The science is the moat

Most venture capital is optimized to evaluate distribution, not physics. That is a reasonable bet in software, where the hard part is usually getting a working product in front of the right customers. It is a bad bet in frontier technology, where the hard part is whether the thing works at all.

I spent a decade at JPL building instruments that had to survive contact with the real world: reflectometry systems, a NASA mission, receivers that either closed the link budget or did not. That work leaves you with a specific instinct. You learn to ask whether the demo generalizes, whether the physics scales, and whether the team actually understands the failure modes they are waving away.

What most investors can’t price

When the science is genuinely hard, the market underprices the company for a simple reason: the people writing checks cannot evaluate the moat. They fall back on proxies — logos, pedigree, narrative — because the underlying claim is illegible to them.

That illegibility is the opportunity. A company whose defensibility lives in something only a handful of people on earth can assess is, almost by definition, mispriced early. The moat is real; it is just invisible to the median diligence process.

Where I look

• The claim is falsifiable. The best deep-tech founders can tell you exactly what would prove them wrong, and have already run that experiment.
• The hard part is the whole company. If the technology works, everything else is execution. If it doesn’t, no amount of go-to-market saves it.
• The team has scar tissue. People who have shipped hardware, or run a mission, or watched a system fail in the field, price risk differently than people who have only modeled it.

If you’re building something the market underestimates because it can’t read the science, that’s precisely the room I want to be in early. These notes are where I think out loud about the companies and categories I’m watching.