The Pulse: ‘Tokenmaxxing’ as a weird new trend

2026-04-23 · Source: The Pragmatic Engineer · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Emerging Technologies & Innovation · Depth: Intermediate, medium

Summary

Meta Platforms recently abolished an internal "token leaderboard" that ranked employees by AI token usage, a practice dubbed "tokenmaxxing." This system, which aggregated AI usage from over 85,000 Meta employees, led to significant waste, with engineers reportedly burning massive amounts of tokens for minimal outcome and even causing outages due to careless AI code generation. The Information reported Meta employees used 60.2 trillion AI tokens in 30 days, potentially costing hundreds of millions of dollars. While Meta removed its leaderboard after social media backlash, some engineers speculate the initial goal was to incentivize AI usage to generate real-world traces for training Meta's next-generation coding models. Microsoft and Salesforce also employ similar internal token tracking and incentive systems, leading to engineers deliberately inflating token usage to avoid being perceived as "AI-unnative" or to meet minimum spend targets. In contrast, Shopify successfully implemented a usage dashboard with circuit breakers and manager oversight to encourage AI adoption without fostering wasteful tokenmaxxing.

Key takeaway

For CTOs and AI Product Managers evaluating internal AI adoption strategies, avoid implementing raw token usage leaderboards or minimum spend targets. Such metrics incentivize wasteful "tokenmaxxing," inflating costs and potentially degrading product quality, as seen at Meta, Microsoft, and Salesforce. Instead, focus on qualitative outcomes and integrate circuit breakers for runaway spend, similar to Shopify's successful model, to foster responsible AI integration and prevent unnecessary expenditure.

Key insights

Gamified AI token usage metrics can lead to massive waste and counterproductive behaviors within organizations.

Principles

• Metrics drive behavior, for better or worse.
• Incentivizing raw usage can lead to inefficiency.
• Oversight and circuit breakers prevent runaway costs.

Method

Shopify's approach combines a usage dashboard for adoption, circuit breakers for cost control, and manager review of high-spenders to prevent tokenmaxxing.

In practice

• Implement circuit breakers for AI token spend.
• Review high AI usage for genuine value.
• Avoid gamifying raw token consumption.

Topics

• Tokenmaxxing
• AI Token Usage
• Internal Leaderboards
• AI Productivity Metrics
• Generative AI Waste

Best for: CTO, AI Product Manager, VP of Engineering/Data, Software Engineer, Director of AI/ML, Consultant

Related on AIssential

• Amazon employees are "tokenmaxxing" due to pressure to use AI tools — Internal AI usage metrics can create perverse incentives, leading employees to inflate activity rather than optimize genuine productivity.
• In Defense of Tokenmaxxing — Experimentation with AI token consumption is crucial for enterprise adaptation to agentic AI, despite potential for misuse.
• The Annual AI Slowdown Panic is Here — AI's current "slowdown panic" reflects market adaptation to scarce compute and usage-based pricing, not a bubble burst.
• More Compute Is The Answer — Misguided AI performance metrics, like token count as a KPI, hinder actual outcomes despite increasing compute capabilities.
• Why Uber has Already Burned Through its AI Budget — Aggressive AI tool adoption can rapidly deplete budgets, necessitating cost management and strategic expansion.
• ‘Tokenmaxxing’ is making developers less productive than they think — Increased AI code generation correlates with disproportionately higher code churn and revision rates, undermining productivity claims.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The Pragmatic Engineer.