Fast Net News, DSL Prime, DOCSIS Report

Imagine if five new networks with the capacity of Verizon's were built across the United States. Canadian wireless prices are dropping 10-15% because of new entrants. Xavier Niel's fourth French mobile net will bring down prices 17%, Merrill Lynch estimates.

    The heart of the plan is freeing more spectrum to bring in competition, with 300 MHz expected by 2015 and 500 MHz by 2020. Letting Phil Falcone's Harbinger use satellite spectrum for terrestrial as well, an AWS-3 auction, and other moves already under way will bring in 100 MHz, more than enough for 3 new Verizon-sized networks. That can be huge, if new entrants can succeed in taking major market share from AT&T and Verizon. More spectrum is not guaranteed to work – V & T are at 60+% and pulling away – but politics excluded any of the more effective moves. Although The FCC has apparently lost on using the 10 MHz 'D' block and some of the plans like relocating TV stations have issues, even half of what the plan asked could be huge.

    Verizon's LTE network, one of the best wireless networks in the world, is running on 20 MHz of spectrum. CTO Dick Lynch is confident he will soon bring 5-12 megabits down to 92% of the U.S.. The broadband plan determined that the coverage could be raised to 97-98% at manageable expense. Engineers tell me both goals are realistic.

     Wireless is shared and the total capacity limited. As a practical matter, that means the network breaks if many people want to watch lots of quality TV. But the capacity is pretty darn good. Back of the envelope, a 2015 LTE network probably would allow most people to do all the surfing they wanted and watch 5-15 hours/month of video.

A few could watch more. These are guidelines because number of towers, frequency, network load and usage patterns will vary widely. Adding an antenna outside each home can nearly double capacity.

    That doesn't call for an abusive cap like AT&T's 2 gigabytes. Very few users any month come close to the cap so a cap between 25 and 75 gigabytes/month would be probably be right for the primary service. More efficient measures might include a 100 gigabyte cap of which only 20 gig could be used in peak hours and various complicated time of day or congestion pricing systems. Wireless therefore is an interesting partial substitute in a few years if competition keeps down prices.

There's enough spectrum in sight to double the number of state of the art U.S. mobile networks. To facilitate that, the FCC should immediately change the auction rules that are a huge obstacle to competition and almost certainly reduce the amount collected. In most of the world, part of the spectrum fee is paid over time. In the U.S., it all has to be paid up front, three or four years before significant revenue (the network has to be built) and even more years before breakeven. With U.S. licenses going for $5B and more, that's a hell of a lot of cash to tie up without even a hint of return. That's a huge obstacle for most bidders, who would gladly pay more if the spectrum payments were more closely tied to when they have income. A new network competing with powerful V & T has to be considered speculative, which means their cost of borrowing the spectrum fee money for the three years of construction is very high.

In addition, even if the bid made sense, there are very few able to front $5B except the incumbents. So the current auction rules are an incumbent protection plan, especially valuable to the larger incumbents, V, T, and Deutsche Telekom. Verizon and AT&T made that clear by bidding up and if necessary buying spectrum in the last few auctions that they didn't really need. They and Sprint have enormous amounts of spectrum not in use, so this was preemptive to keep new companies out.

Or maybe, as Genachowski fears, V & T scale will dominate no matter what the FCC does.