[JumpCrisscross]

> finds that "doing AI" is not a stable enough differentiator to compel him as a VC

This is a better thesis than the article's. Uber for X / AI for Y is not a valid pitch.

But the article goes further. It surmises the only two valid moats are a capital advantage (compute) or intellectual property (data). These are the most trivially-verifiable moats for a third party. Which makes sense for a VC to prioritise them. But they're far from the dominant mode of differentiation.

Plenty of "AI" start-ups will do well because they found a niche, had the right team to sell to it, and developed quickly enough to keep customers hooked. They won't win because of AI per se. But they won't lose for lack of access to more compute or special data either.

[danielmarkbruce]

Exactly. Scale, stickiness and getting the details right via focus are all things.

The author seems to be misguided - no one is really suggesting AI is a way to win. It's a value prop thing, not a competitive thing. SaaS was a value prop thing, not a competitive thing. Mobile was a value prop thing, not a competitive thing. WWW was a value prop thing, not a competitive thing.