Stage capital to physics, not to growth.

A software company raises when its metrics say it earned the right to. Users, revenue, and retention climb far enough to justify the next round at a higher price, and the round arrives as a reward for traction already shown. A hard-technology company often has none of those metrics for years, and the capital still has to arrive — which means it can't be priced off growth. It has to be priced off proof: the specific technical result that says the thing is one step closer to working.

In practice this means capital is staged against engineering milestones rather than released against a growth curve. A tranche unlocks when the engine completes a full-duration firing, when the model clears the failure cases that actually matter, when the cell survives its cycle target — each event chosen in advance because it retires a specific, nameable risk. The company isn't being paid for getting bigger. It's being funded to answer the next question that determines whether it can exist at all.

What staging to proof selects for

Funding against technical proof changes who can credibly provide the capital. Pricing a round off user growth is a skill a generalist can develop; judging whether a firing test or a model evaluation actually retired the risk it claimed to requires understanding the physics well enough to tell a real milestone from a cosmetic one. This is why hard-technology capital concentrates among investors with technical depth on staff or on call — not as a matter of taste, but because the funding model doesn't function without someone who can read the result. An investor who can't evaluate the milestone is left trusting the founder's account of it, and trust is not underwriting.

A software round is priced off what a company has already shown. A hard-tech round is priced off what it has just proved it can survive.

The discipline cuts toward the company too, and that is the part most decks skip. Milestone funding protects the investor from committing a decade of capital to a science that never works; it also forces the company to define, before the money moves, what would count as failure — a conversation founders running on conviction tend to resist. The healthiest hard-technology financings are the ones where both sides argued hard about what the milestone actually was, because a vaguely defined milestone funds nothing and de-risks nothing. The clarity of the milestone is the quality of the underwriting.

The software model made capital patient by making it cheap to deploy and quick to judge. Hard technology can borrow neither property, so it substitutes a different one: money that moves only when reality confirms the last step, released by people who can read the confirmation. Growth tells an investor a company is winning. Proof tells them it is still in the game — and in hard technology, staying in the game long enough is most of the return.