The other approach is to design the vehicles for inherent modularity.
That is kind of where the future is going. It was (and may still be) a key component of the FCS MGVs (manned ground vehicles). I think given the threats, future vehicles will probably be considered that way. However, there are some issues with it - anything modular implies it can be adjusted, or tailored - and provide flexibility - and this is good, but while a multi-tool, or even a crescent wrench is good - it may not be optimal - meaning a socket wrench might be the optimal tool.

This could get to some interesting (and tough) choices - you maybe can improve the "modular" choice to make it better at some or all of the jobs you might give it, but the cost to do so might be disproportional in the short run to just buying multiple vehicles - where you get a return on your investment is not having to sustain a larger inventory vs. the "one". This is still a tough call, because we've often dealt with over-engineered "things" that were so designed to everything equally well, that they wound up not anyone thing (including the original impetus for the thing) very well at all.

In some cases (armor being an excellent example) that might require substantial support, but that would be available from both organic units and contractors.
I think this part may be unavoidable. I'm not exactly set on how I feel about this one. On the one hand the contractors I've seen in steady state operations have flat out provided incredible support (to our unit), on the other the risk of relying on contractors too much makes my gut hurt. They are fine folks, and certainly pulling their weight, and I believe will be a feature on the future battlefield - but we always need to acknowledge the risk of relying on them (or not)
Best, Rob