NNSquad - Network Neutrality Squad
[ NNSquad ] Costs of broadband deployment
------- Forwarded Message
From: David Farber <dave@farber.net>
To: "ip" <ip@v2.listbox.com>
Subject: [IP] WORTH READING The cost of bb deployment
Date: Fri, 3 Oct 2008 04:20:46 -0400
Begin forwarded message:
From: Dave Burstein <daveb@dslprime.com>
Date: October 2, 2008 11:21:03 PM EDT
To: dave@farber.net, Tom Bleha <tbleha@earthlink.net>
Subject: The cost of bb deployment
Dave: If this is too long for IP, Tom's ansers in in the first four
paragraphs and I'll put up the rest on my site.
Quick answer: $5-15B if you'll accept 50 meg down, 50 up, 95+% of the
time (DOCSIS 3.0) and only use the money to fill in the holes the
cablecos and telcos aren't about to reach anyway. $100B if nothing
was already being done and you wanted fiber almost everywhere. The
later is realistic on the engineering costs, tough on the companies
that will be hurt badly and have enormous political power.
Some back of the envelope numbers. These should reach 94% +-
3% . The $5-15B will sound low to most people who haven't been
watching DOCSIS 3.0. The cable upgrades are looking to come in very
cheaply, $100-400/per home.
Fiber at $1,000/home for 95-100M homes yields the standard
estimate of $100B. Verizon's Pete Thonis says they are spending less
than $700 to pass each home, with additional costs for the
installation. Since they already done 11M homes, that's probably a
good estimate for most of the country. At least a third of that is
already in the pipeline, especially Verizon. 150+ meg symmetric,
highly reliable, low maintenance.
DOCSIS 3.0 for about $25B should offer 90M U.S. homes 50 meg up,
50 meg down 95+% of the time. Most of that is already committed, so
that $5-15B (much in loans) should get good speeds to all but 10% of
the U.S. Comcast alone has promised service to 20M homes by 2010.
Adding up the projections of the other cablecos, we can expect 40-60M
homes servable by 2012-2013, half the country, without any government
intervention, program or money. Cablevision's Tom Rutledge estimates
less than $100/per home for each of their 4M homes, although I think
the upgraded modems and set tops will raise that some. Other systems
may cost more to upgrade, but a fraction of what fiber everywhere
would cost.
So the hard part to estimate is the cost to serve the last 5-10%.
I'm confident that $10B invested could get the the U.S. from the 90%
(above) to 94-97%. 96% of the U.S. is passed by cable, of which 96%
can get cable modem service today, or 93% of the U.S. (Kyle McSlarrow,
NCTA.) Telcos can offer high speed to some of the 7% cable doesn't
reach, but I don't have a good source of data.
Some equipment doesn't upgrade that easily, so Cablevision's cost
may be lower than others and we don't have enough data for me to be
more precise than $100-400 per for all but the last 10%. I believe
many of those 10% can be served by upgrading older cable systems and
extending some lines, but I don't have the data to give a good answer.
Recap
Somewhere between 45M and 70M of the U.S. homes will be servable at
50 meg or more by 2012-2014 by the current company plans. 10M, mostly
rural, will be hard and expensive to serve at 50+ meg, and it's hard
to estimate. But there's almost $10B/year going into the rural subsidy
under ICC and USF, much of which is not being productively used.
There's plenty in that program to cover what it needs.
The question then is why 20-40M homes probably won't get those
speeds by 2013, and how to upgrade them. Nearly all have cable
available that is only moderately expensive to upgrade. Some will be
in Charter territory, a cableco on the edge of insolvency that has to
hold back spending. Others may be in Time Warner Cable territory,
where the CFO projects reducing capex by 5-8% each year, to pay off
the billions being moved to the parent company in the spinoff. Some
will be on older cable systems more expensive to upgrade, in rural
systems with prohibitive backhaul costs, or be those remaining at
companies like Cox that may not finish their rollout as quickly as
Comcast.
For Charter and the other cablecos that have difficulty securing
financing, a government loan program should solve the problem. The
high cost of backhaul for some rural carriers is usually due to high
rates because of little competition. That can be fixed directly. Many
parts of the cable networks face weak telco competition, so they will
fall behind. A small amount of government support, offered to cable or
telco, whichever moves first, has proven effective in the U.K. in
inducing the carriers to invest faster or lose too many customers.
Incidentally, somewhere between 20% and 40% could be saved doing
fiber to the basement/curb, with 100 down/50 up standard speeds.
That's what Verizon is doing building hard to fiber now, and half of
Japan's 8M (?) "fiber" homes are really fiber to the basement + VDSL.
100/50 is pretty good. FTTN at AT&T is really DSL from up to a mile
away, with indicated speeds of 1 up, 10 down. Brilliant pr calling it
"fiber to the node," but it's virtually the same as SBC's 1999 Project
Pronto DSL, and slower than most DSL homes in Japan have been getting
for five years.
The ranges here are wide, because I'm calculating based on
equipment only now coming to market, and speculating on investment
plans five years out. I have reliable sources for most of this, but
would welcome any better data or corrections. European costs look a
little higher.
The most difficult unknown is the speed of DOCSIS 3.0 for a given
price. Cable modems are on a shared circuit, with 3.0 beginning with
120 up, 160 down, with products announced for 320 down. The sharing on
cable has proven to work remarkably well, with most customers able to
get reliable 15 and even 20 meg speeds on 37 megs shared. With more
video going over the net, things might change. On the other hand,
160/120 gives much more headroom if a few users ask a lot of
bandwidth, so the sharing should be more effective.
This doesn't mean that DOCSIS 3.0 can't hit the 50 meg suggested
above consistently. It is specified up to a gigabit (shared) and one
very respected vendor was considering put in a bid for Singapore's
gigabit network. It will cost more for the higher speeds, but I've no
good way to guess that, or the opportunity costs of requiring so much
of the cable spectrum. We have pretty good estimates of future
bandwidth demands (see Cisco), but that's not guaranteed.
So all of this is back of the envelope, but I've reported on
broadband since 1999 and think the assumptions are realistic.
Improvements welcome.
db
Designed to scale up to a gigabit.
Note caveats at end before using these numbers for serious analysis.
- - - -------------------------------------------
Archives: https://www.listbox.com/member/archive/247/=now
RSS Feed: https://www.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com