An update on the state of broadband competition in the U.S.

and comparing different digital divide denominators.

I’ve written previously about how the new FCC maps change the digital divide denominator — instead of measuring “how many Americans are unserved by broadband?” there will be a tendency now towards measuring the digital divide in terms of “unserved broadband serviceable locations”. We can move between the denominators, but it’s a little bit complex. Overall, the new maps show that 4.4% of Americans are unserved by 25/3 broadband, or 14.7 million people. An additional 4% of people are underserved, or 13 million people.

To understand why the numbers change as we change denominators, it’s helpful to start with the “served” numbers. There are 98.5 million served BSLs, and 128 million served housing units. 88% of BSLs are served, but 91% of housing units are served. This increase is from apartment buildings that represent a single served BSL but represent many served housing units. For every served housing unit there are 2.37 people on average, so 128 million served housing units is 303 million served people, and still just over 91% are served.

There’s a different story in the unserved. The new FCC maps show there are 7.8 million unserved BSLs, 7% of all locations. But only 4.8% of housing units are unserved. In raw numbers there are actually fewer unserved housing units than BSLs, the opposite of what happened with served BSLs converted to housing units. This is likely what I identified in a previous post: in rural counties most likely to be unserved, there are more BSLs than housing units. 6.7 million unserved housing units is 14.7 million unserved people, 4.4% of the U.S. population.

I’m personally partial to residential housing units as the most intuitive policy-oriented digital divide denominator. It’s days may be over. The Broadband Serviceable Location Divide may be here to stay.

---

Critically, the new maps give us a view of the what the broadband competition situation is in the United States. We want to answer the question of how many options a household has for broadband service. Using the new maps, and a 100 Mbps download and 20 upload throughput as the threshold for acceptable broadband, 37% of households have access to 1 offering, 34% have access to 2 offerings, and 18% have access to 3 or more offerings.

This is slightly more concentration than was present in the most recent Form 477 data. In that data, as of June 30, 2021, 21% of households had access to 3+ offerings, 36% had access to 2 offerings, and 34% had access to only one. In each case, the new maps show a few percentage points less competition than the Form 477 data. This is the expected direction of movement: in the old maps if two providers evenly split a Census block, the block appeared to have 2 offerings; now it will show that every house has access to only one or the other single offering.

I looked briefly at what the numbers look like if the threshold for acceptable broadband is 25/3 instead of 100/20. In short, because of questionable FWA and DSL advertised speeds of 25/3, as detailed here, those charts are so unrealistic they aren’t worth showing. Even if it’s true that almost half of American households have access to 3 or more 25/3 providers, that doesn’t mean half of American households have robust competition for home broadband.

Using a threshold of 100/20 is intended to provide a more realistic picture of the options an American household faces as they choose home broadband. I’m not sure it’s working. All the same questions about whether all technologies are equal creep in. Almost 18 million locations have an offering for licensed fixed wireless at 100/20. Over 4 million have a DSL offer at that throughput. Can those technologies reach 100/20 in all cases to all those households? I’m not sure. Given the choice between a fiber or cable wired option at 100/20, how many households are choosing the FWA or DSL option?

A chart that likely rings more true — certainly for the urban and suburban population — looks at just fiber and cable offerings. Under these assumptions, 13% have access to zero offerings, 50% have access to 1 offering, 31% to 2 offerings, and only 6% to 3 or more offerings. Two things stand out to me: 1) 87% of American households having access to cable or fiber seems pretty good, and 2) there is a lot more concentration (less competition) when measured this way.

It’s hard to know what we should expect of this number. The FCC publishes a stealthy Internet Access Services Report. The last one was released March 9, 2022 with Form 477 data current through June 30, 2019 (a long time ago). It includes satellite service, but that isn’t a factor at the 100/20 level in 2019. The Y-axis is “Percentage of developed Census blocks” which is a weird denominator. It shows a lot of concentration. Only 4% of developed Census blocks had 3 or more provider options at the 100/20 level. Still, it was a long time ago and there has been a lot of buildout since then.

Fun fact: If you reply to the substack email it will come to my inbox if you’re trying to reach me.