Second look: New FCC Maps

Last week I published an analysis of what the new FCC maps show for locations unserved and underserved by broadband. With some time to think about the data, here are more thoughts on what these new maps mean for new deployment programs.

7.8 million unserved locations, or 6.9% of the total locations, meet expectations almost exactly. It’s 118% higher than the 3.58 million unserved housing units in the Form 477 data. Most people, including me, assumed the number of unserved would double. It did.

The number of underserved dropped, which is a surprise, but maybe shouldn’t have been. In the Form 477 data there were 7.35 million underserved housing units. That fell to 6.0 million locations in the new maps. That’s a 19% decline. It makes sense. Under the previous maps, a Census block was “served”, “underserved” or “unserved”. It makes sense that previously-partially-served and underserved blocks both gave up locations to the unserved category.

In total, the new maps show 13.8 million unserved and underserved locations. My previous model used 23.1 million unserved and underserved housing units; Jim Stegeman of CostQuest estimated 23-25 million locations wouldn’t have access to technology that would make them served. While the new maps met expectations for unserved locations, there are significantly fewer locations to reach than we thought when we include underserved.

It will be some time before the dust settles on the new maps. It seems like there will be at least one more round of location and availability challenges before the new maps are used for allocations by NTIA. Things could still change.

Regardless of what happens with challenges, I believe the conclusion of my previous analysis will not change: $42.25 billion is enough money to reach all the unserved, and make significant progress on the underserved, though high cost states will have a much tougher time. There are still about 30% of unserved locations that will be covered by RDOF and ineligible for BEAD funding (not to mention other programs). The private capital match (at least 25%) will bring down the burden on BEAD capital, particularly in low cost areas that might get multiple ISPs bidding on the same areas. If it costs an average $6,214 to reach an unserved or underseved location with broadband (the number from my previous model), we’d need $85.7 billion to reach the 13.8 million unserved and underserved locations. (Again, that’s before RDOF locations are excluded, and before private capital is factored in). $42.25 billion can go a long way.

---

Home Internet below 100 Mbps download and 20 upload throughput (with consistently low latency) is not good connectivity. States, and the federal government, should be aiming to get everyone above that threshold. The chart below shows our starting point. Montana and Illinois are above 30% unserved and underserved. Alaska, West Virginia, and Arkansas make up the next set.

Since the funding is based on the number of unserved, there’s no doubt that’s the most important number. I previously highlighted how Kansas looked to be in trouble: under the Form 477 data they would receive funding based on 7,432 unserved housing units, and would be trying to connect 140,000 housing units including the underserved. Things look better for Kansas now. They’ve receive funding based on 56,000 unserved locations, with the goal to connect 149,000 total locations.

Nationally, each unserved location is worth $5,400. States with small numbers of unserved get more money per unserved location because of the $100 million minimum allocation. For example, D.C. has 93 unserved locations but gets an allocation of $100,391,000. Ten percent of the money is based on the number of “high-cost” locations in a state, but that isn’t enough to push significant money into high-cost states.

The chart above has the BEAD allocation dollars per unserved and underserved locations. Higher numbers are better - it means more money to reach fewer locations. There are surprises on both ends. North Dakota gets about $160 million to reach 9,800 unserved, and only has 2,600 underserved to reach after that. That’s almost $13,000 per location, which is better than anything other than D.C. and Rhode Island which are outliers.

Illinois is interesting to watch. They get $1.1 billion to reach 192,000 unserved locations. But they have 1.1 million underserved locations that they also should reach. At $882 per location, they are in last place.

All unserved locations are not created equal in terms of the cost to reach them. To reach an unserved location in Kansas, I previously estimated it would cost $16,100. To reach an unserved location in Maryland, it would cost an estimated $1,000. Using these estimates, nationally it would cost an estimated $6,200 to reach an unserved location. Remember, if we only endeavored to reach unserved, there would be $5,400 per unserved location available.

At the state level, the cost to reach all the remaining unserved and underserved is the average cost per location multiplied by the number of locations. The cost to reach those locations minus the allocation in BEAD funding gives a representation of which states have the biggest gap, and might have the hardest time reaching all their underserved. At an average cost of $16,100 per location, and 149,000 locations to reach, Kansas needs $2.4 billion in investment. With a BEAD allocation of $435 million, that leaves a shortfall of $2 billion. That’s the 9th biggest shortfall.

Illinois is still the one in trouble. Their cost to serve is estimated at $9,400, but as discussed their huge number of underserved pushes their cost to $12.6 billion with an allocation of only $1.1 billion. It’s so bad it’s possible there’s an issue with their data but I haven’t found it yet and the FCC site appears to show the same relationship.

Again, more work needs to be done to refine this analysis. Many locations are covered by RDOF and will not be eligible for BEAD projects. Private capital matches further reduce the necessary BEAD capital. I’ve long had a theory that locations in partially-served Census blocks (we now know which ones they are) should be less expensive to reach since there is already service in the block. We now have the data to test that hypothesis.

Update 12/8/22: Comcast appears to have filed the vast majority of its locations in Illinois at 300/10 which makes those locations “underserved” in the absence of other service. I believe this to be an error. In previous Form 477 filings, Comcast filed most of its Census blocks in Illinois with Docsis 3.1 at 1200/35.