Each Tintin satellite (demo for Starlink constellation) is 400kg. Assuming each satellite+ fuel (and that the weight doesn't increase for other reasons) is 1kg, 22 could fly in a single F9 per inclination (there's probably not even that much space in the fairing, but we'll ignore that). The total # of (expendable) F9s launches needed is 546; assuming the first comes online in 2020, and the constellation is finished by 2025 (the official date is the 'mid 2020s'), you need 110 F9s per year. The total cost for development and buildout is supposed to be $10 billion.
SpaceX has one facility for Polar launches right now, SLC-4E (Brownsville maybe, but a polar launch from there would GRAZE the Mexican coast- I have a feeling they won't be happy about that...) So each complex would launch 54-55 F9s a year.
Let's be VERY optimistic. Every satellite will be launched on an expendable Falcon Heavy, each satellite will be EXACTLY 400 kg (unlikely, you need payload mass margin, and likely fuel from having a lower satellite orbit), we'll ignore whether each satellite can actually fit in the payload fairing, we'll ignore launch inclination entirely (launching at the wrong inclination means hell for a satellite wanting to change orbit; meaning 160 satellites per launch (almost certainly impossible due to not enough fairing volume). That means 75 launches of the FH, minimum (almost certainly unrealistic.) We'll also assume it launches first Jan. 1, 2019, and put the finished launch date at Dec 30, 2027. This means 8-9 FH launches per year. Sounds easy, right? But that's ~23% higher than the current # of launches SpaceX does from SLC-4E, every year. It would literally have to double its polar launches, at the absolute bare minimum. A mimimum which is almost certainly physically impossible.
I don't think building the satellites will the the issue, Elon...
No one at SpaceX did these back-of-the envelope calculations, and asked Elon- maybe it was a little ambitious?