Distribution Is the Moat
with Partner, generalist fund
A venture partner makes the case that distribution, not capability, is the durable moat in AI.
- 01 Why the application layer is undervalued again
- 02 Channels that compound
- 03 The contrarian portfolio cut
My guest is a partner at a generalist fund with a significant AI portfolio. They came on to argue a position I find increasingly persuasive: that the open-source-eats-everything thesis has stalled, and the durable moats in AI are distribution moats, not capability moats.
The conversation covers what “winning the channel” actually means in vertical AI, why the application layer is undervalued after the 2024 correction, and how to structure a portfolio around distribution bets rather than model bets.
Capability commoditizes on a timeline measured in quarters. Distribution compounds on a timeline measured in years. Bet accordingly.
Key Moments
- 04:15 — Why the application layer is undervalued again after the 2024 correction
- 16:40 — Distribution channels that compound — and the ones that do not
- 25:30 — The contrarian portfolio allocation: fewer model bets, more channel bets
- 33:10 — What “winning the channel” looks like in vertical AI
Takeaways
This is a shorter episode — under 40 minutes — but it is dense. The framework for evaluating AI companies by their distribution position rather than their model quality is one I have been using since this conversation, and it has changed several of my investment opinions.