Oracle says they removed cost uncertainty. They didn’t.

Oracle has introduced “AI Units” to simplify pricing.

But “AI Units” don’t stabilize cost.

Because the problem isn’t the price of a unit.
It’s how many units a real workflow actually consumes.

In SaaS, you can model this:
users → seats → predictable usage → stable margins

In AI, you can’t.

Because cost isn’t tied to a user.
It’s tied to execution.

And execution expands.

A single interaction can:

• branch
• retry
• call multiple tools
• trigger downstream processes

So the same “task” can consume very different amounts of work.

That’s where the uncertainty comes from.

AI Units don’t fix that.

They just bundle the cost into a cleaner label.

To be fair, Oracle did fix something:

• they reduced frictions
• implified SKUs
• made it easier to experiment

But they didn’t solve the core issue:

• AI behaves like a utility
• Cost scales with usage
• usage expands dynamically during execution

So instead of removing uncertainty,

they moved it.

From:
“what does a unit cost?”

To:
“how many units will this actually consume?”

And that’s the part no pricing model can answer —
because the variability isn’t in pricing.

It’s in the system itself.

Until cost is measured against actual compute
and constrained before execution, uncertainty doesn’t go away.

It just gets harder to see.

– Published on Tuesday, March 31, 2026