Also of note, the launch market has changed for the better considerably since then. Despite having the same constellation size for Iridium Next and slightly larger satellites (by about 15%) they'll be able to get them deployed with only 10 launches at a total cost of $490 million.
Of course, now the market has largely caught up with their vision so they're doing a lot better financially.
The fact that you can do TCP/IP data over the first generation Iridium network is actually kind of a minor miracle, the data rates are so low. Voice calls have to fit in 2400 bps with a special codec. Data sessions used a lossless compression system not very dissimilar from v42bis to squeeze up to 9600 baud out of a 2400 bps connection... Well, at least if all you're transferring is a plain text file, if it's content that's already compressed you literally see 0.2KB/second.
They're going after the revenue that Inmarsat enjoys from BGAN services, Thuraya's revenue stream from BGAN-like L/S-band services, etc. Maritime and aircraft of course. And military... and military and civilian UAVs... all sorts of stuff. M2M stuff that has to work literally anywhere.
This is why Iridium has kicked the ass of Globalstar, which was a bent pipe repeater satellite arrangement and relied on dozens of earth stations worldwide, also rendering it unable to provide service in polar regions and in the middle of oceans.
The only other telecom satellites that form backbone/trunk links in space are geostationary and military. The network architecture was WAY ahead of its time considering the design was finalized in 1996-1998 or thereabouts.