The Venture Half-Life

There is a comfortable fiction in venture: that every position is a future outlier waiting to be discovered. The arithmetic of power laws says otherwise. Most of what we underwrite will not compound. The work is not to deny this — it is to know, early and without sentiment, which half of the book we are holding.

We borrow a term from physics. Every venture position has a half-life: the time over which its probability-weighted value decays absent new evidence of compounding. A position that has not produced that evidence is not neutral. It is decaying, quietly, on the balance sheet of our attention.

We underwrite the cost curve. Narratives are cheap; unit economics are not.

Evidence over narrative

The signal we underwrite is rarely the pitch. It is the slope of the cost curve after the first cohort of customers. Does the next deployment cost less than the last? Does reliability improve as scale increases, or does complexity tax it away? A business that gets cheaper, safer, or stronger as it grows has a half-life that extends. One that does not has a half-life that contracts, regardless of how large the addressable market looks on a slide.

EXTENDS falling unit cost · improving reliability at scale · pricing power CONTRACTS subsidized growth · complexity tax · narrative outrunning evidence

Concentration is honesty

When the evidence arrives, we concentrate. Doubling down on a position that is demonstrably compounding is not aggression — it is honesty about where the edge actually lives. The hardest discipline is the inverse: declining to reflexively support a position that has stopped producing evidence, even when sunk cost and relationship pull the other way.

The venture book is not a collection of bets. It is a living estimate, revised constantly, of where conviction has earned the right to compound.