The moat decides whether you can play. The asset decides whether you get paid.

Discovery is priced before anyone knows the molecule works. That single fact explains most of the value distribution in AI-enabled drug discovery.

Schrödinger co-discovered the compound that Nimbus sold to Takeda for $4 billion, and took home $147 million. The science was shared; the economics were not. Whoever holds the asset at the moment of clinical de-risking captures the step change in value.

A data moat tells you nobody can copy you. It says nothing about who pays you. Defensibility and value capture are separate questions, and in target discovery the gap between those two is the whole business.