Sure the iPhone has problems, but John Stankey of AT&T thinks restoring a $2B capex cut will fix them. It may take a little more money than that, but Glen Campbell of Merrill Lynch has confirmed he’s on track. In a 50 page report that’s one of the best I’ve read in years, Merrill destroyed the common belief that wireless has a significant spectrum shortage. He estimated the likely traffic demands (including iPhone) through 2012 (and possibly longer) and the realistic capabilities of the technology. AT&T and O2 obviously have to catch up some investment, but Campbell concludes “For most wireless carriers, we expect that capex increases will be temporary and/or modest ... network equipment cost declines will continue.” He reviewed AT&T’s plans to fix things and a myriad of technology moves already beginning that improve voice and data performance.

Off the record, two of the best engineers in the world have confirmed to me that’s perfectly plausible. It corresponds to my research, an FCC conclusion, and what John Stankey of AT&T is telling investors.

Campbell calculates it costs less than $3/month for a gigabyte of added capacity, falling. If the average “unlimited” (really 5 gig) customer uses about 1 ½ gig (a likely average), that means a carrier collecting $20-50/month needs to spend about $5 of that for the bandwidth.

It should fit in a capex budget similar to todays and will drop with Moore’s Law.

An important point is that the carriers can essentially control the bandwidth demand by plans and pricing. If they stay with today’s 5 gigabyte cap or even double it, the numbers work. Behind the higher estimates for 2012-2015 is an implicit assumption that carriers will raise double their caps every two years and be at 20-40 gigabytes for the almost all customers. That’s unlikely.

Those caps will limit the use of wireless for folks who want to watch all their TV over the net. It is therefore only a partial substitute for landlines, so I believe most families will maintain both. With a landline in most homes, 30-50% of wireless traffic can be diverted to femtos and WiFi. Vodafone and AT&T are well along making that happen, especially as femto prices drop to $50 and eventually $20. They pay for themselves just with the spectrum savings. Vodafone is already giving them out for free as part of the bundle and 2Wire has promised gateways with femtos inside.

While some of his conclusions are U.S. centric, most of his data applies across the world. Glen is Merrill’s world lead on telecom and he has some conclusions about both the emerging and developed world. He has long been acknowledged as one of the dozen best in the world. This report is the most interesting I’ve seen in a while. It should be available through the standard investor databases but unfortunately is not public.

There are predictable technology shifts like double or triple capacity in 3G/4G, huge IP voice efficiencies that will free up spectrum used for voice today, mass deployment of WiFi phones and/or femtocells.

My guess is we probably can go a decade without a serious problem based on some almost certain FCC moves. Improved spectrum use is the most important part of the broadband plan, as I wrote more than a year ago.