Yes, if one had the budget to do all those things, from scratch, better than an existing component manufacturer.

Not many companies have NASA levels of "throw money at it until it works, and every part has been signed off on five times."

Absent that, I'm having trouble seeing how custom > COTS.

In all probability, anything in-house would have been worse and added new failure modes.

Better to buy, analyze, and adapt as needed.

And if it turns out you don't need to adapt, because failure modes aren't safety-critical or components are viable in the environment, then spend your time on something more useful.