Fast Net News, DSL Prime, DOCSIS Report

FIOS is fastest (around 200 Mbps), then DOCSIS 3.0 (about 50,) then AT&T's U-Verse (1-10, like other DSL and older cable) slowest.

By technology, Verizon would win in 20% of the U.S., while cable takes over 70%. AT&T, Qwest, and the RLECs are out of the game. That's the "secular tendency" for the next decade, although the very weak U.S. competition could easily override the network issues.  AT&T and Verizon have a crucial advantage in their near-control of the wireless network, and $tens of billions more cash available.

Result:

Cable may or may not hold its own against FIOS. DOCSIS 3.0 at 50 meg may be enough, although not including upstream may be a bad mistake. Still unclear.

If cable prices aggressively, they should clobber AT&T, Qwest, Bell Canada and the other telcos. 100 meg down, no cap, costs between $25 and $40 in France and Japan as part of the bundle.

If cable and telcos "wink and nod," and both keep prices high, the U.S. stays behind. Nine out of ten executives would try to reach just such an understanding in a two-player game, splitting the market. As a carrier VP explained to me, "We've gotten so good at signalling we don't need to meet each other to split the market."

That implicit collusion describes Canada since George Cope came to Bell; France and Canada with three wireless players; and probably the U.S. in 2001 when AT&T, SBC, Verizon, Comcast and just about everyone else put through a huge broadband price increase soon after Covad, Rhythms, and Northpoint died.  That was when I started worrying that U.S. competition was inadequate. Except for Ivan and FIOS, every North American telco has tried to "reach an accomodation" with cable; on the other side, only Cablevision and Videotron priced aggressively to beat the telcos.

Here's how it looks from each company's point of view:

Randall Stevenson decided to put AT&T's copper network into "harvest mode" and cut capital spending literally in half in 2002-2003. A "good numbers guy," he looked at the cold equations of the business and decided not to put in the money to protect the business. If AT&T's plain old phone network dies, so be it. As CFO during Pronto, Randall had signed too many checks for too much money to upgrade the old SBC. "Don't spend a penny more than you must," was the implicit order to his new CTO, Chris Rice. Capital spending fell to 30% below depreciation, less than the minimum to maintain the network."We are a wireless company." He moved the DSLAMs into field cabinets closer to the customer to get a little more speed, just enough for a modest TV offering. A brilliant pr campaign called that "fiber to the node," fooling most of the press and even two FCC commissioners.

Ivan at Verizon decided to stay in wireliine and fight. He put $20B on the line for FIOS, the best large network in the Western world. "We have to get cable out of the house," he told me five years ago.  Ivan does not want to do that by cutting prices, hoping his superior network will give him a long term advantage. So far, the results are fair at best.

Verizon FIOS is the best network in North America, designed for 200 megabit speeds to every home. DOCSIS is designed for 50 meg. U-Verse, Qwest DSL, and cable before 3.0 are best thought as 1 meg up, 5-10 down.

Implications, if technology were the primary factor:

Verizon will have an advantage over cable where they have FIOS, soon to be about 20% of the country. As I write in early 2009, they are finding few customers willing to switch

In wireless, WiMax and LTE are designed for low megabits, 3G for kilobits, but there are numerous variations.

Wireless and DSL overlap in the 500K to 5M space, with wireless winning on mobility. A significant percentage of mobile data users will drop their landline broadband. Common estimates range from 5% to 25%; a handful think it will be even more extreme.

...Verizon has passed 12M homes, with 18M set for 2010 and millions more afterwards. Only about 25% of the U.S. will be covered, unfortunately. It's expensive to build, about $1,000/home.

DOCSIS 3.0 was developed to compete, although most DOCSIS deployments are best thought of as 50+meg. It can go to a shared 1 gig. DOCSIS doesn't need digging, so is closer to $100/home.
... Expect 50-60% of the U.S. in 2010-2011, 70-80% by 2012-2013.

FTTN is designed for one meg up, 10 down, plus of which 15 are required for TV. It's really a DSL upgrade and is also known as "fiber to the press release."
... AT&T's half of the U.S., as well as the territory of Qwest, Embarq, Century, and others will have to depend on cable for high speed.

---------------

This first became clear in 2005. Late in 2005, both Brian Roberts of Comcast and and Bob McIntyre of Scientific Atlanta were confident of deployments before the end of 2007. Like many other new technologies, the deadlines slipped. As I write in late November, 2008, only the downstream is shipping in production quantities.

Will Comcast and the other cablecos price to win? That's the open question