100 Mbps to everyone for $350 billion

We finally have a realistic estimate of the cost of bringing 100Mbps to every home in America... and Light Reading labeled the cost "jaw-dropping."
Want to provide 100-Mbit/s broadband service to every U.S. household? No problem: Just be ready to write a $350 billion check. Federal Communications Commission (FCC) officials shared that jaw-dropping figure today during an update on their National Broadband Plan for bringing affordable, high-speed Internet access to all Americans. The Commission is schedule to present the plan to Congress in 141 days, on Feb. 17.
Don't get me wrong, I agree that $350 billion is a lot of money. On the other hand, we spent nearly $300 billion on surface transportation over 4 years from 2005-2009. $350 billion buys a fiber-optic network that will last considerably longer. Additionally, such a network will generate considerably more revenue than a highway. In fact, these networks will pay for themselves in most areas if they can access to low-interest loans. Consider the comments of Deputy Administrator Zufolo (of the Rural Utilities Service) from my recent panel at NATOA:
Zufolo explained the RUS decision to use its $2.5 billion in funds primarily to subsidize loans and not provide grants, as the agency's best opportunity to make the more efficient use of the federal money and have maximum impact. Because the default rate on RUS loans is less than 1% and the subsidy rate is also low, only about 7%, it costs the government only $72,000 to loan $1 million for rural network development, she said.
Let's say that RUS decides to embark on getting 100 Mbps to everyone in a rural area - some of the projects will be riskier than the standard portfolio, so let's assume it costs the federal government $100,000 to loan $1 million (makes it easier math too). In order to spur the $350 billion investment for these networks, the government would have to put up $35 billion. But it would probably be more than that because some areas - Montana, Alaska, Wyoming, and other beautiful places will need partial grants on the upfront costs because even a loan at 0% interest may not be serviceable due to the challenges of spreading high fixed costs across so few people. If the U.S. were to commit to this, it would not do it in one year. Industry would have to scale up significantly - it would take a number of years to produce all the of materials needed to build a network on this scale, as well as training all the additional people who would be needed to enter the workforce in this sector. So we are talking about what - $7-12 billion a year to build the single most important infrastructure of the coming decades? This is a jaw-dropping figure in that we have not yet done it. No wonder other countries are leaping ahead of us. Remember though, this cost would build one network. We need to abandon the idea of each competitor building a distinct network with which to compete and to embrace a publicly owned open access network that enables many service providers to compete. The publicly owned network will deal with the largest costs of building the infrastructure - just as the public sector does in most aspects of our lives (roads, sewer projects, bridges, rail, canals, and significant portions of the electrical infrastructure 80-100 years ago). Update: I wanted to echo the calls of the Knight Center for Digital Excellence in calls for a much bolder vision than we have seen thus far:
Imagine if we had made the mistake of building ordinary roads when, in the 1950s, true progress required an interstate highway system. We are at a similar juncture, which is why the time calls for the high ambition of gigabit speeds.
Image used under Creative Commons License from Briar Press.