I'm sure there are various ways the work spaceX has done on reuse could apply, perhaps in reliability engineering the most. I recall a statement along the line

I'm sure there are various ways the work spaceX has done on reuse could apply, perhaps in reliability engineering the most. I recall a statement along the lines of 'we could reuse it, but if X, Y, or Z seriously fails in-situ, we'd spend at least as much as a new machine in trying to extricate and repair it. So we don't.' No doubt bits of the machines are modular and can be salvaged and reused, but especially in urban tunneling, the initial bore access can be very tight, and machines are often assembled in-situ and could never be removed in one piece. That, and the sheer punishment they receive during operation makes an easy-to-dismantle design either more expensive or less performant. Finally, my understanding is that the bulk of the cost (especially in overruns) has little to do with the TBM itself, but rather the delays incurred when the territory doesn't match the map. So the solution would be to make a boring system capable of rapid reconfiguration to handle as many expected and unexpected regions as routinely as possible, rather than going for flat-out speed or ultimate machine reuse.