Tuesday, April 9, 2013

If Google Can Do Google Fiber, Can Telcos and Cable Companies?

AT&T says it is prepared to build an advanced fiber optic infrastructure in Austin, Texas, capable of delivering speeds up to 1 gigabit per second, provided it gets the same terms and conditions as did Google Fiber. That’s interesting.


Some of us have been skeptical that Google Fiber would be able to build its 1-Gbps fiber to home network at costs markedly different from what a telco might expect to be able to do, in the same markets.


But always also has meant that if a telco really wanted to do so, it could build a 1-Gbps access network, today, particularly if “spot upgrades” in the way Google Fiber builds, prequalifying neighborhoods by requiring a certain percentage of households to indicate they will buy.

To be sure, there are overhead and other operating cost issues. But AT&T’s indication of interest also shows that, when push comes to shove, AT&T will respond to new competition in its markets.

Google Fiber buys from the same suppliers and uses the same construction contractors ad a telco would do.

Some might argue that inducements provided by Kansas City, Kan., Kansas City, Mo. or Austin, Texas can materially affect network cost. Some have doubted that, all along.

Google should have lower overhead, and perhaps lower operating and marketing costs based on the way it is prequalifying construction. But nothing of that sort should be meaningful enough for Google Fiber to magically undercut current telco FTTH pricing.

Instead, the issue is the business model, in particular the ability to make money selling 1-Gbps Internet access for $70 a month, while giving away free 5-Mbps service, and selling a video entertainment-plus-broadband package for $120 a month.

AT&T says it will build a 1-Gbps network in Austin if is gets the same geographic scope of offerings, rights of way, permitting, state licenses and any investment incentives Google Fiber has gotten.

Google insists the deal it has with Austin is non-exclusive, and involves no economic incentives, so it is possible AT&T could get such permission.

AT&T also says doing so in Austin would not materially affect its announced 2013 capital investment plans. How could that work? Assume AT&T already has allocated capital for Austin network upgrade purposes in 2013.

AT&T might do what Google does, polling neighborhoods and setting minimum thresholds for take rates before agreeing to build. So, in principle, AT&T could do its own polls, find out where a critical mass of potential customers exists, and then spot build in those neighborhoods, just as Google will do.

So far, there isn’t too much evidence that Google has been able to make a dramatic breakthrough in terms of basic infrastructure cost, one might conclude from an analysis of Google Fiber costs in Kansas City, Kan. and Kansas City, Mo.

That doesn’t mean Google Fiber has “failed.” AT&T’s offer suggests Google Fiber already has won.

Already, Google Fiber might be showing that it is possible to build 1-Gbps fixed networks at costs roughly comparable to what others are spending to create slower-speed networks.
Bernstein analysts Carlos Kirjner and Ram Parameswaran estimate that it costs Google $464 to connect a Google Fiber “broadband access only” service, and $794 to connect a customer buying broadband and video service.

That first wave, of 12,000 homes on “day one” of the service equates to an eight percent penetration rate, and implies a cost of $10 million for Google, further implying a total cost of $94 million for the Kansas City project. That includes about $42 million for the Kansas portion and $52 million for the Missouri.portion.

But those are the “activate a customer” costs, and do not include the cost of the actual backbone network.

Kirjner and Parameswaran say it will cost $84 million to pass (but not actually connect) 149,000 homes. Some $38 million will go into Kansas City, Kan., and $46 million into Kansas City, Mo., with the cost per home, for the network, costing  $674 in Kansas and $500 in Missouri.

The actual cost “per customer” then hinges on assumptions about take rate. If a third of homes passed actually buy service, then the full cost of the network is apportioned against 33 percent of passings. At that level of adoption, the network alone will cost about $1500 to $2000 per customer.

That is quite consistent with what other service providers could reasonably expect. You might also argue that $120 revenue per household likewise is within general norms for the fixed network business.


Bernstein says “the incremental cash investment to grow to 18 percent penetration in the first year will be of approximately $2 million, with $15 million in incremental cash costs offset by $13 million of contribution from users.”

Bernstein estimates that double-play customers will bring in $64 a month and broadband-only customers will bring in $47 a month.

Kirjner and Parameswaran estimate that a wider rollout around Kansas City for 300,000 homes would more than double build-out costs to $170 million (before acquiring customers and connecting those homes).

The point is that Google Fiber might be succeeding, at least for the moment in Austin, in prodding AT& to upgrade to 1-Gbps.

No comments:

Many Winners and Losers from Generative AI

Perhaps there is no contradiction between low historical total factor annual productivity gains and high expected generative artificial inte...