[1] I'm not too familiar with the aerospace world, but for nuclear that would include shooting gamma rays at the valve in order to see if there's any microscopic crack in it, like you'd do with X rays for humans, but X rays don't go through metal so you need more penetrating radiations.
I worked on microscopes sold to wet labs. When you have limited production runs, the fixed costs, that most of us ignore in saas, become very large on a per-piece basis. Even on 6 figure instruments, to take one that I spent 18 months on, approximately $55k per instrument sold was recovering R&D costs ($20k software, hardware the rest).
How many of them fail to pass the stringent quality tests? The price includes all of the defective valves that failed QA/QC testing.
You still need all of the engineering and tooling, but the costs are spread across a small number of units vs a larger volume product.
What about the cost of that gamma ray machine? Those don't sound cheap. And all the other likely very precisely engineered testing apparatus. Overhead is also a cost center.