Friday, December 7, 2012

Google Cannot Afford to Become a National Fixed Access Provider

Google’s Kansas City, Kan. and Kansas City, Mo. 1-Gbps access network naturally tends to
encourage speculation about what else Google might have in mind, including the notion that Google might try something on a wider scale.

It isn’t a completely wild notion, given earlier Google investments in municipal Wi-Fi, Clearwire and even a commitment to make a minimum bid on 700-MHz spectrum. Add in Google’s production of its own Nexus smart phones and tablets, as well as tests of a mobile virtual network operator network in Spain (albeit only as a closed test using Google employees), and one can understand the speculation.

Also, new European Union rules might encourage new providers to think about becoming specialized providers of service that is “roaming only.”

An analysis by Goldman Sachs should reassure potential “victims” that Google would not actually undertake an expansion of Google Fiber on a major scale in the U.S. market. The simple reason is excessive cost.

The analysis suggests that if Google really wanted to build out a new national 1-Gbps networks in most major U.S. cities, it would cost more than $140 billion. Most experienced telecom or cable executives would say it could easily cost that much, and take many years.

A more modest goal, of becoming a service provider for 50 million households, less than half of all U.S. homes, would cost perhaps $70 billion.

Ignore for the moment the likelihood that Google has lots of other users for its cash. Assume Google wanted to do so, and was willing to proceed by using internal funds, rather than borrowing money.

Goldman Sachs Telco analyst Jason Armstrong noted that if Google devoted 25 percent of its $4.5 billion annual capital investment,  it could build network out to about 830,000 homes each year, or 0.7 percent of U.S. households.

Armstrong estimates Verizon has spent roughly $15 billion building out its FiOS fiber network covering an area of approximately 17 million homes, by way of comparison.

At such a pace, Google would find it had to replace its first networks long before it even finished the first round of construction on the initial 25 million sites.

Also, if Google really wanted to do something about the economics of broadband access, you might argue the timetable and cost would be more amenable if Google spent money on mobile or at least wireless infrastructure instead.

Some estimate it will cost T-Mobile USA about $9 billion to build a national LTE network, for example.

The point is that 1-Gbps networks are expensive and time-consuming to create. Google’s primary interest likely is creating new market pressures for all other providers to match its own services in Kansas City.

The only issue is whether Google can do so by building in one city only, or would have to threaten building in other key cities as well.

But in all likelihood, Google would never want to waste capital by building a national network of fixed access networks. It is simply too expensive, and too limited, in terms of return from other investments it could make with that capital.

On the other hand, it is clear why speed matters to Google: speed matters for users and consumers.

Some 64 percent of consumers surveyed by Brand Perfect say slow loading was among the top two irritations when shopping online. Fast loading improves end user satisfaction, and also allows more ad inventory to be displayed.

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...