42.45 billion dollars for broadband, appropriated in 2021 — no deployment project funded through August 2025

What is happening?

In late 2021 Congress appropriated 42.45 billion dollars for the Broadband Equity, Access, and Deployment program, the largest federal investment in broadband infrastructure to that date. NTIA notified the 56 states and territories of their allocations on 2023-06-30 and stated on 2024-12-23 that the full 42.45 billion dollars had been obligated. As of August 2025, no eligible broadband deployment project had been funded from the BEAD money allocated to any state.

On 2025-06-06 NTIA issued a Restructuring Policy Notice that rescinded every final proposal approval already granted, removed seven non-statutory requirements including the low-cost service option, eliminated the fiber preference, and required all 56 to run a fresh subgrantee selection round. Bids came in far below the allocations. By 2026-02-09, 50 of the 56 final proposals had been approved and NTIA put the unspent remainder at about 21 billion dollars — roughly half the appropriation.

Whose problem is this?

The layer that resets the schedule and the layer that waits for a connection never deal with each other. NTIA approves states, states select subgrantees, and subgrantees connect households — a household has no counterparty at the level where the calendar is actually set.

Where does this problem end?

The boundary matters here because the money was never the binding constraint. It was appropriated in full, obligated in full, and then sat.

What is the state now, and what should it be?

Now

Should be

Needs a new measurementno national completion date and no numeric coverage target appears in any source opened for this dossier. The only dated obligation confirmed is per subgrantee and statutory: a subgrantee must deploy the network and begin providing service to customers no later than four years after receiving the subgrant, and the state may extend that deadline under certain circumstances. That clock starts at the subgrant, which is the very step the program spent four years reaching.

How big is it?

The affected population is counted in broadband serviceable locations, not in persons. An analysis of the approved final proposals dated 2026-02-09 counted 3,379,711 locations in the United States without 100/20 Mbps service, of which 2,278,355 are funded by BEAD and 1,101,356 are funded by nothing at all.

The lower bound used here is 2,278,355 — locations that BEAD has now promised on paper and has not yet connected. The upper bound is 3,379,711 — every location still without service, funded or not. The unit is broadband serviceable locations rather than persons, so the number of people affected is larger than either bound. No household size figure was confirmed in this round, so that conversion was not attempted.

The same analysis reports that 33.7 percent of BEAD locations already have terrestrial service at 100/20 Mbps. The list of BEAD eligible locations and the list of locations actually without service are therefore not the same set, and the gap runs in both directions.

Under what conditions does it arise?

Three conditions hold this state in place, and a shortage of money is not one of them.

1. The approval chain has four layers. Congress appropriates, NTIA writes the rules and approves each state twice, the state selects subgrantees, and the subgrantee builds. A requirement added at any layer resets the work of every layer below it. The 2025-06-06 notice did exactly that — it rescinded final proposal approvals that three states had already obtained by 2025-01-16. 2. Nothing in the program measures elapsed time. The reported indicators are proposals approved, dollars obligated and dollars saved. None of them is a date by which a household is connected, and the only statutory clock starts at the subgrant rather than at the appropriation. 3. Delay produces no visible failure. An unserved household stays unserved, which looks the same as it did before the program existed. There is no interruption, no closure and no queue anyone can see, so a year of delay never registers as an event.

Each layer of approval can be reopened by the layer above it, and reopening resets the schedule of every layer below.

What has been tried?

What was found?

Why is it still unsolved?

Coordination failure — the money, the rules, the selection and the shovel sit in four different places, and none of them is accountable for the time between them.

Congress put 42.45 billion dollars into a structure where NTIA must approve each of the 56 twice, each of the 56 must run its own competitive subgrant process, and each subgrantee then has a statutory four years of its own. Every one of those steps has an owner and a rule. No single actor in this chain owns the elapsed time between the appropriation and the first customer connected.

The 2025-06-06 notice shows what that costs. Three states had already cleared the last statutory step and were ready to sign agreements with the providers they had selected. The notice rescinded those approvals and sent all 56 back through subgrantee selection. The rebid was not pointless — bids came in roughly 21 billion dollars below the allocations, and Nevada reportedly covered every eligible location at lower cost — but the same change reset the calendar for every state, including the three that were finished.

The dispute over whether that change was lawful adds a layer rather than resolving one. On 2025-12-16 the Government Accountability Office issued an opinion that the notice was a rule subject to the Congressional Review Act, and reporting on 2026-01-05 described the finding as a violation of that act. That question sits above NTIA and below Congress, in the one place where no schedule exists at all.

What observation would mean it is solved?

Three candidates, and the trap in each.

a. The count of locations connected by BEAD funded networks rises toward the 2,278,355 that approved final proposals cover. The trap is that this is the paper target rather than the need — 1,101,356 locations without service are outside it from the start.

b. The count of locations in the United States without 100/20 Mbps service falls from 3,379,711 toward zero. The trap is attribution — other federal programs and private builds move the same number, and the BEAD share cannot be read off it.

c. All 56 sign award agreements and construction begins everywhere. The trap is that this is what the program has been measuring all along. Initial proposals and obligations each reached completion during 2024, and as of August 2025 no deployment project had been funded.

None of the three is sufficient alone because the program reports progress in the units of its own process. A completion measure has to be dated at the household, not at the filing.

What is it connected to?

Fills with researchplausible links run to the high cost programs of the Universal Service Fund, to the end of the Affordable Connectivity Program, to rural depopulation, and to the federal permitting timeline that the 2025-06-06 notice also addressed. Relation type and evidence grade were not researched in this round.

What these sources do not say

• How much has actually been built. No source opened here reports a national count of locations connected by a BEAD funded network as of 2026-08. Reported milestones stop at proposal approval and award signature, which are filings rather than connections.
• What the roughly 21 billion dollars will be spent on. NTIA held a listening session on 2026-02-11; the sources here record no published rule for non-deployment use afterwards.
• Whether the rebid produced savings or scope reduction. Ohio saw its approved location count rise 13 percent from its draft and Indiana saw its fall 7 percent, and no source opened here reconciles the national location count before and after the 2025-06-06 notice.
• What the technology shift means at the other end. About 21 percent of BEAD eligible locations are slated for low earth orbit satellite in the draft plans, and none of these sources reports a performance floor, a price commitment or a capacity ceiling for those locations.
• Whether the Congressional Review Act finding changed anything in practice. The legal implication was reported: the restructuring cannot take effect until Congress is notified and given time to review. None of the sources here reports whether that notice was ever submitted, whether Congress acted on it, or whether any approval already granted was reopened because of it.
• The distribution of the delay. Only national totals are reported. No source opened here breaks elapsed time down by state, so whether some states are years ahead of others is unknown.

See the evidence

The Congressional Research Service report was read directly as a PDF, and it is the source of every pre-2025-09 date and dollar figure above. Nothing published by NTIA was read at first hand — the NTIA hosts could not be opened, so the agency statements above reach this dossier through the CRS report and the trade press that quote them. The underspend figure of roughly 21 billion dollars was confirmed twice from independently opened documents dated 2025-12-02 and 2026-02-09. One commercial funding tracker was opened and then discarded: it stated that 39 states remained in the challenge process, which contradicts the approval counts in every other source here, and none of its figures were used.

People affected

Estimated range 2,278,355–3,379,711 As of 2026-02

Sensitivity The interval is not a confidence interval. It is the distance between two definitional boundaries measured in the same document on the same date: what BEAD promised and has not built (2,278,355) against everyone still without service (3,379,711). The unit is broadband serviceable locations, not persons. No household or occupancy figure was confirmed in this round, so the conversion to people was not attempted, and the number of people affected is larger than the upper bound. The same source reports 1,342,667 unfunded units against 1,101,356 unfunded locations, so a unit count runs about 22 percent above a location count and the funded slice presumably carries a similar ratio. A correction runs the other way as well: the source reports that 33.7 percent of BEAD locations already have terrestrial service at 100/20 Mbps, which means the BEAD eligible list and the list of locations actually without service are not the same set. Neither adjustment was applied because the source does not give the cross-tabulation needed to apply it.

Regional breakdown No document opened in this round gives state level location counts and the national total together. The Louisiana figure of roughly 124,000 locations and the Delaware figure of roughly 4,700 locations come from a legal trade analysis, while the national totals come from a separate analysis of final proposal data, so summing them against the national figure would be mixing bases. Splitting the national count by state population would be proportional allocation, and the distribution of locations without broadband does not follow population.