I have no credentials, so correct me. Do the high 5G frequencies (Frequency Range 2) lend themselves to tiny cells with no contention? I'm talking a radio for every ~50/100 subscribers or so, placed on light posts and telephone poles.
OP could've admittedly done a better job with this, but let me give it a try.
The problem comes down to spectrum, how much power you can push across that spectrum, and how you share that spectrum with other devices. The more bandwidth you have available, the more power you can send across it, and the fewer devices you have sharing it, the higher your information rate between a sender and receiver.
5G, in theory, brings a lot of spectrum to the table, so they can talk about, say 1 Gb/s or 10 Gb/s connections. The problem is, you've got to find a way to share that among users. Sharing spectrum is never a 1/n process because of inefficiencies, so you end up with a bit under 1/n speed for n users.
Meanwhile, cables might have less spectrum, but you can push comparatively more power through them, but more crucially, everyone gets their own connection and there's much less noise. So, more aggregate bandwidth across cables than over the air.
This is all very handwavy, so if someone would like to put some numbers to this and/or fix any mistakes I made, please have at it.
Technically, everyone on the same cable node shares the same spectrum (typically targeting 200-500 subscribers.) Which is effectively 1GHz at the moment, a decent chunk of which is dedicated to TV.
But certainly there's less noise, neighboring nodes effectively don't interfere with each other, and nodes can be made arbitrarily dense.
