Why does Amazon say not to use AI for its leaderboard? Amazon reportedly shut down an internal AI token leaderboard, KiroRank, after employees abused artificial intelligence to inflate their rankings. The move comes amid growing questions from tech giants about whether the increased use of AI is actually delivering measurable business value.
A story that has ignited a widespread discussion across the AI industry as it underscores a colossal challenge companies will grapple with in 2026: AI adoption vs. AI effectiveness.
Reports say Amazon executives told employees:
“Please don’t use AI just because it’s AI,”
The message is a move away from measuring AI usage metrics and toward measuring real-world business outcomes.
What Is Amazon’s KiroRank AI Token Leaderboard?
KiroRank was an internal dashboard that tracked AI tool usage by employees and AI tokens. Tokens are the input units of large language models (LLMs) such as Anthropic Claude, OpenAI GPT models, and Amazon’s internal AI systems.
The leaderboard was designed to incentivize AI adoption across teams. But there are reports of some employees getting into something called “token maxing,” where they’d invent AI requests just to inflate their token counts and move up the leaderboard.
Consequently:
- AI infrastructure costs went up.
- Staff concentrated on usage metrics, not results.
- AI-generated tasks seldom offered business value.
- Leadership was concerned about return on investment (ROI).
What is tokenmaxing?
“Tokenmaxxing” is a term to describe the practice of using AI tokens to the max to seem more productive or engaged with AI systems.
In Amazon’s case, employees reportedly used AI agents to perform trivial or unnecessary tasks just to boost their leaderboard positions. This created the classic performance-measurement problem: When a metric becomes the target, it ceases to be a useful metric.
Examples of Tokenmaxxing:
- Running AI prompts repeatedly with no business reason.
- Developing unnecessary AI-generated workflows
- Automating existing efficiencies
- Only utilize AI tools to increase token counts
The practice underscores the dangers of measuring AI success on raw usage data rather than on business impact.
Why does Amazon say not to use AI for its leaderboard?
The decision was largely fueled by two factors at Amazon:
1. Rising Costs of AI
Every AI prompt costs computational resources and GPU power. As AI adoption increases, token usage directly impacts operational expenses.
Industry analysts estimate that major tech companies, including Amazon, Microsoft, Google, and Meta, could spend between $650 billion and $700 billion on AI infrastructure in 2026, and spending could exceed $1 trillion by 2027.
These costs are worsened when employees make unnecessary AI requests that do not generate commensurate business value.
2. No Clear ROI
Many companies are now asking whether their big bets on A.I. are paying off in terms of productivity.
Tech executives are reportedly asking if AI spending is sustainable. Some organizations have reported that the cost of AI is outweighing the benefits being delivered.
Amazon’s move is part of a larger shift from the following:
- How much AI is used
- Tracking token counts
- Looking for adoption metrics
for:
- Monitoring business results
- Efficiency gain tracking
- Evaluating customer impact
Real-Time AI Industry Data (2026)
Key AI Adoption Statistics
| Metric | 2026 Data |
| Big Tech AI Infrastructure Spending | $650B–$700B |
| Projected AI Infrastructure Spending by 2027 | Over $1 Trillion |
| Amazon Developers Encouraged to Use AI Tools | 80%+ |
| AI Cost Concern Among Enterprises | Growing Rapidly |
| New Focus Area | AI ROI and Efficiency |
Sources indicate that companies are increasingly shifting from measuring AI activity to measuring AI effectiveness and profitability.
What This Means for Businesses Using AI
Amazon’s decision sends a strong message to organizations adopting AI:
Use AI to Solve Problems
AI should improve:
- Customer service
- Software development
- Marketing performance
- Business automation
- Operational efficiency
Rather than being used simply to meet adoption targets.
Focus on AI ROI
Companies should track:
- Revenue impact
- Productivity gains
- Cost reduction
- Customer satisfaction
- Time savings
instead of token consumption alone.
Avoid Vanity Metrics
But the numbers on AI use can be misleading if they’re not connected to measurable results.
Amazon’s experience shows that more AI activity doesn’t necessarily lead to better business results.
Why This Story Matters for the Future of AI
The Amazon AI leaderboard controversy is a microcosm of a growing trend in the tech industry. The initial rush to adopt AI is fading, and companies are now in a new phase where AI efficiency, AI ROI, AI governance, and AI cost optimization matter more than just raw usage numbers.
The lesson for companies that are investing in generative AI, machine learning, AI automation, and enterprise AI solutions is clear:
It is not about using more AI. The aim is to deliver more value with AI.
As the costs of AI infrastructure continue to rise, it’s organizations that focus on meaningful outcomes rather than token counts that will likely have the greatest competitive advantage over the coming years.