AI builders. How are you handling the compute cost vs pricing conversation with early users?

I recently had a conversation with a founder building an AI SaaS product.
The product was working. Users were active. Usage was growing.

And that’s where the stress started.

Every new user felt like a small win and a small liability at the same time.
More prompts meant more value delivered.
But also more tokens burned, more retries, more unpredictable costs.

Early users wanted simple pricing. “Just give me one number.”
The founder wanted clarity too. He just didn’t want to wake up one day realizing usage had scaled faster than revenue.

What struck me wasn’t the math. It was the tension.

Charge too little and you subsidize learning forever.
Charge too much and you scare away the very users helping you shape the product.

I believe, early pricing isn’t about perfect margins. It’s about learning where value actually sits.

That often means:

• Starting simple on the outside

• Tracking usage deeply on the inside

• Being honest about what’s expensive

• And accepting some messiness while patterns emerge

Predictability builds trust. Transparency builds patience.

How are you approaching pricing at this stage, and what has surprised you once real usage kicked in?