One Million Locations Might Remain Unserved Post-BEAD. Leftover Funds Should be Used to Connect Them.

funding
BEAD
Based on an analysis of all available data, the ACLP estimates that upwards of 1 million locations across the U.S. could remain unserved post-BEAD. With more than $20 billion in leftover BEAD funding likely, the ACLP proposes that some amount of those funds should be deposited into a BEAD Reserve Fund and used to connect any remaining unconnected.
Author

Alex Karras, Michael Santorelli

Published

October 23, 2025

Key Takeaways

  • The ACLP has compiled BEAD Final Proposal data from 49 states, D.C., and 3 territories encompassing $18.3B in proposed allocations across 3.6 million locations.
  • For a variety of reasons, it is likely that some number of locations will remain unserved post-BEAD. These reasons include:
    • Defaults on BEAD grants are likely, especially given the high cost to reach some locations, the large number of small and untested ISPs receiving funds, and the use of unproven technologies in many project areas.
    • At least 196,603 locations have been excluded due to “enforceable commitments” from other programs, which could be subject to default.
    • Some 1,063,812 locations appear unserved based on FCC data but are not covered by BEAD or commitments from other federal funding programs for which data is available.
  • With more than $20B in BEAD funding likely to be left over, the ACLP proposes that some amount of those funds – maybe as much as half – should be deposited into a BEAD Reserve Fund and used to bring service to any remaining unserved locations in a BEAD Round 2.

Final Proposal Data

The ACLP has compiled data from the BEAD Final Proposals (FPs) released by 49 states, D.C., and 3 territories (FPs have not yet been released by California, Puerto Rico, or U.S. Virgin Islands). Basic statistics on locations and funding awards can be viewed by hovering over a specific state in Figure 1, which is shaded by average cost per location.

Figure 1: States by average cost per funded BEAD location

As an overview:

  • The data encompasses $18,337,893,848 in BEAD funding allocated across 3,575,361 Broadband Serviceable Locations (BSLs).
  • Fiber (65% of locations) and low-earth orbit satellite (20%) top the list of technologies receiving awards by number of locations.
  • Some 896,168 BSLs deemed eligible for BEAD funding as of states’ post-challenge lists have been denied BEAD funding, largely due to already being served (46%), having an enforceable commitment (22%), or updates to the FCC’s maps (22%).

Potential For Locations to Remain Unserved

Despite its aim of universal service – or as close to universal service as possible – it appears unavoidable that some locations will remain unserved after BEAD. In addition to BEAD grant defaults, which are likely (and discussed separately below), other potential reasons why locations might remain unserved post-BEAD include:

  • “New” unserved BSLs not included in BEAD lists, both from new construction and corrections to FCC maps.
  • Defaults on grants made during previous programs. At least 196,603 locations were left out of BEAD after the challenge round due to enforceable commitments. These commitments can theoretically be broken, leaving locations unserved. Indeed, many of these commitments are from pandemic-era state grant programs that doled out dollars from ARPA’s Capital Projects Fund (45%+) and State & Local Fiscal Recovery Fund (15%+).1 An unknown number of these locations were also excluded during the challenge process due to such commitments. This figure is harder to ascertain because, unlike the Benefit of the Bargain (BOTB) round, states were not subject to standardized public data reporting requirements during the challenge phase.
  • Locations otherwise left off BEAD lists that are truly unserved. As of the latest FCC data, at least 1,063,812 locations across 49 states appear to be unserved and were not included from BEAD. These locations did not receive BEAD funding, were not marked as intentionally excluded from BEAD during BOTB, and were not included in other recent funding programs as indicated in FCC data. Many of these may indeed be correctly excluded due to challenges, map errors, or other enforceable commitments not covered by currently available FCC data. However, it is likely that at least some were erroneously excluded, the exact number of which will only be clear after BEAD projects are complete.

Combining data from FPs with data from the FCC’s National Broadband Map and Broadband Funding Map provides an illustrative look at this dynamic, shown below in Figure 2.

Figure 2: Flow of served & unserved locations

BEAD Defaults

As with prior federal broadband grant programs, some degree of defaults will likely occur with BEAD. With upwards of 3.6 million funded locations to date, even a small proportion of defaults would result in a sizable number of locations remaining unserved.

Technology may play a role in where defaults occur. Fiber, which received 65% of grants by number of locations, is relatively costly to deploy and could be vulnerable to cost overruns and other issues, especially for small or inexperienced providers. Low-earth orbit (LEO) satellite, which received 20% of grants by number of locations, is still untested at scale, with one large awardee (Amazon Kuiper) yet to provide residential connectivity anywhere in the country.

Defaults in prior federal broadband programs may provide a hint regarding the potential magnitude of defaults:

  • RDOF’s legacy is tainted by a high number of defaults (37%) and clawing back of funding from poorly vetted awardees. Per an analysis of FCC data by the Benton Institute, about 1.9 million RDOF locations were defaulted on, representing about 37% of total awarded locations. Thankfully, according to ACLP analysis of recent FCC location data, only about 300,000 locations in RDOF-defaulted Census Blocks remain unserved, meaning that, in the interim, private ISPs have largely filled the gaps.
  • CAF II, despite its poor design and execution, saw a low default rate of about 1.5%.2 This might have been due to low performance standards: many ISPs were able to easily provide 10/1 Mbps. Shortly after the program rolled out, the FCC raised its broadband speed threshold to 25/3 Mbps, effectively reclassifying many CAF II served locations as unserved.

BEAD appears to have been designed, at least in part, with the complicated legacies of RDOF and CAF II in mind. However, as has been detailed extensively elsewhere, including by the ACLP, BEAD was made unwieldy by the Biden administration’s layering on of a host of unnecessary rules and requirements of prospective subgrantees. The result was many years of planning and no allocations of grant funding.

The Trump administration’s BOTB round brought BEAD more in line with RDOF in terms of its focus on “lowest bid wins.” RDOF was a reverse auction; BOTB has largely operated as one. However, the possibility exists that some BEAD applicants were poorly vetted, a problem that forced the FCC to cancel significant RDOF awards to two of its biggest winners, leaving many thousands of rural households without broadband access. The ACLP highlighted the importance of robust vetting of prospective BEAD subgrantees to NTIA in a March 2024 report. The long tail of preliminary BEAD winners across the states that have released their FPs appears to include many small ISPs. Some of these might not be able to follow through on the promises made in their BEAD applications.

In addition, as in RDOF, other factors, like higher-than-expected costs, bankruptcies, and other unforeseen challenges, could force ISPs of any size to give up their BEAD award.

As such, it is likely that there will be defaults after BEAD awards are finalized. Even at a relatively conservative estimate of 10% default (by number of locations), over 300,000 locations would go unserved.

Proactive Solutions to Finish the Job

Regardless of assumptions, it seems certain that some locations will remain unserved after BEAD concludes. Whether BEAD is deemed successful or not will hinge on the number of these locations. If there are 1 million unserved–or even 100,000–post-BEAD, then BEAD will not have achieved the mission that everyone understood it to have: pushing broadband to as many people as possible and finally achieving universal or near-universal availability.

If this occurs, the country will once again return to the cycle of subsidy programs failing to achieve their goals, thus necessitating another grant program to make up for the shortcomings of the ones that came before.

A better approach would be to hold a second round of BEAD.

The ACLP proposes that NTIA use some amount of the $20B+ in leftover BEAD funds—perhaps as much as half—to seed a BEAD Reserve Fund. BEAD Round 2 could be run one year after the final FP is approved by NTIA, or sooner if it becomes clear that a sizeable number of unserved locations remain and/or if defaults start stacking up. It will also give NTIA, the FCC, and the states a chance to update the relevant maps plotting out served and unserved locations. Indeed, there is a chance that a BEAD Round 2 could be guided by the most accurate broadband maps ever if all relevant data from all relevant stakeholders is rationalized.

Funds could be apportioned to the states per the same formula used for the initial BEAD allocations. BEAD Round 2 could follow the BOTB framework, with a focus on “lowest cost wins” and carveouts for higher awards if some locations do not receive a bid. Any remaining funds could be returned to the Treasury, repurposed for other uses (e.g., supporting demand-side programs), or left in the Reserve Fund as a kind of rainy-day insurance policy in case additional defaults arise.

Ending BEAD with leftover funds and hundreds of thousands of unserved locations would mean NTIA did not accomplish its mission to meet the objective of Congress when it created and funded the program. Fortunately, it is not too late for NTIA to begin making plans to finish the job and make sure every American has access to a robust broadband connection.

(The ACLP discusses ideas for how NTIA should use the other $10B or so in leftover funds on discrete supply- and demand-side issues in a companion piece available here.)

Data Methodology

Final Proposals

Our aggregated Final Proposal datasets can be downloaded here: BEAD Final Proposal Data.

Data is sourced from public comment draft and submitted-to-NTIA version of states’ Final Proposals. When submitted versions were available, these were used instead of draft versions. Data was available for the following states:

  • Public Comment Draft: AK, AR, AS, AZ, CT, DC, DE, FL, GU, IA, ID, IL, IN, KS, KY, LA, MA, MD, ME, MO, MP, NC, ND, NE, NJ, NM, NV, NY, RI, SC, SD, TN, TX, UT, WA, WI, WV, WY
  • Submitted to NTIA: AL, CO, GA, HI, MI, MN, MS, MT, NH, OH, OK, OR, PA, VA, VT

As part of the final proposal, states were required to provide a standardized set of CSV data files. For each state, four of these were used in our analysis:

  • fp_subgrantees.csv: Listing out all subgrantees (i.e., ISPs) who are to receive funding.
  • fp_deployment_projects.csv: Listing out all projects which will receive funding.
  • fp_locations.csv: Listing out all BSLs which are included in a funded project.
  • fp_no_bead_locations: Listing out all BSLs from a state’s post-challenge eligible location list which will be intentionally excluded from funding, and why.

These files were archived, cleaned as necessary to bring into compliance with the standardized format, and aggregated to provide the figures above.

While NTIA provided clear, easy-to-follow guidance on how to format these standardized data files, states’ adherance to this guidance varied widely. While some states made proper, compliant CSVs available, some did not. A full gamut of data issues, errors, and non-compliance were present, including omitted columns, illegal values, non-ASCII text, and many others. In some cases, these included omission of or errors in critical, uniquely identifying variables like location_id, project_id, and uei, which made proper merging of data files difficult. Hopefully, these issues will be corrected during the submission and review process. Indeed, we observed that some errors in draft versions were corrected in subsequent submitted versions.

Due to these errors, several judgement calls were necessary during the process of cleaning the data. When possible, we attempted to err on the side of inclusion. For example, if a given funded location_id appeared to be duplicative across projects, was not present in CostQuest’s Fabric, or was missing/improperly entered, we still counted the row in the data as a funded location for that project.

National Broadband Map Data

Utilized in this analysis are lists of “served” and “unserved” locations from the FCC’s National Broadband Map. These were computed using data reflecting availability as of December 30, 2024, the latest available. Served and unserved locations were computed as those with and without, respectively, at least one such connection available:

  • max_advertised_download_speed >= 100
  • max_advertised_upload_speed >= 20
  • technology IN ('10', '40', '50', '70', '71', '72') (copper, cable, fiber, all fixed wireless)
  • low_latency IS TRUE

National Broadband Funding Map Data

Data from the FCC’s Broadband Funding Map available, generally, at the location (BSL) level or with GIS shapefiles representing served areas. Data was incorporated for the following programs:

  • Location: ARC, FCC_Tribal_NOFO2, Treasury_SLFRF, NTIA_BIP, FCC_Tribal_NOFO1, Treasury_CPF
  • Shapefile: FCC_Tribal_NOFO2, NTIA_BIP, FCC_EACACM, RUS_CommConn, FCC_RDOF, FCC_ConnectUSVI, FCC_PR, FCC_CAF2, RUS_Rural_EConn, FCC_Tribal_NOFO1, RUS_TelephoneLoan

For location data, data was matched using location_id. For shapefile programs, the provided shapefiles were overlaid over CostQuest’s Fabric and any locations within were assumed to be served by that program.

Footnotes

  1. Some states listed their ARPA and/or CPF programs using state-specific names which may not have been captured in our filtering, meaining these figures may underestimate the total share from ARPA and/or CPF.↩︎

  2. Total HHs to be served (per announcement of awards, August 2018): 713,176 | Pre-authorization defaults (as of June 2022): 4,682 | Post-authorization defaults (as of April 2024): 6,336↩︎