Articles

AI Strategy

Your Team Is Burning Tokens. That Is Not the Same as Creating Value.

Rising AI usage can look like progress, but token consumption does not prove better decisions or business outcomes.

Bharat Kumar6 min read
Executives compare AI usage metrics with a business outcomes display
Table of contents

Everyone is talking about how much AI their teams are using.

Almost nobody is talking about whether any of it is producing a business result.

Three of the biggest technology companies in the world just gave us the clearest answer yet. And the pattern is uglier than the headline.

The scoreboard that rewarded the wrong thing

At Meta, an employee-built dashboard ranked roughly 85,000 workers by token consumption. It was taken down. But not before it shaped behavior.

At Amazon, employees reportedly assigned agents unnecessary work to lift their usage scores.

At Uber, teams burned through the annual AI coding budget in four months. Heavy users were running $500 to $2,000 a month per person.

None of this was sabotage. It was exactly what the scoreboard rewarded.

When you measure token usage, people use more tokens. When you measure prompts, people write more prompts. When you measure agent actions, people set agents loose on work that did not need doing.

That is not transformation. That is dashboard compliance.

The numbers behind the failure

PwC surveyed 1,217 executives and found that the top 20% of companies captured 74% of AI's economic value. They were twice as likely to redesign workflows rather than simply add tools.

Gartner estimates that a task costing a chatbot $0.01 can cost an AI agent up to $1.50. Same task. Different architecture. Same silence on whether the outcome changed.

McKinsey found 80% of employees say AI made them more productive. Only 37% of companies can find any of that in EBIT. Essentially unchanged from last year.

The spend is real. The value is not showing up where it matters.

The real gap nobody owns

Most companies measure AI activity.

Prompts generated. Tokens consumed. Documents produced. Agents deployed.

Almost nobody measures decision quality.

Which call got better because of this tool. Which number moved. Which risk was caught. Which customer outcome changed.

The reason is not laziness. Activity is easy to count. Decision quality takes longer to see and is harder to defend in a review.

So the scoreboard defaults to what is countable. And the scoreboard shapes the behavior.

Why this keeps happening

1. Consumption feels like progress.

More prompts feel like more work. More tokens feel like more investment. Both feel like forward motion. Neither proves anything about outcomes.

2. Cost ownership disappears the moment it scales.

When one person uses a $20 tool, finance does not care. When 5,000 people run agents across production, the bill becomes a line item nobody is accountable for. Uber discovered this in four months. Most companies will discover it in a quarter.

3. The metric becomes the goal.

The moment a leader says "We need 80% of developers using AI every week," the metric stops describing reality and starts replacing it. That is what happened at Meta and Amazon. The number went up. The value did not.

The AI Value Equation, extended for the consumption era

The five questions from earlier issues still apply. There is a further one that has to be answered before a single token is budgeted:

What outcome improves because of this usage, and who is accountable for proving it did?

Not "how many tokens were consumed."

Not "how many agents are running."

A specific business result, tied to a specific decision, owned by a specific person.

If the honest answer inside your organization is "we are still figuring out how to measure that," then you are not measuring usage.

You are paying for activity and calling it progress.

Decision of the Week

Before your next AI spend review, ask one question in the room:

If our AI usage doubled tomorrow, which business metric would improve first, and who would notice?

If nobody has an answer, you do not have an AI program.

You have a subscription.

Thanks for reading Issue #9 of Strategic Edge.

Every week, I will break down a real business event, not to report the news, but to uncover the strategic decisions hiding beneath it.

Because in the AI era, technology is becoming abundant. Sound judgment is not.