No it would be "preserve institutional knowledge about building OUTDATED rockets"
If you are a tech company are you going to run your own outdated servers at 10x the cost because you would like to preserve institutional knowledge about running outdated hardware ?
Your argument would be correct if the cost difference was not that huge, the fact that it is so huge indicated that the tech they are using is very very outdated. No point in preserving anything about it.
Preserving knowledge is easy and cheap.
Preserving capability (that is to say, people and companies with knowledge and ability to use it) is not.
To use the metaphor, a huge part of the military procurement budget is dedicated to continue building, deploying, and operating on-prem servers.
Not because it's more efficient, but because at an international level, having a cloud provider (or in this case Russia or China) tell you "No" when you ask to renew or expand, is unacceptable.
Unlike tech, machine shops filled with trained people can't be provisioned with a click...
> If you are a tech company are you going to run your own outdated servers at 10x the cost because you would like to preserve institutional knowledge about running outdated hardware ?
As a startup CTO, I will think twice about using the freshest framework or programming language, which is sexy and cool - just because I haven't seen enough people use it long enough yet. I will build a hobby project with it and love it, but or a small product that is allowed to fail, but I will not start an important new product with something like that.