AI doesn't lower SaaS prices. It widens margins.

The reflexive take on AI in software goes like this: anyone can build anything now, the moat is gone, prices race to zero, and SaaS as a category gets compressed into a thin commodity layer. It's a clean narrative. It's also, for any operator who has actually run the numbers, wrong.

AI does change the cost structure of software. It changes it in the operator's favor, not the customer's.

Lower cost to build is not the same as lower price to charge. Those are two different lines on two different income statements.

The unit economics of AI-native SaaS

A five-person AI-native team can ship what a fifty-person team shipped five years ago. That fact is real. What people miss is that the customer doesn't know or care. They're not buying engineering hours. They're buying an outcome: more pipeline, fewer churned accounts, a cleaner close.

If anything, the willingness to pay for those outcomes has gone up, because every buyer is now under pressure to do more with a smaller team and a tighter budget. The product that credibly replaces a headcount is worth more, not less, than the product that merely assists one.

Where AI savings actually go in a SaaS P&L

In a well-run AI-native SaaS business, the cost-side gains from AI show up in three places, and the order matters:

1. Gross margin. First and most importantly, the gap between what the customer pays and what it costs to serve them widens. That margin is what funds everything else, and it's what makes the business durable when the next cycle turns.

2. Product velocity. Second, the team ships more, faster. Not in features-per-quarter theatre, but in actual customer outcomes delivered. That velocity becomes a moat in its own right, because the gap between you and a slower competitor compounds.

3. Price discipline. Only after the first two are secured does it make sense to use any of the savings to expand into adjacent segments at lower price points. And even then, you're expanding the market, not discounting the core.

The operator's playbook for pricing AI-native software

The operators we back at Cobalt Glacier all run this play, whether or not they'd describe it in these terms. They use AI to widen margin first. They reinvest the margin into product velocity. They protect price. And they let the slower, less disciplined competitors race each other to the bottom of a market that no longer exists by the time they get there.

The companies that compound for the next decade in B2B software won't be the cheapest. They'll be the ones that quietly turned every AI-driven cost gain into a structural advantage, and kept charging what the product is worth.