A MHz-POP is just bandwidth times population covered by the Geographic Service Area (i.e., where the company is licensed to operate). For example, in the US, they would have 11.5 MHz x 330 M = 3.79 Billion MHz-POP.

Thank you! Doesn't that seem like a weird metric? I would think MHz/POP would make more sense. I guess the idea is to assume unlimited and independent channels to everyone as a first order?

MHz-POP makes the most sense in cell networks, where an operator (AT&T, T-Mobile,...) wants to acquire a spectrum license in a particular region of the country. Evaluating the MHz-POP makes sense as the price they are willing to pay varies a lot depending on the population density in that region area. In general, cell networks can reuse spectrum more easily (deploy more towers, add more sectors), and they design their network deployment to hit whatever MHz/customer they are targeting (which mostly depends on the technology 3G/4G/5G).

In sat-networks, well, MHz-POP doesn't matter that much, because, generally, every operator is licensed to operate in the whole country. As you mentioned, what really matters is (a) the bandwidth of their license allocation (e.g., Globalstar is 11.5 MHz), and (b) how efficiently can they reuse spectrum:

* how many beams can they land (# satellite x # beams / satellite)?

* how much freedom do they have to chunk bandwidth and allocate it to individual beams based on demand?

* what type of satellite are they using, bent-pipe or regenerative payload?

* how big are these beams?

* can they allocate resources dynamically or is everything fixed?

* how much power does the satellite have? how big are the terminal antennas? what kind of link-budget can they close?

In the end, the MHz/customer they can achieve depends on the answer to all these questions.

I think multiplying by population serves as a way to normalize for link speed. Ten people who use a lot of spectrum are probably bigger customers than ten people who use a tiny sliver of spectrum, and thus constitute a bigger user base.