Hard technology is always five years away.

Ask when a fusion reactor, a full self-driving stack, or a grid-scale battery chemistry will be commercial, and the answer has a way of staying constant while the calendar moves. It was five years away in 2015. It was five years away in 2020. Told sincerely each time, by people doing serious work, the estimate advances at roughly the speed of time itself. The pattern is common enough that parts of the industry have a name for it: the perpetual five-year technology.

The reason isn't that the people building these systems are dishonest or careless with estimates. It's that each solved problem exposes the next one, and the next one wasn't visible from where they were standing. Software compresses that loop into weeks; a physical system runs it over years, because every iteration has to be built, powered, and tested against reality rather than compiled. The timeline slides not because progress stalls, but because progress keeps discovering what it didn't yet know it had to do.

Underwriting a moving target

An investor has two ways to respond to a timeline that won't hold still, and one of them is a trap. The trap is to take the estimate at face value, build it into a model, and treat every slip as a deviation from plan — which turns each predictable delay into a small crisis and pushes the investor toward exactly the wrong instinct, stepping back right as the hard problems start to yield. The alternative is to underwrite the slippage itself: assume the five years is really eight or ten, fund against engineering milestones instead of calendar dates, and size the position so that time is a cost the investment can absorb rather than a threat to it.

The five-year estimate is not a lie and not a plan. It's the distance to the next problem the team can currently see.

This is why the steadiest hard-technology investors anchor capital to physical events rather than quarters. Not the deck's revenue curve, but the concrete result: the engine that has to fire for a full mission duration, the model that has to hold on the cases that decide safety, the cell that has to survive a thousand cycles without degrading. Money released against those events tracks the only clock that governs the company — the one set by physics, which does not care what the fund's term is. A milestone met retires real risk; a quarter passed retires none.

The firms that do well in hard technology are rarely the ones with the most accurate timeline, because an accurate timeline was never available to buy. They are the ones that priced the inaccuracy in from the start, and were therefore still holding when the technology that was always five years away finally arrived — later than anyone promised, and roughly when the people doing the work had always quietly suspected.