Your enterprise AI bill is probably higher than you think. While everyone's racing to adopt AI, a quiet storm of "token maxing" is brewing, inflating costs and challenging the promised ROI for smart, ambitious founders.

John Coogan, from TBPN, didn't mince words, pointing out the elephant in the room: “Token maxing dashboards have been reportedly led to potentially ROI negative AI use in at name brand companies like Meta.” Think about that. Even tech giants might be bleeding cash on AI initiatives that aren't paying off. He even cited reports of AWS spending “something like half a billion dollars in a single month” on AI-related costs. This isn't small change; it’s a reorientation of organizational spending.

The initial promise of AI was cheaper, faster work. The bull case, as Coogan explains, is that “the cost per task completed by AI will decrease very quickly.” And for many tasks, it does. But here's the kicker: as efficiency goes up, total spend often follows, not falls. We're seeing what some call "lumpy ROI" – massive investments that don't always yield immediate, clear returns. Is this justifiable for reorienting an entire organization, or is it misdirected spend on low-leverage problems? Brad Gerstner, also on the podcast, reminds us, "What does that tell us? It tells us that of course people optimize along the way but we are so early in the adoption curve." His point is valid, but early adoption doesn't excuse blind spending.

This is where an older economic principle and a social law collide, helping us understand why your AI costs are spiraling.

The Jevons Paradox and Goodhart's Law Applied to AI Token Spend

The conversation synthesized two powerful concepts to explain this seemingly irrational behavior in enterprise AI spending:

  • Jevons Paradox: When something becomes more efficient to use people often end up using more of it not less.
  • Goodhart's Law: When a measure becomes a target it ceases to be a good measure.

When This Works (and When It Doesn't)

The synthesis of Jevons Paradox and Goodhart's Law is in fact what's important if you're trying to run an efficient organization in the context of AI token spend. This framework is particularly sharp when evaluating internal AI initiatives aimed at boosting operational efficiency or automating specific tasks. If your goal is to reduce costs or free up human capital, these two laws are your early warning system. You'll catch scenarios where a 10x cheaper AI task doesn't save you money, but instead creates 100x more tasks, ballooning total costs.

However, this lens isn't always perfect. If you're building a new product or service entirely powered by AI, where the token spend scales directly with user engagement and revenue, the Jevons Paradox might simply reflect healthy growth. The key difference lies in whether the AI is a cost center you want to minimize or a revenue driver you want to maximize. For internal tools, always assume cost minimization is the goal, and scrutinize any spend increases through this framework.

What to Do With This

As a founder, don't just accept your AI budget line item at face value. This week, pull your company's AI token spend report from the last quarter. Don't simply look at the total. Instead, task your head of engineering or product to identify one internal AI workflow that's experiencing significant "token maxing" – perhaps a team now generating ten times more reports with AI than they used to manually. Apply the Jevons Paradox: Are engineers using AI for tasks they previously wouldn't, simply because it's now 'cheaper' per task, even if the total spend is higher?

Then, apply Goodhart's Law: If you've set a target like 'increase AI adoption by 20%,' are teams using AI for trivial, low-value tasks just to hit that metric? For that identified workflow, ask a pointed question: "For every dollar spent on tokens here, what is the measurable business value created, beyond just task completion?" You might find that even if a task is 'cheaper' per execution, the sheer volume of new, low-value executions makes the overall spend irrational, directly impacting your bottom line without a clear return.