Jim Thompson explains the T-1 market to me and others: it's cheap when you're a carrier: I confess a career-long misunderstanding of the difference in cost for a T-1 line (1.544 Mbps) sold to a business, sold to an ISP, and sold to a telephone carrier. Jim Thompson was the CTO of Wayport a few years ago and more recently at Vivato.
In his post, he notes that even when he was at Wayport, his average T-1 cost was $250 per month, and that T-Mobile could be paying as little as $90 per month in many locations in which they have co-located equipment to carry their cell traffic.
Combine that knowledge with Carlo Longino's analysis of our post yesterday about T-Mobile's revenue averaging to $400 per location per month, and you start to see that they could, in fact, be running far cash-flow positive. (Carlo wonders if the infrastructure is all T-1 still; that's what T-Mobile confirmed for me all last year for Kinko's, Borders, and all Starbucks past and future.)
Carlo notes, by the way, that T-Mobile said 67 percent of users are subscribers. This doesn't give us apples to apples on sessions, but if you ascribe $1.2 million to 48,000 subscribers at an average rate of $25/month ($20 for T-Mobile cell users, $30 for one-year commitment regular users, $40 for month-to-month) that would leave you about 25,000 hourly/day sessions a month. Is it really possible T-Mobile has 48,000 subscribers?
Even if T-Mobile is paying prices all across the board, Jim's information leads me to believe that their recurring costs might be substantially lower than anyone has previously estimated. Carlo's factoid contributes the knowledge, if it's all connected correctly, that the cell carrier might be signing up many more regular users of Wi-Fi service than previously imagined.
Couple this with their iPass deal, which resells access at the same $9.95 per day rate to iPass's combined hundreds of thousands of business users, and perhaps this isn't a money-losing venture after all.
It may be hard to get reliable numbers to confirm or deny each of the aspects of this chain of logic that Carlo, Jim, and I have assembled. Nonetheless, with more of the numbers exposed, it may be simpler to confirm any one of them.