Depends on a bunch of variables of course but ballpark a few million monthly users would probably be good enough.

The falcon heavy launches are going to be ballpark somewhere around 90 million. So excluding the satellite construction that's what it costs to launch around 70-100 satellites. They need at least about 30-35 launches. So ballpark that means a it is going to take probably 3-4 billion. Of course that is excluding the production cost for the satellites and other infrastructure. Lets call it 10 billion altogether, give or take a few billion.

Not a crazy amount to spend on infrastructure in the telco world. German operators paid tens of billions just for spectrum licenses. Considering you'd be able to compete world wide with this, 10 billion is a bargain.

It's still a lot of money of course but with a few million customers paying in the order of 10-40$ per month it would earn it self back quite rapidly. Even at the low end of that price range. Arguably they could charge more.

We also need to factor in the cost of replacing all the satellites every 5 years or so, as they only last this long in the super low orbit that they will be in. So it's ~10 billion every 5 years, not just once up front.

So 10.000.000.000 / 5 / 12 = 166.666.666 bucks per month

Say an average subscriber pays 20 bucks.

166.669.666 / 20 = 8.333.333

You'd need about eight million subscribers paying an average of 20 bucks a month to keep the business sustainable.

Compared to the $200 Comcast bill I would be more than willing to pay $50 a month for this, plus another $100 for video content. The big question I have is how many subscribers can they support per square mile in higher population density areas?