AI Anxiety Syndrome Is Here! Multiple Companies Use KPIs to Push Employees Into Token Grind: Can't Stop Even During Rest

TechNews, March 25 — The workplace competition in the AI era has taken on a new form, with a frenzy around Tokens spreading across global tech companies.

Last week, an OpenAI engineer set an outrageous record, consuming 210 billion Tokens in just seven days. This amount is equivalent to reading the entire Wikipedia 33 times.

Tokens are the most basic unit of computation in AI, but now they have transformed from a technical term into a new currency in the workplace, infiltrating salary negotiations, performance evaluations, and promotion assessments.

Not only OpenAI, but another AI giant, Anthropic, reports that a user of their Claude Code product generated over $150,000 in compute bills in just one month. Stockholm-based software engineer Max Lindt admitted that his expenses on Claude even exceeded his salary, with his company covering these hefty Token costs.

What’s troubling is that these astronomical Token consumptions do not produce proportional output. The so-called maximization of Tokens is not about doing better work but simply about consuming more Tokens. Metrics once considered standards for engineers, like lines of code or submission counts, have become meaningless. Top tech companies like Meta, Shopify, and OpenAI now measure employee performance solely by one metric: how many Tokens they consume each day.

Tokens have quietly become a benefit, KPI, or even a fourth salary for workers. Insiders reveal that internal leaderboards tracking Token usage are everywhere at Meta and OpenAI, updating in real-time to show employees’ AI consumption, like a game leaderboard that motivates every worker.

In the job market, generous Token budgets have replaced traditional benefits like meal allowances and transportation subsidies, becoming a new attraction for talent. NVIDIA founder Jensen Huang predicted this trend early on, stating that Token budgets are as important as cash in an engineer’s annual salary, potentially accounting for half of it. He warned that if an engineer earning $500,000 annually consumes less than $250,000 worth of Tokens in a year, they should be warned.

This trend has also spread to domestic companies. Tencent and Alibaba are both offering Tokens as employee benefits, encouraging staff to use AI tools more freely. Alibaba provides large Token quotas for employees, allowing reimbursement for paid AI tools like Wukong and Qoder used in work and R&D. When Xiaomi launched MiMo V2, team leader Luo Fuli urged everyone to use the models more and even suggested that those who had fewer than 100 conversations could be deemed unfit to continue in their roles.

Companies investing heavily in Tokens are essentially exchanging this cost for ten times the work output. The instant, quantifiable, seemingly objective nature of Tokens makes them a favored KPI for managers, but it has also led to absurd artifice of busywork.

More worryingly, Token anxiety has completely taken over workers’ lives. Many tech employees give up leisure time, constantly running AI agents even without overtime, for fear of being eliminated due to insufficient Token consumption.

A San Francisco venture capitalist described how Token anxiety destroys young people’s social life and relaxation.

“To keep AI agents running 24/7, tech workers give up weekend fun, feeling guilty even when watching movies or reading novels because of idle compute. The old small talk at dinner—‘What project are you working on?’—is gone. Now, it’s about how many agents you’re running simultaneously.”

Token anxiety has become more intense than time anxiety. Behind companies pushing employees to compete for Tokens through KPIs is the internalization of anxiety by workers— even without overtime, they can’t stop. This has become a new mental health issue this year: if AI agents don’t consume Tokens, workers feel like they are wasting time.