a16z: Don't Price AI Apps per Token, Price the Value
a16z Newsletter · rss · 2026-08-27
a16z argues that while token pricing makes sense at the model layer, importing it to the application layer is often a mistake.
Key Points:
- Don't anchor on tokens: Falling token costs make this a poor anchor for app value; it ignores data/workflow/orchestration value and makes costs unpredictable.
- Price at the value layer:
- Selling model access: Price tokens.
- Selling useful work: Price recognizable value units (often via Credits).
- Selling clear business outcomes: Price the outcome.
- Credits done right: Should map to customer-recognized work (e.g., small bug fix vs. feature), not hide tokens. Good systems abstract complexity, explain relative effort, and offer flexibility.
- Protect margins: Decouple "work value" from "delivery cost" to retain benefits from falling model costs.
More from Companies & People
- OpenAI kicks off first DevDay Exchange of the year in Bengaluru — paw_lean · 2026-10-15
- Dan Shipper: run product and labs in parallel, even with a labs team of one — danshipper · 2026-10-15
- AI Safety PhD Quits in Final Year After 15 Papers at Technion — IgorCarron · 2026-10-15
- Cohere to scale Aisle's agent-driven cyber defense for air-gapped critical infrastructure — aidangomez · 2026-10-15
- curl maintainer confirms: Aisle found 29 new CVEs in curl, rival Mythos found zero — aidangomez · 2026-10-15
- Cohere acquires Aisle, whose defensive cyber tech found 200+ High/Critical CVEs in curl, FFmpeg and more — aidangomez · 2026-10-15