As service providers gear up for increased fiber builds, financial analysts at MoffettNathanson are questioning whether another fiber bubble could be coming.
“Our conclusion is that there is simply too much enthusiasm for too many overbuilds,” the analysts wrote in a note released today. As a result, the return on many of the planned projects will fall below the cost of capital, the analysts’ caution.
As the analysts note, the “market’s embrace” of fiber deployment is dependent on four assumptions:
- The cost-per-home to deploy fiber will remain low
- Fiber’s eventual take rates will be high
- Take rate gains can be achieved even at relatively high average revenue per user (ARPU)
- The capital to fund these projects will remain cheap and plentiful
The analysts question all these assumptions but are particularly concerned about deployment costs measured on a per-home basis. These concerns are based on an analysis of population density nationwide and the differing costs between aerial and buried cabling.
A Fiber Bubble?
“The densest 10% of the country is comprised of census blocks with an average density of 19.1K housing units per square mile of land,” MoffettNathanson observes. “The drop-off from there is dramatic, with the second densest decile less than half as dense.”
This suggests a doubling of cost per home passed from the first 10% to the second 10%, the researchers note. They hasten to add that the difference may not be quite so extreme because homes tend to cluster along roads, even in lower-density areas.
Nevertheless, the U.S. is a very low-density country overall – and as the researchers note, fiber deployments to date have focused on the most densely populated areas, which means that areas that have not yet seen fiber deployed are in areas that are costlier to serve.
Exacerbating this: Lower-density areas are more likely to have fiber if providers were able to use existing aerial infrastructure – and it costs about twice as much to deploy buried fiber, which suggests that even some relatively dense areas that do not yet have fiber will likely have above-average deployment costs.
“Based on commentary from the various telcos, the cost of high-density, largely aerial builds has averaged around $400-$500 per passing on the low end,” today’s fiber bubble note states. “For the next phase of fiber builds, Lumen and Frontier, for example, expect average costs of around $1,000 per passing or less, as they move into areas with slightly lower densities and/or slightly higher mixes of buried infrastructure.”
Even less dense areas have substantially higher deployment costs. In areas where Charter won RDOF funding, the company expects an average cost to pass of about $4,700 gross, or around $3,600 per passing net of RDOF subsidies, the analysts note.
The analysts ran numbers on a variety of scenarios, assuming different ARPU, different margins, different costs of capital and different deployment costs and quite a few of those scenarios yielded negative net present value per subscriber.
Factors that could threaten ARPU and margin include how widely fixed wireless is adopted and how cable companies respond to new fiber competitors. The analysts note, for example, that cable companies see their wireless offerings as a means of retaining customers by requiring customers to purchase home broadband from the cable company to get discounts on wireless – discounts that can be quite substantial for families with multiple cellphones.
Assuming MoffettNathanson’s concerns do indeed materialize and the internal rate of return for some new fiber projects falls below the cost of capital, a key question will be how soon financial markets recognize that, the researchers note. And that, in turn, will impact the cost of capital.
I think there’s another important question, too, though.
What About Government Subsidies?
The analysts note that they didn’t factor in the impact of new government broadband funding in their analysis, but it seems like the picture might not be so bleak if we assume some of the upcoming fiber deployments could be subsidized.
That, however, raises another question – how costly-to-deploy an area must be to qualify for government funding. The government isn’t looking at it that way; it’s looking at whether an area has high-speed broadband available, but the two factors undoubtedly will be closely correlated.
We won’t know the answer to that, though, until we know how many homes lack high-speed broadband, and we won’t know that until we have accurate broadband maps. The expected date for that is late this year or sometime next year.
It’s true, though, that areas with cable broadband generally won’t qualify for subsidy programs, and if we believe MoffettNathanson’s fiber bubble analysis, that means the next digital divide could be between areas with fiber, including the most urban and most rural areas, and those in between who can only get cable.