SpaceX Starlink Constellation, page-6

Narrow bands real advantage is cost. SpaceX are using 800 microsats with ultimately 4425 and then another constellation of 7518 orbiting at a different altitude. So 11,943 satellites in total. These are more capable, but much larger (400Kg each) and presumably much more expensive.
SAS is using much smaller (<10Kg each) and much cheaper nanosats. The constellation will consist of 200 for equatorial coverage and will ultimately expand to around 1,000 for global coverage.
I'm sorry I don't know the cost of the SpaceX sats but it is fair to assume they are significantly more expansive to build and to launch (the $22,000 per Kg figure has been thrown about a lot lately).

SpaceX = Broadband 12,000 Satellites and 4.8Million Kg to launch
SAS = Narrowband 1,000 Satellites 10,000 Kg to launch

IMO SpaceX is selling a car, SAS is selling a bicycle. A car is usually better, but when your current option is walking and you're strapped for cash, a bike starts looking pretty good. Plus for some niche applications where a bike (narrow band) will do the job just as well (ie. SCADA, M2M, IOT), why would you pay for a car?
I do see some potential risk from with this SpaceX plan. For example SpaceX could decide to make their money in the developed world and give services to the developing world at a cheaper subsidized rate. Or they could sell a limited broadband packages at a cheaper rate that could compete in the same space as SAS. I would also expect as certain areas and technology develop people will migrate to broadband networks. My personal opinion is SpaceX aren't planning on targeting the same space SAS, hence the differences in the technology and costs. I also think all going well SAS should beat them to market by some ways.

On a side note:
Does anyone know the answers to these questions?
Does SpaceX build their own satellites?
How fast can they produce and launch their first 800 satellites?
How many ground stations will they require?
I'm also curious about their cost per satellite.