Discussion about this post

User's avatar
Sebastian AB's avatar

Agreed on the statistics of many shots on goal (Andrew Lo et al) — the main bottlenecks are:

1) Contrived unproductive animal models of disease. WT aging and lifespan would be better

2) needlessly overpriced clinical trial costs in Western countries (high CRO margins)

Ket Mhatre's avatar

Best version I've read of the value-migration thesis: stand at the bottleneck rather than in the path of the improving capability. The tension I keep returning to is that your sharpest point cuts at your own platform. If proving a PoS edge takes 10–20 prospective readouts over a decade, then Delphi can't validate its own predictions any faster than the clinic resolves them. which leaves the flywheel unfalsifiable on exactly the timescale you're underwriting, no better off than anyone else's prediction stack. That puts the observable near-term moat on the Apollo and capital side, where the advantage shows up before the readouts land, rather than on prediction. So I'd be curious whether you treat Delphi as a genuine edge today, or as a decade-long bet you're underwriting at industry averages until it proves itself out.

Ready for more?