Also, Starlink doesn't have to be too costly for them. The constellation's operating costs could potentially be fully paid by North American and European users. The satellites would still incidentally fly over the southern hemisphere, and any revenue at all from Africa/South America/Asia would mostly just add to Starlinks bottom line profit. The marginal cost of additional users should be minimal once general revenue targets are met.
I'd like to hear whether the throughput will increase with additional satellites, or whether it requires re-designs. I wouldn't expect the current satellites to be saturated with users by the currently "limited" number of beta test users.
Cost of end-user terminals also will likely have to come down significantly before the worlds poorest would have any access. Although sharing the cost of that among a large group would help, it would also help a lot if the speeds were actually closer to gigabit for sharing the connection.