Keep in mind current cell phone coverage is spotty, with many unserved or underserved areas, and towns where Verizon is clearly better than AT&T, or vice versa.
Running that kind of cell tower network might be cheaper, but running a cell phone network that spans every inch of the globe would be an engineering and logistics feat that would almost be impossible to profit off of.
Even maintenance-wise, servicing a cell tower requires keeping a bunch of contractors around to drive to drive, fly, or even boat to remote areas, and climb a hundred feet once they get there. With starlink, they can just piggyback replacement satellites on customer payloads.
Also, there are around 4 million cell towers in the world to achieve spotty coverage, whereas starlink is targeting 40k satellites for global coverage. I'd have to believe that launching 40,000 similar payloads via similar methods has huge vertical integration advantages over the extremely heterogeneous approaches we use to put up those millions of towers.
So the ideal situation would be Starlink for remote areas, and in medium to higher density areas put up towers that talk to Starlink, then can speak to local receivers using terrestrial radio technologies.