Karl picked up my thoughts on AT&T spectrum slightly differently than I meant to imply. I wrote
"Does AT&T need the T-Mobile spectrum? (no - 70-90% of the AT&T
spectrum capacity is currently unused.)"
not 70-90% of the spectrum. I including in that figure a great deal of spectrum that is currently "used" with older technologies but carries far less than it would with current technologies already being deployed by AT&T, Verizon and everyone else.
I'm working from a comparison of what the spectrum could carry with current technology (LTE, HSUPA+) compared with the capacity in use. That's 1.5-2.5 megabits/megahertz, depending on whether you're measuring average versus edge of cellsite, fixed antennae, etc.
That's 120-300 megabits in most markets. Let's call it 200 megabits average.
Currently, the heavy majority of AT&T cellsites are still served by T-1's carrying a total of less than 12 megabits, often far less. It's typically used for "up to 7.2 megabit" data and lots of voice. Some percentage - surprising high, but not 70-90%, is currently unused.(?20-60% on average as a wild guess, but that's unprovable without internal AT&T data.)
Much of the spectrum is currently used for voice, using older technologies that are far less efficient than today's. Glen Campbell of Merrill Lynch estimates that by refarming that spectrum and using it efficiently (if only for voice) you double the carrying capacity. Similar is true for all the spectrum being used for 2G and even 3G data. They only use 10-?50% of the capacity of the spectrum using today's technologies.
Carriers around the world have begun this "re-farming" for more efficiency, including UK and Canada. Everyone has it in their plans because it's more efficient and hence cheaper. Sprint intends to do that with the Nextel spectrum and AT&T has discussed similar. It takes time, because you have to change out all the handsets, but using existing spectrum more efficiently saves so much money the carriers are doing it almost universally.
Much spectrum lies purely fallow, about enough to carry us without upping capex about 5 years (FCC figure, badly calculated) or 10+ years (Ivan Seidenberg of Verizon and most technical people as opposed to lobbyists.) Most of the rest is used by older 2G and 3G tech (both voice and data) and has 2-4x the capacity with today's technology.
Hence, T (and almost every one else) is using only 10-30% of the capacity of their spectrum. They know this and are rapidly upgrading backhaul (2010-2011 primary problem) and radios. They are discussing plans to switch users from 2G voice - still what's in 3G and 4G handsets - to 4G voice over IP/LTE over the next few years.
AT&T's announced 2011 backhaul upgrades - from 20% GigE fiber or 100 meg microwave to 70% - will yield 500% more capacity for data this year alone. »fastnetnews.com/a-wireless-cloud ··· -in-2011
They are using it to go from 7.2 meg to 20 and 40 meg HSUPA+ and LTE.
A 300% improvement in bandwidth implies 75% of capacity was unused.