Inside the halls of Meta, a new symbol has emerged that defines an employee’s standing among their peers. It is neither job title nor salary, but rather the amount of artificial intelligence processing units—technically known as “tokens”—an employee consumes. The use of artificial intelligence tools has shifted from a mere supporting technology to accelerate work into an open, frantic competitive arena among the company’s workforce of roughly 85,000 employees.
The leaderboard and the “Token Legend” title
A recent report published by “The Information” last Monday revealed an internal leaderboard created by employees to rank the top 250 users by their consumption of artificial intelligence computing power. The striking report noted that those who claim the top of this competitive ranking earn a prestigious title now known internally as the “Token Legend.”
To grasp the staggering scale of this hidden consumption, the internal dashboard recorded a cumulative consumption exceeding the 60 trillion token mark over the last 30-day period. This massive figure reflects a deeper cultural shift within Meta, where the adoption of artificial intelligence technologies has moved from encouragement and strategic guidance to expectation and job commitment, eventually transforming into something resembling a competitive sport where everyone races to prove their efficiency.
The strategy of shifting toward an AI-driven enterprise
This leaderboard is the latest manifestation of a strong push led by CEO Mark Zuckerberg to transform his tech giant into what he termed in statements an “AI-first company.” Building this direction required more than a year of continuous strategic planning and gradual execution across various departments and teams.
In late 2025, the company launched an innovative internal gamification and behavioral incentive program called “Level Up,” through which employees receive digital badges and moral rewards upon completing daily tasks using their available AI tools and models. The motivational push did not stop there; by early 2026, the company took a more decisive step by formalizing the use of artificial intelligence as a core and fundamental standard in periodic performance reviews, linking career progress and promotions to what senior management calls “AI-driven impact.”
Unprecedented engineering targets for writing code
According to leaked internal documents reviewed by “Business Insider” earlier this year, some engineering teams received explicit, unambiguous targets. For example, the company’s infrastructure division expects 65% of its engineers to write more than 75% of their approved code using AI tools by mid-2026.
Along the same lines, the scalable machine learning department set an ambitious goal for 50% to 80% of written code to be generated with direct assistance from AI models. These high figures definitively confirm that the role of the human programmer is gradually shifting to become closer to a supervisor and auditor of machine productivity, rather than being the exclusive author of code from scratch.
The “token maximization” phenomenon sweeps the tech sector
Meta is not the only company resorting to measuring employee efficiency and future-readiness through AI consumption. An analytical article published by the “New York Times” by author Kevin Roose last March documented the rise of a new spreading phenomenon across the tech industry known as “token maximization.”
The article detailed that many major tech companies, including Meta and OpenAI, maintain internal leaderboards that strictly track the number of tokens consumed by each worker. In one shocking case cited in the report, a single engineer at OpenAI processed 210 billion tokens in just one week—a massive volume of text processing sufficient to fill the entire Wikipedia encyclopedia 33 times consecutively.
Additionally, a single user of Anthropic’s “Claude Code” incurred token consumption fees exceeding $150,000 in a single month, highlighting the exorbitant financial cost accompanying this unprecedented technological race toward everyday workflow automation.
Restructuring: Boosting productivity or replacing humans?
This frantic race to consume AI computing power arrives at a very delicate time, as the company simultaneously lays off hundreds of employees as part of restructuring plans. These workforce reductions primarily aim to save liquidity and redirect investment spending toward building and developing the massive infrastructure required to run advanced AI models.
This unsettling synchronicity for some between laying off human workers on one hand and driving remaining employees to consume more machine power on the other raises fundamental questions and genuine concerns about whether the push for token consumption is a true strategy to boost productivity or merely a practical prelude and preparation for further workforce reductions and machine replacement in the near future.
«Job evaluations and bonuses primarily rely on the scale of positive impact an employee makes in their work, rather than just the raw and excessive use of artificial intelligence tools.»
To counter these criticisms, a company spokesperson rushed to respond to these growing concerns, confirming in statements that performance bonuses and promotions are fundamentally based on actual “impact,” added value, and tangible results the employee brings to the company, categorically denying that raw use of AI tools or burning millions of tokens pointlessly is the sole criterion for success and positive evaluation.
Frequently Asked Questions
What is the recently surfaced “Token Legend” title?
It is an informal internal title granted to employees who top the company’s ranking list in terms of their consumption volume of computing power and AI models, where an internal dashboard tracks the top 250 users and awards the highest consumers this title.
How did artificial intelligence become a standard for evaluating employee performance?
By early 2026, senior management began linking career progress and periodic performance reviews to the impact an employee achieves using AI tools, while also setting targets for engineers to write 50% to 80% of their code with machine assistance.
Does the growing reliance on tokens threaten human jobs?
The push driving employees toward excessive consumption of AI technologies coincides with notable workforce reductions aimed at redirecting funds to support technological infrastructure, sparking fears that this move paves the way for reduced reliance on humans. However, the spokesperson emphasized that evaluations rely on productivity and positive impact rather than random consumption.
