Connect with us

Broadband Mapping & Data

Mike Conlow: There Really Is Enough Money to Reach Most of the Unserved and Underserved

Estimating how far the money will go does not require advanced math.



The author of this Expert Opinion is Mike Conlow, a longtime broadband data and Internet policy analyst.

Last week a wireless company published a study about the cost to reach all the Unserved and Underserved. They estimated we need $307 billion to reach everyone. Similar reports have come out before. Generally they make a case that BEAD will run out of money, so the federal government should settle — for wireless technology, or FCC subsidy programs outside the scope of BEAD.

Like previous versions, this one too doesn’t represent the real situation. We have almost enough money to reach all the Unserved and Underserved, except for states that are particularly high cost.

Estimating how far the money will go does not require advanced math. We need an average cost to reach a location, and the total number of locations. There are 12.8 million Unserved and Underserved locations nationally. Instead of using the FCC data, the wireless study estimates this number from Census data, then adds 10% assuming growth from challenges to arrive at 16 million locations. That’s one difference.

The study breaks down these locations by the density of their Census tract. That’s a reasonable approach. Density is the primary driver of cost to serve a location, and is a decent proxy for other important things like terrain. When we break down the 12.8 million location Digital Divide by density, the peak is between 20-50 housing units per square mile. There are 2.7 million locations in that density range. There are about 1.5 million locations each in the least dense bucket 0-5 housing units/sq mile and the bucket between 5-10 hu/sq mi.

Now turning to the price per location at each of these intervals. I use the reserve prices from the RDOF reverse auction to estimate cost per location served by fiber. The reserve prices are not the winning bid from the RDOF auction. Instead, they are the price below which the FCC was willing to fund a project. If you add up all the RDOF reserve prices, it was $26.5 billion for 5.3 million eligible locations, or $5,000 per location. These prices are very clearly derived from the FCC’s cost model (built by CostQuest) and may be the cost model itself. Regardless, they certainly order locations correctly — more rural locations have a higher reserve price than urban locations.

As I’ve done previously, here I’m inflating the RDOF reserve price by 25% to capture changes since the RDOF auction like inflation, supply chain issues, and other things. When you plot these prices against the density of the Census block groups they cover, you see the expected hockey stick-like increase in cost. At less than 5 housing units per sq. mile the cost per location is $13,243 per location. It drops quickly: between 5-10 hu/sq mi the cost is $9,154, and when it reaches the mode at 20-50 hu/sq mi the cost is $6,015. In the most dense areas, the cost per location is $1,150. The overall average is $5,266 per location.

(As an aside, there are 373,000 BDC Unserved and Underserved locations in Census block groups with zero density. How does that happen? The Census counts housing units, not businesses. A Census block that is entirely businesses has 0 housing units and thus 0 density according to this metric using Census data. )

If you add up all these costs — the number of locations in each group times its average cost — you get $64.5 billion to reach all of them with fiber. That number includes the private capital match of at least 25%. The tab to BEAD alone is $50.6 billion. Remember also that this rough estimate doesn’t factor in the significant RDOF commitment to fund locations, nor ARPA, nor some locations which will be Extremely High Cost and therefor they won’t get fiber. Even if you inflate the RDOF cost by 50% it only brings the total to $81 billion, without any deductions for RDOF, private capital match, etc. I see no way to get $307 billion as this study did.

Below is the summary of how the wireless study reaches the conclusion about needing $307 billion in investment. The distribution of locations by density is not all that different. They have 6 million locations below 20 housing units/sq mile. I have 4.9 million. However, according to the BDC data there are over 1 million locations in blocks with density above 750 hu/sq mi which have a very low cost and is not reflected in this study.

Probably the biggest difference is the cost per location applied to that lowest density group. They appear to give a cost of $18,100 for every location under 20 hu/sq mi. I don’t disagree that the cost curve bends upwards for the lowest density areas, but I think the bend starts happening at 10 housing units / sq mile or even lower. There probably are locations that cost $18,000 to reach them, but it isn’t fair to apply that average across every location below 20 hu/sq mile. I can’t replicate how they get $191 billion to serve that lowest density group. Six million locations times $18,100 per location is $108.6 billion, not $191. That’s a big difference. Maybe there’s some methodological aspect I’m not following.

About the same time that this study was published, Cost Quest put up a blog post about cost curves. Using a state which they don’t name, they provide a cost curve chart by percentile for Unserved locations. The average cost per location in this state is $5,589, in line with the estimates I use for average cost per location nationally. Their cost curve starts to increase significantly for the last 10% most expensive locations in the state. It really hockey sticks for the hardest to reach 5% of locations.

No alt text provided for this image

If we apply this logic to the national distribution of Unserved and Underserved locations by density, it means costs start to rise for least dense 10% of locations — 1.3 million locations. And really hockey sticks for the last 650,000 locations. While it is interesting to consider these locations, no state is going to try to build fiber to them. NTIA created the Extremely High Cost Location threshold for exactly this purpose. According to NTIA, the EHCL threshold is a price above which a state “may decline to select a proposal if the use of an alternative technology meeting the BEAD Program’s technical requirements would be less expensive.” How high the cost curve reaches doesn’t matter. They’ll be EHCLs and won’t get fiber.

Implicit in my contention that there’s [almost] enough money to reach all the Underserved and Unserved nationally is that BEAD grants are awarded competitively and close to the true cost to serve the location. Consider the example Cost Quest state above. They have a statewide average of $5,589 per location. But about 62% of locations are below that average. Some of these locations are on the “network edge” — an ISP provides service on the block, or next door. Some of them are dense towns served by DSL now. It is critical that this state generates competition for these areas and finds a proposal close to the true cost to serve the location. If they fund all these locations at the statewide average of $5,589 they could waste $2,000 or more per location.

We know private capital is willing to bring most of the capital for these 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, let’s find a way to bring the same level of private capital to state-administered grant programs.

There’s an important practical takeaway from all this: a public cost curve for every state is critical. States need it to set a reasonable Extremely High Cost Location threshold. But more importantly, states and the public need it to evaluate grant proposals. If states don’t have the cost curve data and the ability to run a grant program that optimizes on it, BEAD will run out of funding for sure, and it doesn’t matter whose numbers we use for estimation now.


  • NTIA should procure a national cost model that represents the unique contours of every state.
  • NTIA should give that cost model to every state, and make it public.
  • NTIA should provide guidance to states on how to set the Extremely High Cost Location threshold based on their individual cost curve.
  • NTIA should provide guidance, and only accept plans from states, that encourage competition between ISPs, and that fund locations based on the underlying costs to reach those locations.

Mike Conlow is a longtime broadband data and Internet policy analyst. Prior to being involved in Internet policy he was in political technology, and was on both of President Obama’s campaigns, in 2012 as the Deputy CTO. This piece was originally published on Mike Conlow’s substack on April 25, 2023, and is reprinted with permission.

Broadband Breakfast accepts commentary from informed observers of the broadband scene. Please send pieces to The views expressed in Expert Opinion pieces do not necessarily reflect the views of Broadband Breakfast and Breakfast Media LLC.

Broadband Breakfast is a decade-old news organization based in Washington that is building a community of interest around broadband policy and internet technology, with a particular focus on better broadband infrastructure, the politics of privacy and the regulation of social media. Learn more about Broadband Breakfast.

Broadband Mapping & Data

Tribes Must Be Ready to Challenge State Broadband Maps: Tribal Ready

Tribes needs to be prepared to approach states on what coverage data is not included in state maps.



Photo of Lori Adams of Nokia, Joe Valandra of Tribal Ready, Megan Beresford of Learn Design Apply, E.J. John of the American Indian Policy Institute (left to right)

WASHINGTON, May 31, 2023 – Tribal governments should gather broadband coverage data for the state mapping process, said Joe Valandra, CEO of newly formed Native American-owned data company Tribal Ready at a Broadband Breakfast Live Online event Wednesday. 

Historically, tribal data has been excluded or misinterpreted in broadband maps, he said. The $42.5 billion Broadband Equity Access and Deployment program will be allocated to subgrantees by state governments according to state broadband maps. 

Tribal governments need to be prepared to approach the state with a data-driven argument about what coverage data is not included in the state map and what changes need to be made, said Valandra. 

In turn, state broadband offices need to listen to tribes, added Megan Beresford, director of broadband programs at grant writing company Learn Design Apply.  

The $3-billion Tribal Connectivity Program of the National Telecommunications and Information Administration received over $5 billion in grant requests from its application process last year. BEAD allocations, expected to be announced by the end of June, can play a part in addressing the undersubscription of funds to tribal programs, said E.J. John, senior research analyst at the American Indian Policy Institute. 

Other federal programs can also support tribal connectivity, said Beresford. The Affordable Connectivity Program allows eligible low-income households to get a discount on broadband of up to $75 per month on tribal lands. 

The NTIA announced in May nine new Tribal Broadband Connectivity Program grants of $500,000 each, bringing the program’s total amount disbursed to $1.77 billion.  

Our Broadband Breakfast Live Online events take place on Wednesday at 12 Noon ET. Watch the event on Broadband Breakfast, or REGISTER HERE to join the conversation.

Wednesday, May 31, 2023 – Tribal Broadband Deployment

As the NTIA continues to issue awards from the first round of the $3 billion Tribal Broadband Connectivity Program, how are the funded projects progressing? How will they interact with the other ongoing broadband initiatives, such as the Middle Mile and Broadband Equity, Access and Deployment Programs?


  • Lori Adams, Vice President of Broadband Policy & Funding Strategy, Nokia
  • Joe Valandra, CEO and President, Tribal Ready
  • Megan Beresford, Director of Broadband Programs, Learn Design Apply
  • E.J. John, Senior Research Analyst, American Indian Policy Institute
  • Drew Clark (moderator), Editor and Publisher, Broadband Breakfast

Panelist resources

As senior director of broadband policy and funding strategy, Lori Adams is a key member of the Nokia Government Affairs Americas Team. She is responsible for developing strategies and tools to enable increased company participation in state, federal, and international programs supporting infrastructure deployment by several of Nokia’s business organizations. Additionally, she focuses on external government relations and communications with stakeholders at all levels of government through direct engagement, filings, and participation in public forums.

Before leading Tribal Ready, Joe Valandra served as the executive director of the Native American Contractors Association (NACA). He also served as the managing director of VAdvisors, LLC, a specialty advisory firm in Washington, DC, and as the chief of staff for the National Indian Gaming Commission (NIGC), a federal regulatory agency with Indian gaming oversight responsibilities. Joe has served in senior executive roles in private and public sectors, including as a board member of numerous companies in multiple industries.

Megan Beresford is the director of broadband programs at Learn Design Apply Inc (LDA). She joined the company mid-pandemic as the digital divide became glaringly evident. Since then, she and her team have helped states, public entities, tribes, and private internet service providers secure over $300 million in broadband infrastructure and digital equity funding.

E.J. John is the senior research analyst at the American Indian Policy Institute at Arizona State University. He is a member of the Navajo Nation who uses his experience working in Tribal government and policy research to promote digital equity for Tribal communities.

Drew Clark (moderator) is CEO of Breakfast Media LLC. He has led the Broadband Breakfast community since 2008. An early proponent of better broadband, better lives, he initially founded the Broadband Census crowdsourcing campaign for broadband data. As Editor and Publisher, Clark presides over the leading media company advocating for higher-capacity internet everywhere through topical, timely and intelligent coverage. Clark also served as head of the Partnership for a Connected Illinois, a state broadband initiative.

Painting by Paul Cézanne used with permission

WATCH HERE, or on YouTubeTwitter and Facebook.

As with all Broadband Breakfast Live Online events, the FREE webcasts will take place at 12 Noon ET on Wednesday.

SUBSCRIBE to the Broadband Breakfast YouTube channel. That way, you will be notified when events go live. Watch on YouTubeTwitter and Facebook.

See a complete list of upcoming and past Broadband Breakfast Live Online events.

Continue Reading

Broadband Mapping & Data

South Carolina’s Innovative Broadband Maps Verifies ISPs’ Internet Speeds

South Carolina performs mapping audits to hold ISPs accountable for coverage claims.



Photo of Jim Strizinger

WASHINGTON, May 21, 2023 – South Carolina’s innovative state broadband map can accurately identify areas of over-reporting by internet service providers, the director of the state’s broadband office said in a Friday Ask Me Anything! session in the broadband community.

South Carolina processes the same data as does the Federal Communications Commission as it creates its broadband map. However, it also performs audits on the ISPs to ensure they are submitting accurate data. Hence, the state can determine errors in reporting data based on where the ISP’s networks had been deployed previously and where state investments have gone, said Jim Stritzinger, director of the state’s broadband office.

Providers are required to file amended returns with the FCC in the event that South Carolina’s state broadband office flags errors in their reporting information. Errors include misreporting of technology types.

If the reporting errors are not corrected, the state will report the defaulting ISP to the FCC, said Stritzinger, a software engineer with a passion for mapping broadband in the Palmetto state.

A big flaw of the FCC’s maps is that ISPs were able to report advertised speeds, which Stritzinger said were useless.

To enhance the accuracy and reliability of the maps, Stritzinger partnered with broadband data collection company Ookla, and integrated speed test data directly into the mapping system. More than 12 million Ookla speed tests have now been incorporated into the map, with some census blocks containing over 15,000 tests.

In 2021, South Carolina made the decision to no longer accept Digital Subscriber Lines as reliable service anywhere in the state. Doing so opened large regions of the state to investments, said Stritzinger, and will reduce the number of underserved locations.

The state’s next iteration of its map is set to come out sometime before June 30, and will be the state’s first address-level broadband map.

Stritzinger estimated that investments from the Broadband Equity Access and Deployment program will be deployed in 2025. In the meantime, the state will continue working to deploy the American Rescue Plan Act dollars, which allocated $25 billion in several broadband projects, $8 billion of which will go to states and local governments directly.

Continue Reading

Broadband Mapping & Data

In Ask Me Anything!, Jim Stegeman of CostQuest Says Broadband Fabric Will Improve

Fabric data will continue to improve with feedback implementation and process changes, said CostQuest CEO.



Photo of Jim Stegeman, CEO of CostQuest

WASHINGTON, May 12, 2023 – There are unrealistic expectations for the broadband mapping process despite its continual improvement, said Jim Stegeman, president and CEO of CostQuest, the mapping company that was hired to deliver fabric data to the Federal Communications Commission, during an Ask Me Anything! event in the broadband community on Friday. 

“We will never get to allocation if we are after perfection,” said Stegeman, claiming that the company has been working hard to improve its processes over time.  

The broadband fabric data is a dataset that maps all locations at which “fixed broadband internet access service has been or could be installed.” The FCC populates its National Broadband Map with the fabric data. The National Telecommunications and Information Administration has committed to allocating federal broadband funding by June 30 based on a version of the map. 

Since its initial roll-out and subsequent challenge process – in which providers and state broadband offices have been able to challenge coverage claims by submitting contrary evidence – many have complained about the map’s many inaccuracies. 

CostQuest maintains that the fabric identifies 99.3 percent of all broadband serviceable locations correctly and is “very good” where it stands. 

Fabric will be continual improved 

Stegeman outlined in his remarks areas in which the fabric and its processes will continue to improve in the coming iterations.  

Each version of the fabric implements feedback, he said. Version two incorporated over a million new BSLs, 178,000 of which came from the FCC’s challenge process and 860,000 from CostQuest’s internal efforts to improve the fabric.  

The FCC is currently in the process of releasing the next version of the National Broadband Map based on version two of the fabric data. This version of the map is expected to be the map which the NTIA allocates BEAD funding. 

Subsequent iterations of the map will support state deployment and challenge processes, said Stegeman. 

Version two also improved tribal land BSL identification by updating the logic that identifies whether a parcel of land holds a BSL and whether it contains multiple or single units. 

Furthermore, through contractual agreement with the FCC, CostQuest has retained a file – facetiously titled the detritus file – that stores information on where water towers, sheds, chicken coops, and other agricultural buildings are located.  

Precision agriculture, the process of using broadband to optimize agricultural production, requires broadband connection to these locations, often located miles away from homes and other BSLs. The FCC anticipates that this location data may be of future use, said Stegeman.

Continue Reading

Signup for Broadband Breakfast News

Broadband Breakfast Research Partner