# Moats, Wrappers & Commoditization

**Track:** Foundations of the AI-Native Venture — AI for Entrepreneurship — complete (29)
**Framework / surface:** venture strategy
**Level:** Beginner
**Prerequisites:** The New Cost Curve

**In one line:** What remains valuable when everyone rents comparable intelligence — and which layers above the model compound into companies.

## Theory, aesthetics & inspiration

When every competitor can rent the same frontier model, the model itself is no moat — so what remains valuable when everyone rents comparable intelligence? Hamilton Helmer's "7 Powers" (2016) still names the durable sources of advantage — switching costs, network economies, scale, brand, cornered resources, counter-positioning, process power — and none of them is "better weights." That is also the honest answer to "just a GPT wrapper": some wrappers are companies and some are features awaiting absorption, and the difference is whether the layer above the model accumulates anything — workflow depth, proprietary data exhaust, integration into systems of record, a customer relationship the lab will never service. Joel Spolsky's "Strategy Letter V" (2002) supplies the uncomfortable corollary: platforms commoditize their complements, and the labs treat application-layer cleverness as a complement to be absorbed. Every capability jump deletes the wrappers whose only asset was compensating for model weakness, and strengthens the ones whose asset was context the model cannot reach. The moat question is therefore not "is our AI better?" but "what do we own that gets stronger each time everyone's AI gets better?"

**Founder question:** What do you own that gets stronger when everyone’s model gets better?