Nuclear power is not new. Small reactors have run submarines since the 1950s. What is new is a company that builds one in a factory, puts it on a truck, and delivers it like a generator.
The science was settled decades ago. The manufacturing was not. We see the same pattern across everything we invest in. The hard part is rarely the idea, it's producing the thing, at volume, on a schedule, at a price a customer can pay.
The old model
For most of the last fifty years, the United States bought its most important hardware on cost-plus contracts. The government pays a contractor's costs and adds a fee on top. The incentive is easy to see: the longer and more expensive the program, the larger the fee. That system built remarkable machines. It did not build them quickly, and it did not build many of them.
A newer group of companies works the other way around. They pay for their own research, build a finished product, and then sell it. They carry the risk themselves, which changes how fast they move.
The talent moved
Ten years ago many of the best engineers in the country were working on advertising and consumer apps. A meaningful number of them now build aircraft, reactors, rockets and batteries. They brought software habits with them: ship, measure, fix, ship again. Applied to hardware, those habits shorten the distance between a design and a production line.
The factory is the proof
A deck can describe a factory. It cannot substitute for one. We put more weight on a production line than on a projection.
Anduril began building aircraft at its Ohio plant in March 2026, ahead of the date it had set for itself. Radiant sent its first reactor from the factory floor to a national laboratory for testing, on the back of a truck. Redwood Materials takes batteries at the end of their life and turns them back into the metals a new battery needs, in Nevada.
We think of Radiant as building the Model T factory for reactors. The comparison holds for the others too. The product matters. The way it is produced matters more.
Why it is a national question
Battery materials, advanced chips and nuclear fuel have something in common. The United States depends on other countries for much of its supply. That was an acceptable trade when the only goal was the lowest price. It is harder to defend when the product is the electricity grid or the defense of the country.
What we look for
So we look for three things. Hardware that exists. Customers who are paying for it, or have contracted to. And a plan to manufacture that is as serious as the plan to engineer.
In hard technology the main risk is usually delay, not physics. We would rather own the company that is late with a real product than the one that is early with a slide.
This note reflects the views of Inflection Ventures as of the date written. It is not investment advice or an offer to sell securities.