of tokens.

Executives have fanned the fire. Nvidia CEO Jensen Huang warned that an engineer earning $500 k annually but spending less than $250 k on AI tokens would be "shocked". Meta CTO Andrew Bosworth claimed a top engineer’s token spend matched their salary while boosting productivity tenfold, calling it a "no‑loss investment". Former Tesla and OpenAI researcher Karpathey highlighted token throughput as the key pressure point in AI‑driven programming.

Anthropic’s recent financial disclosures illustrate the broader market impact: since 2026 its annualized revenue from Claude API calls has surpassed $30 billion , up from $9 billion at the end of 2025, with over 1,000 enterprise customers spending more than $1 million each year. The company secured multi‑gigawatt TPU capacity from Google and Broadcom starting in 2027 to support this growth.

These developments suggest that token consumption is becoming the most concrete indicator of AI demand, reshaping competition among large‑model firms from pure model capability to the efficiency of producing, processing and bearing tokens.

Reference links: https://www.theinformation.com/briefings/anthropic-says-topped-30-billion-annualized-revenue?rc=jn0pp4, https://www.theinformation.com/articles/meta-employees-vie-ai-token-legend-status