But making it up on volume

Martin Peers of the WSJ’s “Heard on the Street” column reports today that data suggests that Cingular AT&T is losing money on iPhone customers:

Users of iPhone download games, video and other Web data at two to four times the rate of other smartphone users, according to comScore. Yet AT&T charges iPhone subscribers the same fee of $30 a month for data that it levies on other smartphone customers. And aside from restricting certain activities, like file sharing, AT&T doesn't limit how much data can be downloaded.

Peers estimates that iPhone 3G users (from July 11 to March 28) are 7.5% of all AT&T subscribers.

While everyone knows that iPhone users love to surf the web, the bad news is that (based on data compiled by Lucent) web browsing (surprise!) takes 16x the bandwidth of email. Web browsing is 32% of the usage but 69% of the bandwidth, while for email it’s 30% and 4% of the bandwidth. (Web browsing is 1.9x data intensive per minute as P2P). Or, as they say, AT&T is losing money on every unit, but making it up on volume.

While the supply of iPhone applications is a classic software positive network effect, use of the network is a negative externality. As my friend Rudi Bekkers wrote in his book on 3G networks:

Negative network externalities, for instance, when a telephone or computer network becomes congested or overloaded and the value of that network for an individual users decreases.

Peers argues that for AT&T and Verizon Wireless, the only solution is to abandon “all you can eat” data plans for cellphones, the way they have for laptops. The only problem is that there are cracks in the oligopoly:

With competition, the temptation to discount will be hard to avoid.

That competition will be from Sprint (trying to stem its sliding market share) and T-Mobile (still trying to gain share).

American consumers are used to “all you can eat.” They long enjoyed it for wireline voice communications, and forced it upon AOL for dialup ISPs. The wireless voice business is moving in this direction, with $50 unlimited service plans niche carriers like Metro PCS and Leap, as well as Nextel’s seven-year-old prepaid division, Boost Mobile.

All-you-can-eat (possibly with some sort of reasonable cap) is the only way that American consumers will adopt the mobile Internet. The iPhone users have shown there’s a pent up demand for mobile web browsing, but if it means the risk of $100 data bills, they won’t do it: instead, they’ll wait until they get back their wireline Internet.

So if the Big Two aggressively price data services, they may get too many users (shades of AOL’s excess demand for unlimited dialup services). If they charge too much, they’ll lose customers to smaller carriers, or to WiFi hotspots, or people will stay with their wired Internet.

One possibility (as suggested by GigaOM) is congestion pricing: give away megabytes of download bandwidth only when it’s unused, and charge a premium when everyone wants it. In the extreme, it would be like the cellphone (voice) pricing strategy of the 1990s: free night and weekend minutes, but expensive minutes on weekdays and at rush hour.

In the short term, the numbers don’t work for 3G unlimited data plans. In the long run, plans to build 4G networks assume high levels of usage, and proponents claim that LTE networks are 2x-4x more efficient than their WCMDA counterparts. I’m not sure that even that is cheap enough bandwidth to support all-you-can-eat.

I think it’s long past time for American carriers (as do European like Orange and T-Mobile) to embrace WiFi as a complementary service to their mobile networks. Much of the Internet browsing occurs in coffee shops and similar locations, so now that people are starting to embrace the mobile Internet, there’s no reason why phones can’t be programmed to prefer the high-capacity (and easily expanded) hotspot over the scarce 3G bandwidth.

Here, AT&T and T-Mobile are well positioned with their existing hotspot networks, with AT&T growing its network last year when it purchased Wayport. Meanwhile, Sprint has been selling hotspots, and presumably it hopes that its WiMax network will obviate need for 802.11 hotspots. To catch up, Verizon would have to buy Boingo; rumor has it that it will soon announce a partnering agreement.