Meta employees are engaged in a wild competition: Who uses the most tokens?

```

Inside Meta, burning the most AI compute power is becoming a new status symbol.

On April 6, The Information reported that Meta Platforms has an internal AI usage leaderboard called "Claudeonomics." Built by employees on the company intranet, the list tracks AI token consumption for over 85,000 staff members and features the top 250 "super users." Those ranking highly can earn titles like "Session Immortal" or even "Token Legend."

The name "Claudeonomics" comes from Anthropic's flagship AI product, Claude. According to a copy of the leaderboard obtained by the media, total token usage recorded over the past 30 days exceeded 60 trillion. The top individual user averaged consumption of 281 billion tokens—depending on the model, this amount could cost millions of dollars.

Based on Anthropic's latest publicly available pricing, the average cost for input/output tokens with the Claude Opus 4.6 model is about $15 per million tokens. Estimating from this, 60 trillion tokens translates to roughly $900 million. However, which models Meta actually uses and at what prices remain unclear.

"Burning Tokens" Becomes the New Measure of Productivity

This phenomenon reflects the emerging "tokenmaxxing" culture in Silicon Valley—using token consumption as a productivity benchmark and as a competitive metric to determine if employees are "AI native."

Tech executives are backing this trend.

Nvidia CEO Jensen Huang said last month that if an engineer earning $500,000 a year spends less than $250,000 annually on AI tokens, he would be "very cautious."

Meta CTO Andrew Bosworth stated at a tech conference in February, as Forbes reported, that top engineers spend an amount equivalent to their salary on AI tokens, increasing productivity by up to tenfold. Bosworth bluntly said, "This is a guaranteed winning deal, keep doing it, there’s no upper limit."

Former Tesla and OpenAI lead AI scientist, now founder of an AI education startup, Andrej Karpathy, stated on a podcast last month: "The key is tokens. What's your token throughput? How much token throughput can you mobilize?"

How the Leaderboard Works

Employees can track their personal usage on the leaderboard, compare horizontally with colleagues, and earn gamified rewards—from bronze, silver, gold, platinum to emerald badges, as well as achievement titles like "Model Connoisseur" and "Cache Wizard."

According to two current employees, some staff run AI agents for hours to perform research tasks and maximize token consumption to climb the ranks.

Meta also has an official independent token usage dashboard for software engineers, and other staff can also check their usage. Notably, according to an insider, neither Zuckerberg nor Bosworth have made it into the top 250 super users.

On the tool front, Meta employees can use models from Anthropic, OpenAI, and Google, as well as internal tools like Meta's version of OpenClaw (called MyClaw), and Manus, which Meta recently acquired.

A Meta spokesperson said: "As is well known, this is a company priority. We are focused on using AI to help employees with daily work."

Doubts: Does Consumption Equal Productivity?

This competition is not without controversy.

Bloomberg journalist Joe Weisenthal asked directly on X platform: "What’s the point of measuring productivity by total token consumption?"

He went on to mock: "This really gives off a ‘real backyard steel furnace vibe’." He meant the fanaticism of chasing numerical metrics while ignoring actual quality, resembling reckless resource waste.

The skepticism points to a core issue: Token consumption is an input metric, not an output metric. Like measuring employee efficiency by how much paper they print, burning more tokens doesn’t equal producing more results. Some employees letting AI agents "idle" for hours to climb the leaderboard highlights how the metric can be gamed.

In response, well-known tech analyst Noah Brier offered a different view: "I don't think it makes sense, but when you’re trying to steer a giant organization like Meta, sometimes you have to ‘purposely overcorrect.’"

However, Weisenthal immediately followed up: "Even so, what exactly are they trying to turn around—employees' work habits, or the company’s model for making money?"

From a market perspective, this phenomenon itself sends a clear signal: enterprise AI consumption is expanding at a pace far beyond expectations. For Meta alone, the estimated monthly spend on AI compute power could approach $900 million, meaning sustained demand for cloud and AI infrastructure providers.

Risk warning and disclaimerThe market has risks, investment needs caution. This article does not constitute personal investment advice and has not considered individual users’ specific investment goals, financial situation, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article suit their particular circumstances. Investment based on this is at your own risk. ```