And you aren't even mentioning the case of degree programs that are in some essential sense about learning to use equipment costing in the tens to hundreds of thousands of dollars, and in some cases whose availability is strictly controlled by the US government.

I got my undergrad in Nuclear Engineering, and I learned to use Gamma Ray Spectrometers costing easily 30,000 dollars and did Plasma Physics experiments in a Vacuum Chamber setup of similar cost, all the while supervised by people who knew how to safely perform experiments with radioactive material and comply with relevant government regulations. And I was mainly about Computational Neutron Transport - I got what was considered a bare-minimum grounding in Experimental Nuclear Science. This is to say nothing of programs with research reactors students can use for experiments (whose ranks my alma mater sadly left just as I arrived).

This was basically true of all the non-CS engineering disciplines. Perhaps one could cobble together an equivalent to a Mechanical Engineering or Civil Engineering degree from my alma mater with CnC Routers at Hackerspaces and the like. But I really can't imagine that working for Nuclear, Aerospace, Biomedical, the Semiconductor side of EE...

Or rather, I can, but it would require an enormous investment into high-fidelity simulation of the relevant equipment. And that is a dangerous road. At the moment, we verify Engineers' ability to use equipment in a very robust way - having them actually use the equipment to perform engineering tasks and measurements. If you rely instead on training simulators you need to be extremely vigilant about their fidelity (and you will almost certainly fall back on minimum duration of instructor-monitored usage of real equipment, like the nuclear and aerospace industries require).