Specifically, there are a couple of things I'd point out:
1. They have to put at least 1 satellite into service. That's a huge cost for something that you have no intention of making an important part of your service.
2. They're banking on the FCC changing rules surrounding their "satellite" spectrum. Current rules say that it can be used to create a terrestrial network to back up the satellite service, but every device they sell has to be able to communicate with the satellite. That would make it prohibatively expensive to provision customers with devices against companies who don't have that restruction.
The $7B contract to build and maintain the network also seems a bit light. Carriers like Verizon and AT&T spend around $7B annually to improve and maintain their networks and they aren't starting from scratch. $7B over 8 years comes out to under $900M per year.
Granted, 4G should be cheaper to deploy since the equipment takes up less space and there's no need for a lot of expensive and large things to support legacy phone stuff. Still, Clearwire (who is rolling out 4G WiMAX) had capital expenditure of $690M in the first quarter of this year alone. Although that probably includes more than just network expenditure, Clearwire has yet to cover most people.
If they're able to pull it off, it'll be great! If a company can reduce the cost of rolling out a network, it usually means lower prices for consumers (assuming adequate competition). So, hopefully we'll see a great nationwide LTE network from LightSquared in a few years!