How far could the money go? Update with new Enhanced ACAM numbers

I’m asked all the time: “How far will the money go?” and “Which states can’t reach all their Unserved?” My last update to this model was at the end of January. A lot has changed. In this new update, we could theoretically reach 94% of the Unserved and Underserved nationally. We only miss 750,000 locations. The biggest misses by percentage are Iowa (61% of Unserved and Underserved), Idaho (66%), Illinois, Kansas and California (all 71%), Minnesota (76%), and Colorado and Nebraska (about 80%). It’s a startling estimate, so let me explain.

I find it helpful to think about this as a simple math problem: how far the money might go can be estimated by multiplying the number of locations that need service times the average cost to serve them.

There are 11.9 million Unserved and Underserved locations nationally. About 2.4 million locations are Unserved or Underserved and are part of the FCC’s Rural Digital Opportunity Fund program, and broadband will be deployed as part of that program. Also, according to my recent estimates, 1.3 million locations would be served by ISPs under the FCC’s Enhanced A-CAM program. Together, that’s 31% of the remaining digital divide that will get broadband through a program other than BEAD. And it doesn’t factor in the billions of dollars that went into broadband through the Covid recovery bills administered by the Treasury Department (the data is still too messy right now).

That leaves 8.3 million locations nationally that we need to reach with the BEAD program. We allocated just under $41 billion to the 50 states + DC with the BEAD program. When the goal was 11.9 million locations, that was $3,445 per location as a national average. But when we only need to reach 8.3 million locations, it’s $4,939 per location. The two FCC programs are doing a tremendous amount to lower the number of locations that BEAD needs to fund, leaving more BEAD money for the remaining locations.

RDOF and Enhanced ACAM are both targeted at rural areas, so rural states disproportionately benefit. More than 80% of North Dakota’s Unserved and Underserved will be part of either RDOF or E-ACAM; it’s more than 50% in South Carolina, 49% in South Dakota, and 48% in Nebraska.

Now let’s turn to how much it costs to bring fiber-to-the-home to a location. As I’ve discussed many times, I reverse engineered costs from the reserve prices of the FCC’s RDOF auction. There are some caveats to this as I’ll discuss below, but I also add 35% to those costs to capture inflation and supply chain issues. Nationally, I have an average cost to pass an Unserved or Underserved location with fiber as $7,906.

Already we’re seeing how the math could work out in a lot of states. While the national average cost to serve might be $7,906, private capital from the ISPs is expected to fund part of that cost. With an average allocation of $4,939 per location, if states can get high match rates and keep costs low, they might have enough money.

Another thing the FCC did with RDOF, but especially E-ACAM, is take high-cost locations off the board, reducing the average cost. Nationally, the locations that are left after removing RDOF and E-ACAM have an average cost of $7,252, down an important $650 per location. In South Dakota, the average remaining location costs $13,843, down from $19,316 before removing RDOF and E-ACAM. In Nebraska, the remaining locations cost $16,301, down from $19,575 before removing E-ACAM.

I still believe there are states — 11 of them to be exact — that have no chance of bringing FTTH to all their Unserved and Underserved, usually because of high build costs. One of them is Nebraska. In my previous analysis, I had Nebraska able to fund maybe 15% of its Unserved and Underserved. E-ACAM is a game-changer for Nebraska.

In the chart below, we “fund” all the Unserved locations first, from left-to-right in increasing order of the cost to serve the location. We don’t need to fund the RDOF and E-ACAM locations, so those don’t add to the cumulative BEAD total (the yellow line and right axis). With a $400 million allocation in BEAD, Nebraska still runs out of money funding the remaining locations (in blue) before all the Unserved locations have a commitment. But importantly, E-ACAM keeps going and is providing service to locations at the upper end of the cost curve, including the Underserved as well. Nebraska ends up potentially able to reach 88,097 locations with broadband out of 108,601 Unserved and Underserved locations — 81% and way better than before.

The difference has the potential to be no less dramatic in South Dakota, where RDOF is a non-factor but E-ACAM could serve half the Unserved and Underserved, allowing the state a chance to bring fiber to all its Unserved and Underserved with their $207 million allocation.

Now is as good as time as any for caveats. First, all these Enhanced ACAM areas will get an offer from the FCC. We don’t yet know how many ISPs will accept it. Accepting the Enhanced ACAM offer would be a binding commitment to bring 100/20 or better service. It is possible, though I don’t think likely, that an ACAM ISP could accept the offer and serve the locations with unlicensed fixed wireless, which would mean it is still eligible under BEAD. I think the vast majority of ACAM ISPs are going to accept this offer and use a BEAD-compliant technology, so I’m comfortable using the offer data in this way.

Probably the biggest caveat is that my model assumes a best case scenario on the level of competition that will be possible in the grant process in the states. Specifically, it assumes that the state will be able to fund every location at its underlying cost, and that no ISPs will get more grant money from the state than is implied by the cost model. This should be a goal in every state, but it’s not likely to be achieved. If a single ISP bids to serve a location, states will be in a bad bargaining position to drive down costs. It’s one thing if there is 1 ISP bid to serve a remote family. It’s another if the state is essentially funding line extensions in an incumbent’s territory, as is the case here:

There are certain limitations of using a reverse engineered cost model. Data issues such as the below, where an incumbent claims to cover most of the town except a dozen locations are almost certainly filing errors. My cost model will assign a cost to these locations as if they are rural locations. Hopefully, the state is able to pay $0 to fund these locations, though as of this writing, they are still there, and still need “service” according to the official maps. There might be hundreds of thousands of these nationally.

I also believe the model underestimates costs in some cases. The highest cost location in my model is about $36,000. There may be a small number of locations that are more expensive to reach than that. To compensate, in this model I haven’t given states the “benefit” of an Extremely High Cost Location Threshold that, in reality, gives them an “out” for the most expensive locations. Instead, this model assumes states fund the locations through the whole cost curve.

Another factor is built-in assumptions about how much capital the ISPs bring, versus what needs to be provided by the BEAD program. I assume that for locations under $5,000 per location, ISP’s will bring 75% of the capital. For locations above $10,000 per location, ISPs will bring 25% of the capital (the statutory minimum). And for location between $5,000 and $10,000, ISPs will bring 50% of the capital.

We know private capital is capable of bringing most of the capital for low-cost locations because that’s exactly what happened in RDOF. Private capital brought more than 90% of the necessary funding in 20% of the locations that were won by providers planning to offer gigabit service. As a reminder, these location were Unserved and relatively high cost. If private capital can bring most of the funding during RDOF, we can hope to bring the same level of private capital to state-administered grant programs.

I’m not making a prediction here about the number of states that will be able to reach all their Unserved and Underserved. Funding locations at exactly the cost curve is a level of competition and cost control that is simply unrealistic. However, this model does emphasize that certain states will have a lot of trouble reaching all their Unserved and Underserved, and that it behooves every state to incorporate ways of generating competition in their grant program.