Most of what matters in venture happens before the story is written.

By the time a shift is on the front page, it is consensus. And once everyone agrees something is a good idea, the advantage in it is mostly gone. The important work happened years earlier, when it was still hard to explain.

SpaceX spent its best years being ignored. The uncovered years are where the position gets built.

The same is true in AI right now. The loud layer is easy to build, which is exactly why it is hard to win in. When capability becomes obvious, it stops being the moat. The edge moves to what is still genuinely hard: the regulated, the physical, the domain-deep, the problems a model alone does not solve.

So I do not underwrite consensus. I underwrite difficulty.

The harder something is to build, the longer its advantage lasts once it is built. That is where I spend my time, on the companies the front page has not discovered yet, and I wrote last week about where I think that edge is moving.

Two things still earn venture money: technology that is hard to build, or a market position that is hard to catch. Everything in between is a good business. It is just not a venture bet.

Full piece in the first comment. Grove Ventures