There's a lot of push in the defense world to get to so-called "Digital Twin", where cabling, conduit, maintainability and other late-stage product work is done purely in CAD and in simulation. I'm very open to the idea . . however . . when I get a pile of geometry with no fastener material data, no soft bodies of any kind (tape, string, wire, cloth, rope, etc), broke fillets . . and then someone tells me "there you go, you can figure out the maintenance breakdown now. DIGITAL TWIN!" . . I'm not filled with confidence.
Someone, somewhere, needs to spec out the minimum required for each stage of this kind of work . . they might find that doing things like wire channels in new product isn't fantastically easier than doing it in a prototype. You can do it in CAD< but you still have to get the design right. Tool clearance doesn't magically appear in the right place by screaming DIGITAL TWIN.
This is all aerospace. In another industry, with less demanding performance requirements, the easy solution to all problems is always "increase the tolerances". Then you can jam anything in there.
Another problem: logistics systems want to know the sparing level for what's being supported, not what's being designed. This, incredibly, isn't addressed so far as I can see; adopters are just sort of pretending that bleeding edge is the same as the as-built, supported configuration. When I bring it up, the general response is that LSA can be batch-create from the design, totally ignoring the fact that a lot of changes come the other way, from field modifications. The solution for this - for NGAD and Raider - is to just not do logistics as it's been practiced for the last forty years. Spoiler: this is the correct answer. 30 year sustainment programs are hilarious fictions that do nothing but generate high margin work for primes.