I will readily admit that I have a fiber bias when it comes using infrastructure grant funding. This is a policy issue for me and is not limited only to broadband. The federal government is handing out huge once-in-a-lifetime infrastructure grants. I think federal infrastructure grants should be used to build infrastructure that will last as long as possible to create the longest-term public good. I am perplexed when I see cities using ARPA funding to buy firetrucks and computers if that city has big infrastructure deficits for things like water systems or public housing. I obviously have no bias against firetrucks or computers – but they aren’t infrastructure.
My position raises the obvious question of what qualifies as infrastructure. In my mind, infrastructure is an asset with a long useful life. I think everybody would agree that roads, bridges, and water pipes are infrastructure. These are assets that will be useful to the public for a long time.
It’s a little less clear with broadband infrastructure. Conduit is clearly infrastructure, and there is no reason to think that conduit won’t still be functional in a century.
Fiber is a little less clear-cut. I remember when fiber was being constructed in the 1980s, we thought of it as a 40-year asset. There are some fiber routes built in the 80s that are showing wear, but a lot of fiber built in the 1980s is still going strong.
But fiber manufacturing technology has improved significantly since the 80s. Fiber is now much clearer and less likely to grow opaque with age. Fiber today has much tougher outer sheathing. We’ve also learned a lot about fiber installation techniques, and many of the problems that have arisen from older fiber are due to stress placed on the fiber during construction. While the manufacturers won’t go on the record on the useful life of fiber, I’ve been told privately by fiber manufacturing engineers that fiber ought to last 70 or 80 years if installed properly. That sounds like infrastructure.
The biggest weakness of all broadband technologies in terms of longevity is the electronics. This applies equally to fiber and wireless technologies. The conventional wisdom is that most broadband electronics are good for about 10 or 12 years. Part of this is due to true obsolescence, where circuit cards wear out after being used non-stop for a decade. But part of the obsolescence is due to vendors that stop supporting older technology. It becomes harder each year to support a network if vendors aren’t making replacement cards. Everybody that’s owned a broadband network for twenty years can still point to a few pieces of gear that are still chugging along – but for the most part, electronics have to be replaced over time.
If my philosophy is that infrastructure is an asset that lasts for a long time, how do I reconcile any broadband grant with relatively short-lived electronics (at least short-lived on an infrastructure time scale)? I define infrastructure in the same way as lenders. Federal bond rules say that a borrower can’t have a bond term (the years to pay back the loan) that is longer than the average economic life of the assets being funded. A lot of commercial banks have a similar test as part of evaluating infrastructure loans.
What’s the average useful life of a fiber network? Consider the following real-life example of a recent rural fiber project I worked on.
|Average Life||% of Project|
|Fiber||40 – 60||60%|
Folks can disagree about the average life of fiber. I’ve been conservative since I think fiber will last longer than shown in the table. If you assume that fiber is good for 40 years, the weighted average useful life of the above network is 53 years. If you assume the average life of fiber is 60 years, the useful life climbs to 65 years. Aerial fiber networks have a lower economic life without conduit, but the range of expected life is still between 37 years and 53 years.
Other broadband technologies have a much shorter economic life. My guess is that the economic life for Starlink is under ten years since the satellites are designed to fall out of orbit by then. There are probably components in satellite base stations that will last longer – but most of the investment is in the satellites.
It’s hard to do the same math and get a useful economic life for the typical fixed wireless network that is higher than 15 years. It is possible to construct a fixed wireless network with a higher average useful life. Well-built towers can easily last 75 years. Fiber backhaul to towers has the same useful life as last-mile fiber. However, my reading of the BEAD grant rules is that it will be difficult to win funding to build towers or middle-mile fiber. A fixed wireless grant that funded towers and fiber would probably pass my infrastructure sniff test.
I can’t begin to estimate the average useful life of an FWA cellular network, but it’s not very long. These are networks that are built to use the excess capacity of cell phone networks and are not constructed just for broadband. When I consider the rapid evolution of cellular technologies, it seems likely that any system built today will be technically obsolete when real 5G standards are finally implemented.
Hybrid-fiber coaxial systems have an average economic life that is about the same as the lower range of fiber network lives. The coaxial wire won’t last as long as fiber, but forty years is a reasonably assumed life for the coax.
The NTIA tried to express the same sentiment as me without defining why. The NTIA said early on, after it was given responsibility for the BEAD grants, that the agency favors fiber. It would have been a lot clearer if the NTIA said instead that it doesn’t support infrastructure grants for projects that don’t have infrastructure useful lives – I think that is what they meant. If the agency had set a definition of infrastructure as projects with a useful life of at least thirty or forty years, we wouldn’t be having the discussion of funding networks with short useful lives.