Today, almost all AI startups charge per use: “Pay €X for Y tokens.” It's simple. It's clear. But it is also dangerous.

Why ?

Because tokens are a technical unit, not a unit of value. And when you charge on a metric your customer doesn't value, you reduce your solution to a commodity.

The risk is that AI will be perceived like electricity: a measured, interchangeable flow that ends up devaluing itself. While its true value lies elsewhere: in the results, the precision, the productivity, the ROI generated.

What are the takeaways?

If AI wants to avoid becoming a simple “utility”, it must rethink its pricing. Not “per minute of viewing” like Netflix could have done… but to the business value produced. And this value is not limited to “obvious” metrics like time saved or sales generated.

It can also come from non-intuitive but critical factors:

• Trust in a recommendation,

• Lower stress in an automated process,

• The perceived simplicity of an interface that drives adoption.

These “invisible values” could tomorrow become real pricing units. And when that day comes, AI will no longer be sold as a commodity but as a transformative experience.

As long as we remain stuck in the “token mindset,” we trivialize what should be differentiating. And a commoditized product always ends up losing its value.