AT&T didn't want to let T-Mobile steal the initiative on HSPA 7.2: I missed the 5 January 2010 release from AT&T that explained the firm had updated its mobile broadband network to HSPA 7.2; T-Mobile said last week its entire footprint is now HSPA 7.2 as well.
AT&T said last year that it would roll out the 7.2 update--which offers a raw data rate of 7.2 Mbps including network overhead--to most sites by the end of 2010, and its entire footprint by 2011. In Europe, HSPA 7.2 networks seem to operate in a broad range: users report regular performance from about 1 to 4 Mbps. It's unclear how carriers will provision, throttle, or shape HSPA 7.2 here.
AT&T's announcement coming early made it seem like it might have leapfrogged T-Mobile, given that AT&T has more metro areas covered with 3G, and has already sold a fair amount of HSPA 7.2 enabled gear, such as the iPhone 3GS. But that's not the case.
AT&T wasn't trying to pull a fast one, even though it was marketing this move as an enhancement. In the 5 January press release, the company makes it crystal clear that it's upgraded the software to allow HSPA 7.2 even though it doesn't have the necessary backhaul improvements in place to make the network actually faster.
This is refreshingly frank. The release said that the software update would improve network quality even before backhaul improvements that are slated for this year and next are in place to provide increased capacity. I'm sure AT&T is correct. More efficient use of the local link among capable devices means less wasted air time and less congestion, which allows better packet shaping and prioritization. I'm sure local network links function far better at 7.2 Mbps, even if the backhaul can't support the uplink side.
Six cities have the promised backhaul improvements underway: Charlotte, Chicago, Dallas, Houston, Los Angeles, and Miami. The release says the updates are site by site, which means you could experience different speeds driving around the same city while the upgrades are in process.
T-Mobile, on the other hand, has been building its fresh HSPA network with the notion that backhaul for HPSA+ (21 Mpbs in its version) will be flooding the grid, and thus has the advantage of not having to deal with legacy installations. It has been able to make all its choices about towers with the notion that each base station may need 20 to 100 Mbps of backhaul. (These are guesses; the company hasn't released that information.)
I continue to find it ironic that wireless networks rely so heavily on wires.