Abstract
The Federal Communications Commission (FCC) reports that 19.6 million Americans lack access to a fixed 100/20 Mbps broadband service. A nationwide audit using 109,473 ISP‑address tests from October 2024 to March 2025 finds the true figure closer to 26.0 million, exposing a 6.4 million person (33%) under‑count.
Discrepancies are not wholly uniform: they concentrate in rural Plains, Mountain West, and fast‑growing Sunbelt states, where reliance on self‑reported provider filings and long‑range fixed‑wireless deployments skews official maps. Mis‑measurement at this scale may misallocate billions of dollars from the Broadband, Equity, Access, and Deployment (BEAD) program, undermine Rural Digital Opportunity Fund compliance enforcement, and weaken targeting of future affordability subsidies.
This research details the audit methodology, quantifies state‑level errors, examines statistical patterns, and proposes a replicable geospatial monitoring framework to close the “mapping gap.”
Why It Matters
High‑speed internet is now a prerequisite for economic participation, education, and healthcare. Congress and the Executive Branch have tied most contemporary broadband subsidies—including the $42.45 billion BEAD program—to the FCC’s Broadband Serviceable Location Fabric and associated availability maps.
Although recent map iterations are more granular than legacy Form 477 data, they remain provider‑reported and still count a location “served” if a single provider asserts it could deliver the speed tier. Stakeholders across academia, state broadband offices, and consumer groups (e.g., BroadbandNow) therefore question whether the maps reliably steer public‑funding decisions.
This research compares the FCC’s 2025 100/20 Mbps availability claims with an independent audit that blends provider plan crawls and address‑level purchase‑attempt verifications. We ask three questions:
- How large is the national under‑count of unserved Americans?
- Which states and demographic contexts show the largest mapping gaps?
- What policy and regulatory levers can correct these blind spots before BEAD dollars are disbursed?
Key Findings
- 33% under-count nationwide: The FCC says 19.6 million Americans lack 100/20 Mbps; our manual address-checker audit shows 26.0 million—an extra 6.4 million people—still offline.
- Technology-specific over-reporting: Provider filings overstate real-world availability most sharply for newer fiber networks: 66.5 percent of addresses we audited (that the FCC lists as fiber-served) still can’t order a 100/20 Mbps plan today. Legacy DSL exaggerations affect 48.7% of listed addresses, fixed-wireless 44.6%, and cable systems 14.5%.
- State-level extremes: Iowa and New Mexico miss more than 60% of their truly unserved residents; California’s absolute gap tops 446,000 people.
- $14 billion at stake: A 33% mapping error could mis-route roughly $14B in BEAD funds away from the communities that need them most.
| Scope | FCC estimate of Americans without 100/20 Mbps access | BroadbandNow crowd‑sourced estimate | Gap (extra Americans still unserved) | % FCC is under‑counting |
|---|---|---|---|---|
| Nation‑wide (2025) | 19.6 million | 26.0 million | 6.4 million | ≈ 33 % |
The national picture: a digital‑divide alarm bell
One-in-13 Americans that the FCC thinks have broadband still can’t get a 100/20 Mbps service.
The 6.4 million‑person shortfall is larger than the population of Missouri, and it widens the “unserved” pool by roughly one‑third. Methodology drives the wedge.
FCC maps: self‑reported availability at the location level (ISP Form 477 → Fabric), counts a location as “served” if one provider claims it could deliver 100/20 Mbps.
BroadbandNow audit: address‑level availability checks; a location is “served” only when at least one provider actively sells the tier.
The difference punishes edge‑case locations where providers file but never build, and where fixed‑wireless/mobile variants fall short in real‑world throughput.
Technology‑specific over‑reporting
| Access Technology | FCC “served” locations flagged unserviceable in audit | Over‑reporting rate |
|---|---|---|
| Fiber | 66.5% | Highest error; many “planned” builds not yet live |
| DSL | 48.7% | Vectoring/“up‑to” filings rarely meet 100/20 in practice |
| Fixed Wireless | 44.6% | Long‑range links struggle with upload, capacity limits |
| Cable | 14.5% | Still significant where legacy nodes aren’t yet split |
Take‑away: Even premium fiber claims are overstated two‑thirds of the time at audited addresses, undercutting narratives that fiber filings alone guarantee future‑proof access.
Which states are most mis‑measured?
Highest relative under‑counts (≥ top 20 % of states)
| Rank | State | FCC under‑count | Relative gap |
|---|---|---|---|
| 1 | Iowa | 195,000 | +62.6% |
| 2 | New Mexico | 161,000 | +61.0% |
| 3 | Mississippi | 264,000 | +55.2% |
| 4 | Arizona | 198,000 | +55.1% |
| 5 | South Dakota | 68,000 | +52.8% |
Sparsely populated Plains and Mountain West states plus parts of the rural South. Low‑density terrain, heavy reliance on long‑range fixed‑wireless, and a high fraction of RDOF/CAF deployments that have slipped schedules all correlate with larger gaps.
Largest absolute under‑counts
| Rank | State | FCC under-count |
|---|---|---|
| 1 | California | 446,000 |
| 2 | Texas | 444,000 |
| 3 | Florida | 392,000 |
| 4 | North Carolina | 272,000 |
| 5 | Alabama | 270,000 |
Big, fast‑growing Sunbelt states where suburban‑rural fringes outstrip last‑mile builds. Even modest percentage gaps translate into hundreds of thousands of people.
Policy & funding implications
- BEAD allocation risk. The IIJA’s $42.5 B formula weights FCC unserved counts; a 33 % under‑count nationally could redirect ≈ $14 B away from the hardest‑hit states unless NTIA’s challenge‑process data is adopted.
- Oversubscription vs. overbuild. States with > 50 % mapping error may be prioritizing areas that look “served” on FCC maps but test “unserved” on the ground. Challenge data should be embedded directly into the Initial Proposal Volume 2 updates.
- RDOF compliance auditing. Of the 6.4 million under‑counted Americans, ~2 million live in RDOF‑awarded census blocks, suggesting incomplete builds or technologies (e.g., 25/3 FWA) that fail to scale.
- Equitable affordability programs. ACP’s eventual replacement needs more than address eligibility; it needs performance verification so that “served” addresses actually see ≥ 100/20 Mbps when subsidies are used.
Overreporting By State
| State | Overreporting Percent | FCC’s Unserved Americans Estimate | BBN Unserved Americans Estimate | Population |
|---|---|---|---|---|
| Alabama | 38.50% | 701393 | 971517 | 5108468 |
| Arizona | 55.10% | 359677 | 557905 | 7431344 |
| Arkansas | 40.30% | 351255 | 492827 | 3067732 |
| California | 28.10% | 1589780 | 2036000 | 38965193 |
| Colorado | 37.30% | 270370 | 371327 | 5877610 |
| Connecticut | 24.60% | 90791 | 113123 | 3617176 |
| Delaware | 43.50% | 16201 | 23249 | 1031890 |
| Florida | 33.70% | 1164452 | 1556814 | 22610726 |
| Georgia | 35.00% | 687121 | 927741 | 11029227 |
| Idaho | 48.40% | 208457 | 309452 | 1964726 |
| Illinois | 25.40% | 466848 | 585588 | 12549689 |
| Indiana | 29.00% | 373990 | 482533 | 6862199 |
| Iowa | 62.60% | 146560 | 238329 | 3207004 |
| Kansas | 29.90% | 177315 | 230302 | 2940546 |
| Kentucky | 37.40% | 520960 | 716007 | 4526154 |
| Louisiana | 37.20% | 639867 | 878052 | 4573749 |
| Maine | 25.90% | 156321 | 196779 | 1395722 |
| Maryland | 10.00% | 120515 | 132566 | 6180253 |
| Massachusetts | 8.20% | 70014 | 75728 | 7001399 |
| Michigan | 27.90% | 764839 | 977971 | 10037261 |
| Minnesota | 43.60% | 302388 | 434118 | 5737915 |
| Mississippi | 55.20% | 478582 | 742534 | 2939690 |
| Missouri | 32.40% | 477104 | 631609 | 6196156 |
| Montana | 43.50% | 226449 | 325050 | 1132812 |
| Nebraska | 45.30% | 130969 | 190361 | 1978379 |
| Nevada | 45.60% | 25873 | 37676 | 3194176 |
| New Hampshire | 23.00% | 43464 | 53470 | 1402054 |
| New Jersey | 12.90% | 142150 | 160418 | 9290841 |
| New Mexico | 61.00% | 203402 | 327386 | 2114371 |
| New York | 22.30% | 587136 | 718302 | 19571216 |
| North Carolina | 27.90% | 974111 | 1245769 | 10835491 |
| North Dakota | 40.70% | 8623 | 12136 | 783926 |
| Ohio | 27.40% | 723656 | 922209 | 11785935 |
| Oklahoma | 35.30% | 338900 | 458545 | 4053824 |
| Oregon | 41.30% | 308188 | 435622 | 4233358 |
| Pennsylvania | 21.70% | 588460 | 716098 | 12961683 |
| Rhode Island | 28.90% | 29591 | 38137 | 1095962 |
| South Carolina | 34.80% | 310054 | 417805 | 5373555 |
| South Dakota | 52.80% | 49827 | 76145 | 919318 |
| Tennessee | 32.60% | 374141 | 496261 | 7126489 |
| Texas | 27.60% | 1607524 | 2051770 | 30503301 |
| Utah | 40.00% | 132608 | 185692 | 3417734 |
| Vermont | 13.60% | 87667 | 99629 | 647464 |
| Virginia | 17.10% | 630145 | 737802 | 8715698 |
| Washington | 41.30% | 604717 | 854742 | 7812880 |
| West Virginia | 41.80% | 414374 | 587776 | 1770071 |
| Wisconsin | 29.10% | 871275 | 1125199 | 5910955 |
| Wyoming | 33.10% | 63137 | 84018 | 584057 |
| Total | 19611241 | 26040089 |
Data & Methods
| Component | Source | Purpose |
|---|---|---|
| FCC availability baseline | Broadband Data Collection v2 (Jan 2022) | Official “served/unserved” counts per location |
| Manual address‑checker audit | 109,473 ISP‑address tests (October 2024–March 2025) | Confirms whether ≥ 100/20 Mbps plan can be ordered at each address via ISP checkout or API |
| Provider‑plan crawler | BroadbandNow availability engine | Catalogues offered speed tiers and pricing to tag technology (fiber, cable, DSL, FWA) |
Serviceability definition
An address is “served” in the audit only if at least one ISP’s live checker:
- returns the address as serviceable, and
- displays a purchase path for a plan advertising ≥ 100 Mbps download and ≥ 20 Mbps upload.
If every available plan falls below 100/20, or if the checker states “service not available,” the address is unserved, regardless of FCC claims. This differs from our previous audits of national data, which used the legacy definition of 25/3 Mbps for broadband service.
Key verification statistics
- Total ISP‑address combinations tested: 109,473
- Combinations where service was not available despite FCC “served” flag: 19.9 %
- Addresses with zero serviceable providers: 0 % (each had some connectivity, but many lacked 100/20 capability)
Counts aggregate to state and national tallies using 2024 ACS population denominators. Analyses use state‑level observations (n = 51).
