The Pricing of Payment APIs

About this paper

The first two papers in this series looked at categories that price by the unit consumed — a token, a message, a minute. Payment APIs price something fundamentally different, and it’s the most important pricing idea in the whole series: they charge a percentage of the value moved. It costs Stripe nearly the same to process a $5 charge and a $500 charge, but it earns a hundred times more on the second. That’s the take-rate model — the purest value-based pricing in the API economy, and exactly where I argued AI inference is slowly heading.

This paper is a market breakdown — real flat-rate and interchange-plus rates, and the five-layer fee stack underneath both, of which the processor’s own cut is the smallest layer — and a pricing playbook for your own API: price value not cost, separate cost-recovery from value-capture, itemize every pass-through layer honestly, and let the plan decide the price. Even if you never touch a credit card, this is the most lucrative pricing structure in the industry, explained.

What's inside

1. The unit — percentage-of-value plus a fixed fee, and why payments price value not cost
2. The market — real flat-rate and interchange-plus rates, and the spread
3. The five-layer stack — interchange, scheme, markup, gateway, FX, and how little the processor keeps
4. Flat-rate versus interchange-plus — bundling versus transparency, and who each one taxes
5. Who you're for — enterprise minimums versus self-service, plans deciding pricing
6. Pricing your own API — the take-rate model, value over cost, pass-through honesty
7. Anti-patterns I watch for
8. Self-assessment
9. Where this is going — machine-readable rates and agentic commerce

What you get for $25.00