Google Fiber and Creating the Next Generation ISP
Showing ISPs How It's Done
With its plans to bring gigabit broadband to Kansas City, Kan., Google is changing the fate of that city, but it's also setting out to build a new generation Internet Service Provider, one designed for the type of world where connectivity drives innovation and is an irrefutable aspect of our lives. Milo Medin, head of Google's fiber efforts, explains that his goal is nothing short of causing a revolution—a revolution that may spread to other towns.
"As I said, we're still having conversations with other markets and will try to build out in other areas in [Kansas City]," Medin says. So maybe Austin, Tex.—which was passed over—shouldn't lose hope yet?
When he formed the company that became Excite@home, Medin helped take the Web from dial-up speeds to speeds of 4 Mbps to 5 Mbps, and he draws a similar parallel between his efforts there and what Google is trying to do. "It's a little bit like first days of @home, when the Web was all optimized for dial-up, and no one had always-on connections," he says. "Back then, people asked why would we need that, but the need for bandwidth seems to be insatiable … and there are businesses like Netflix that exist today because of it."
Above Ground and Below
At GigaOM, we get the idea that people will always need more bandwidth, but exactly how Google plans to deliver it is still not something Medin is willing to discuss. He declines to explain what vendors Google will use, or what type of fiber network Google will build, but he does say that in 70 percent of Kansas City, Google will be able to string fiber using utility poles. This will lower costs, because it means fewer streets will have to be dug up. The other 30 percent of needed fiber will be underground and will take advantage of existing conduit that extends, in some areas, all the way to people's homes. So instead of digging up streets, the fiber can be threaded through the conduit for much less.
Because this is Google, cheap and optimized are the name of the game. The existing conduit, the utility pole access, and other things the city granted Google (such as on-site inspectors) will cut the cost of laying fiber significantly. Medin credits Kansas City's existing infrastructure, streamlined government rules, and general business friendliness with some of the reasons Google chose the city. Medin is essentially trying to work backward from an ideal cost (he won't share it) and trying to build and operate a network that can run profitably while allowing customers to pay this ideal monthly price.
"We looked at it and said, 'What do we have to do to make a service that has a good take rate?' because if only 1 percent of the residents subscribe, then I'd consider that a failure," Medin says. "I have to say, 'What can I afford to charge?' and work the cost backward to support an architecture to drive the costs for this." He cites Chattanooga, which offers a gigabit network and charges people $350 for access to those speeds. But at that price, Google likely wouldn't see the acceptance rate it wants to see.
Sweating the Small Stuff
Medin adds that Google has managed to build out a huge infrastructure that supports such services as YouTube with only a relatively small number of people actually clicking on the ads that fuel most of Google's revenue. Those clicks add up to billions, but his point is that Google sweats the small stuff when it comes to costs and delivering speed for less. Google's existing infrastructure is one reason it will be able to support its fiber project; most backbone providers would charge an arm and a leg for providing access back to the Internet backbone for multiple gigabit connections.
Medin says Google has already built or contracted for those connections on its own network and doesn't pay what an ISP or provider might otherwise charge. That's an essential link in getting gigabit broadband for homes that is "accessible" from a price standpoint. Medin says one of the ways Google achieves this efficiency is by building what it needs or making do with cheaper alternatives instead of building a network like a telco or thinking in terms of multi-thousand-dollar gear.
That could be the scariest aspect of this whole project for ISPs: watching Google build out a network that can deliver gigabit speeds at accessible prices and then operate such a network profitably.It would blow ISP arguments about the price of bandwidth out of the water.Those arguments are currently leading ISPs to enforce caps on broadband and complain about services such as Netflix. So if Google can show ISPs a new model for offering residential bandwidth at a reasonable cost, then perhaps it could soon have the same influence on broadband costs and architecture as it has had in scaled-out computing.
We'll know soon enough. Medin says construction will begin in the fourth quarter of this year and Google should begin offering service early next year. As for the three-month delay in notifying a city, Medin confesses that the response to Google's announcement last February that it wanted to build a fiber network was so amazing that the search giant wasn't prepared for the onslaught of applications. At least it proves his point: People do have an insatiable need for faster broadband.
Also from GigaOM:
Three Trends Defining the Future of the Digital Home (subscription required)