How far might the broadband funding go? An update with data from the new maps

Last fall, I produced a model to estimate how far the BEAD funding might go, using estimates of the unserved and underserved from the old Form 477 data. The prediction was that with an optimal allocation between states, there would be almost enough money to reach all the unserved and underseved. Well, we’re getting closer to real and final data, and an update is in: $41.4 billion at an average national cost of $6,214 per location should reach 6.7 million locations. Considering the 3.5 million locations already committed to by RDOF, we should be able to reach 10.1 million locations, or 76% of the total unserved and underserved nationally.

As a refresher, it’s really quite easy to make a prediction of how far the money will go. We only need two pieces of data: the number of unserved and underserved locations, and the cost to reach them. First, lets look at the number of locations. Whereas before we needed to estimate the unserved locations, now we know. There are 7.8 million unserved locations and 5 million underserved locations — 12.9 million locations total. Maybe that changes a little bit in the second map. I doubt it changes much.

There’s an easy-to-forget x-factor. As of this writing, 3.5 million locations have been authorized for RDOF funding. These are locations that are intended to be rural and unserved. There should be 100% overlap between RDOF locations and BEAD-eligible locations. While RDOF authorized locations are counted as unserved for the purposes of the allocation of money, by rule BEAD money can’t be used to fund a project for the same location. States with lots of RDOF awards are winning on both ends — the location is counted as unserved which increases the allocation, but will get [mostly fiber] service from RDOF (assuming RDOF requirements are enforced). BEAD only needs to fund 9.4 million locations, not 12.9 million. It’s a 27% decline in the number of locations BEAD has to fund.

There are winners and losers already at this stage. Utah only has 1,743 RDOF awards authorized but 71,341 unserved. South Dakota has 2,552 RDOF awarded-locations and 30,646 unserved locations. California, Idaho, Montana, and Nevada also make the list for relatively few RDOF freebies compared to their unserved locations.

Some states fare better. Iowa has 40,768 RDOF-authorized locations but only 38,714 unserved locations — over 100%. Since my last look at this, some big Kansas RDOF authorizations must have come in — they’re now at 43,909 RDOF locations versus 55,831 unserved.

A big driver of the winners and losers is the “Starlink and LTD effect”. Starlink and LTD together won RDOF awards funding 1.2 million locations, but their awards were later rejected by the FCC. Montana is a good example of this. Starlink won 29,000 of the total 45,000 Montana locations that were originally in the auction, all of which were rejected and now need to built using BEAD funds.

Nationally, the average allocation per unserved or underserved is $3,218. That’s $41.4 billion (the allocation to the 50 states + DC) divided by 12.9 million locations. Once you factor in RDOF, the allocation per location goes up to $4,406. Mississippi, thanks to their high number of unserved relative to underserved, and their high number of RDOF allocations, more than doubles their allocated dollars per location, going from $3,764 per location to $8,376 per location after removing RDOF.

Next we look at the estimated cost to pass a location with fiber service. For this I’m using my previous model of the cost, which reverse engineers the RDOF reserve prices and then adds 25% to those costs. That results in an average national cost of $6,214 per location, with wide variation by state. In eastern states, the costs tend be lower — under $3,000 per location in Maryland, Florida, New Jersey, and Massachusetts. Whereas west of the Mississippi river, the average estimated cost per location in the state is much higher — $16,582 in North Dakota, $16,100 in Kansas, $15,621 in Nebraska, and $14,459 in South Dakota.

Another simple division problem gives us a sense for how far the money might go. Total allocation divided by average cost per location yields the number of locations that can be funded at that price. For example, North Dakota gets an estimated $159 million allocation from BEAD. With an estimated average cost to serve a location with fiber of $16,582, they’ll be able to reach about 9,581 locations, which is 77% of their 12,378 unserved and underserved locations. When we factor in the 1,901 locations that RDOF will reach, with BEAD funding, they should be able to reach 93% of their total. They’re one of very few states western states that might do this well.

North Dakota has a couple things going for it: (1) as a small state, the $100 million base allocation goes a long way on a per location basis. Almost two thirds of their total allocation is from the initial $100 million. And (2) they have a small number of underserved relative to unserved. The allocation is done on the unserved, but we really want to reach both.

Nebraska draws a very different hand. Nebraska has 98,000 underserved locations, 12% of all the locations in the state, but only 20,000 unserved locations, which drive the allocation. Nebraska’s allocation is a meager $2,136 per location, yet they still have high costs at $15,621 to bring fiber to each location. The result, unfortunately, is that Nebraska would only be able to bring fiber to an estimated 24% of their unserved and underserved.

I’m consciously blending together the unserved and underseved numbers — I view it as the real Digital Divide. The presence of licensed fixed wireless offerings at exactly 25/3 counting as underserved instead of unserved is all I need to know. While it obviously makes a big difference for the allocation, I suspect many states will make unserved and underserved eligible for funding, probably along with DSL and fixed wireless offerings if they’re claiming 100/20 service.

There’s another factor that I included previously but haven’t yet included here. BEAD grants require a private capital match of at least 25% — the federal government doesn’t plan to cover all the costs. If states can produce competition for these grants, there’s the possibility that ISPs could put in similar capital level to what they offered in RDOF, which was significant. Private capital will make the BEAD funds go even farther.

Yet another factor that I hope to incorporate in another revision is using localized cost-to-pass numbers instead of a statewide average. Every state will have a different cost curve for their remaining unserved and underserved. In some states, RDOF may have taken the most expensive locations off the table. In other places, the new maps may have exposed partially served census blocks where there is already service on the block and thus we’d expect the cost to serve those locations to be much less. I believe those factors will wash around in different directions but not change the fundamental results here. Time will tell.

And yet another factor is that states will be allowed to draw a line around “Extremely High Cost Locations” and above that price use wireless technologies instead of fiber to the home. That process will substitute high-cost locations for lower-cost locations, increasing the total number of locations reached.

Obviously this analysis is dependent on the cost model. I’m using my cost model because there isn’t anything else that’s public. I’ve heard that my cost model is unrealistically low, and also from some people that my cost model was too high. I think there will be a tendency to hand-wave about inflation, high labor costs, supply-chain shortages — all as a way to justify many increases in costs. A public cost model to compare grants against, and fostering competition in the process, are crucial to controlling costs.

In the fall I was optimistic about how far the BEAD money can go in closing the Digital Divide. The new maps, if anything, make me more optimistic. While I wouldn’t want to be in a high-cost state, overall, if we can control the costs, we might be able to reach 79% of the unserved and underserved. That’s a lot.

Thanks to Jon Wilkins at Quadra Partners for comments on early versions of this analysis.

Note: If you’re trying to reach me, replying to the Substack email gets to me.