Yup to the above, and no, I'm not aware of any (unless it's what Musk is doing). I was just thinking out loud and speculating what could be possible given the technology available today, and the need to support wide bandwidths from a handset that has a relatively poor antenna and low output power, relatively inexpensive and power constrained compact satellites, and significant path loss. Within timing, ephemeris/positional accuracy, and phase resolution, constraints, I think it's definitely possible. The size of and positional accuracy of the spot, and thus gain, is a function of the number of emitters and limitations/uncertainties of the above.
In the end it boils down to channel bandwidth (ie how much is needed - a lot to meet today's expectations), path loss, low NF receiver channels, available power, and spot size.
It would be a fun problem to work on.

I wouldn't be shocked if Apple were exploring something similar.