AI economics library

AI Economics Wiki

A founder-focused guide to AI product boundaries, compute costs, metering, margins, and the evidence required for a defensible data advantage.

Use this hub to connect model architecture and product behavior to cost per successful outcome, pricing, gross margin, and long-run defensibility.

A practical sequence

Start here

Follow these concepts in order for the shortest path from first principles to a usable operating decision.

The full map

Library progress7/7

Track 01

Decide how much autonomy the product should have and how customers consume the underlying capability.

Choose whether AI advises a responsible human or acts inside a defined authority boundary.

Give customers managed access to AI capability without making them own the model and serving stack.

Track 02

Trace training, inference, capacity, and usage units from infrastructure through the customer bill.

Separate periodic investment in model capability from the compute spent each time that capability is used.

Translate capacity, utilization, throughput, and reliability into cost per successful customer outcome.

Connect provider usage, successful customer work, price, and margin through auditable metering.

Track 03

Measure cost per outcome, then test whether proprietary data creates an improvement loop rivals cannot cheaply reproduce.

Measure the revenue and directly attributable cost of delivering one successful AI outcome.

Test whether lawful, hard-to-replicate data creates better outcomes and a repeatable acquisition loop.

Built as a living reference

Published concepts stay linked from their track, while future additions enter the map with the context needed to use them well.

Open license

This content is available for reuse. When referencing or republishing it, please credit Dr. Sarah Zou and link back to the original source.

Licensed under Creative Commons Attribution 4.0 International. You may share and adapt the material with appropriate credit.