The news is all over the web (Ars Technica, MacWorldCNet): AT&T, which had previously had a policy which allowed VOIP apps on its flagship smartphone, Apple's iPhone, to make calls only over WiFi rather than over its 3G network, has relented. This allows fring, Vonage, Skype, and other apps to connect to VOIP services to terminate calls using AT&T's mobile data service, depriving AT&T of per-minute charges for voice calls, and intensifying the shift on AT&T's mobile network from carrying circuit-switched voice channels to being a "dumb pipe" for data.
There are a lot of comments about how great this is, how hypocritical it is of Apple to still not approve the Google Voice app for the App Store, but very little analysis of why AT&T might have done this. There are a number of possible reasons, and the real motivations are probably some combination of these.
In no particular order:
Net Neutrality. I recently commented about how AT&T supported Net Neutrality, except where it would impact them the most: on mobile networks. This may be a recognition that net neutrality will eventually affect how they run their mobile network, but frankly, I think this is less likely. I think they'll fight it tooth and nail until the bitter end.
Public Opinion. iPhone users consume more mobile data than other smartphone owners, so it's only reasonable to assume that there is a reasonably high percentage of bleeding edge users who want to use all the apps they have on their desktop-- including VOIP apps. AT&T might just be caving to demands that it's inconsistent to have these apps on the iPhone (and on other smartphones) and advertise their supposedly blazing-fast Internet speeds, and then disallow the apps they want to use from using that network. This I also feel is a not particularly big factor in AT&T's decision-making process, but I'm sure positive press as a result of the announcement probably figured in a bit.
Good Business. For a long, long time, the big money has been in circuit-switched voice channels, billed by time intervals. Operators knew exactly how much it cost to terminate calls to other networks, and how much it cost to operate those networks. They could set exact prices on a per-direction basis, offset loss leaders in some directions with margin-makers in other directions.
The data business isn't like that at all. Customers don't pay for their data based on direction, and a lot of the time, they don't even pay for certain levels of traffic. They're getting best effort speed and all the traffic they can eat for a low, flat monthly fee. Sure, in the mobile world, you're more likely to have a high fee, a low "bandwidth cap" (even on a service fraudulently labeled 'unlimited') and lower speed and less availability than on DSL or cable. But the end result is largely the same: compared to voice, the operator is getting a fixed income and is forced to cope with unpredictable costs. Some users will be very profitable. Others will not. This is not necessarily less profitable. But it is more unpredictable.
What AT&T may be waking to is the idea that as long as they provide enough data backhaul to their base stations, delivering IP for a flat fee may actually be less costly and more profitable. The data plans for the iPhone are typical in that they are fairly expensive. But if people use Vonage for calling on that network, then they aren't using AT&T's switches. That means no per-minute revenue, but it also means no call termination costs. Since most calling plans have a hefty monthly fee and a large number of included minutes (hundreds or thousands) those unused minutes actually take load off of AT&T's network at no cost. Eventually, dropping voice traffic could free up network capacity for data and alleviate complaints about low data speeds. After all, of all the multimedia services iPhone users could be consuming, VOIP consumes less bandwidth than streaming video or even streaming music.