[loginalready]

You can reduce this argument all the way to writing your own OS and building your own hardware. And no, that is not ad absurdum, there are valid real world business cases for both.

It all depends an a simple cost/benefit calculation. There's no absolute right or wrong choice.

Also, not having the knowledge to do something is a way bigger vulnerability than outsourcing the execution.

My preferred approach is to figure out how to do it ourselves, immediately followed by finding a 3rd party to outsource it to once we know exactly what we need so we can focus on other things. And of course always make sure we are able to take it back.