A slightly disjointed USA Today article discusses the FCC's considering how to auction spectrum for in-flight voice and data for ground stations: Most interested parties are arguing for competitive auctions, even if that makes it harder to coordinate where ground stations would be based. Aircell already has a working system, and it's apparent that they need more spectrum--although that's not stated in this article. Aircell already sells telecom services--voice, data, fax, telemetry--to airlines and private planes.
Ground-to-air stations would allow a potentially cheaper deployment per plane: $80,000 versus $500,000 for Connexion by Boeing; it would also weigh less. The article doesn't mention it, but Boeing is limited to 150 transponder pairs at the moment, so that there are 150 "cells" worldwide, each of which has a fixed maximum bandwidth pool. As usage increases, Boeing must increase their budget for transponder license ($1 million per transponder per year or $300M) or travelers might be contending for bandwidth as planes travel in the same large cells.
Tenzing's system allows for much more discrete bandwidth per plane, although in lower quantities, using a focused system that doesn't rely on cells. But you can't beat ground-to-air stations in which the distance will make high rates per plane possible at this lower cost. (Tenzing estimates about $100,000 to $200,000 for their over-water satellite system for any of the several thousand planes that already use Inmarsat satellite gear for international flying.)
All this to say that William Raspberry's worst nightmare is about to come true. Read his column, and consider: are you the talker or the middle-seat victim?