I hope for Google's sake that it understands the cost to build its 1 gigabit/second ultra high-speed fiber network experiment announced today could be $750 million or more. Even for Google that's a very big number, especially considering the company has said it has no intention of actually pursuing this as a business. Of course, we don't know exactly what Google is forecasting its project costs to be, but using Verizon's FiOS numbers wouldn't be a bad starting point to do the math. So here goes.

Google said it would offer the gigabit service to between 50,000 and 500,000 people. Let's start at the high end of that range. Verizon has disclosed that it will spend $18 billion to pass approximately 18 million homes in its footprint with its FiOS fiber-to-the-home network. It's not fair to do a straight average and assume that Verizon is still paying $1,000/home passed given that its costs have no doubt declined over the years. However, in Google's case, since it has approximately zero experience laying fiber in neighborhoods, and won't get the same level of vendor discounts that Verizon enjoys, it is probably fair to assume Google will spend at least $1,000 per home passed. So if it goes all the way to 500,000 homes, that's $500 million in neighborhood build-out costs.

But that's only to wire the neighborhoods, then the service has to be deployed in the homes themselves. That means in-home wiring, on-premise equipment, labor, trucks, insurance, overhead, etc. Estimates for Verizon's per home cost vary, but $500 is in the range often cited. In Verizon's case they're also deploying a set-top box to deliver TV, which Google hasn't announced plans to do (more on that below), so that cost should be deducted. But on the flip side, once again, because Google has never wired a consumer's home (that I'm aware of anyway) it has a steep learning curve ahead of it, meaning its costs could be much higher than Verizon's.

But to make things easy, let's just use the $500 per installed home. So 500,000 homes at $500 apiece, another $250 million for the project. Add it to the $500 million for the neighborhood build-outs and the total is $750 million. This assumes Google decides to go all the way to 500,000. Obviously if it stopped at 50,000, the costs would be a lot lower.

Executives at cable operators and telcos - who build and operate residential networks for a living - are very familiar with modeling network deployment costs. But I wonder, how familiar do you think Google is? Does it know what it has bitten off here? And for what benefit exactly - to test next-generation apps? Hmm. Everyone knows video is the biggest bandwidth hog; an expensive experiment isn't going to change that. And also remember, Google only plans to sell broadband Internet access, not a full bundle with TV or voice. It says it will do this at competitive prices, which means around $50-$100/mo. At these revenue levels and with operating costs that I haven't even mentioned, it's inconceivable to me that there's a positive business case for Google's gigabit experiment.

I'm all for innovation and for pushing competitors along. But Google's experiment really has me scratching my head. No doubt the folks at Verizon, Comcast and other big broadband ISPs are wondering as well. It's one thing for Google to throw $2.5-$3 million at a 52-second Super Bowl ad, but quite another to be contemplating a $750 million experiment with ambiguous goals. What am I missing?

What do you think? Post a comment now (no sign-in required).