Sunday, May 27, 2012

Google Already Seems to Have Learned Something about Access in Kansas City

Some say Google is backing off its commitment to provide wholesale access to third parties as part of its 1-Gbps Kansas City network. Early on, Google seemed to believe it would offer third parties access, but it might be that Google already has learned how difficult it actually is to make business with an advanced fiber-to-home network.

Google might have concluded that any realistic hopes of operating the network as an money-making enterprise, as well as its hope to encourage lots of application innovation, might only be possible if Google acts as a retail provider and reaps all the revenue.

Also, if the point is to create a test bed for new applications, Google might now be concluding that it will learn more if it takes a more direct and active role in sponsoring such applications. One of the problems with operating as a wholesale provider is that you don't actually learn very much about what retail customers want, and are doing, because you don't actually have those customers.

Some think the wholesale approach would have been more interesting, as GigaOm tends to argue. But some might say the realities of the fiber to the home business now are clearer. As other triple-play service providers can well attest, offering such services in a competitive market, and Kansas City already has strong, entrenched cable and telco providers, is difficult.

If Google only wanted to create a "test bed," and didn't mind losing quite a lot of money, a wholesale approach might be feasible.

Utopia, an open access network operated by 16 municipalities in Utah, is among those entities trying to prove that a wholesale model can work, commercially.

Google might not believe it can learn as much, much less turn a profit, without offering and managing a retail service, allowing it to work with partners to test new services directly.

No comments:

"Tokens" are the New "FLOPS," "MIPS" or "Gbps"

Modern computing has some virtually-universal reference metrics. For Gemini 1.5 and other large language models, tokens are a basic measure...