Nationwide NG911 carries a remaining bill of USD 5.8 billion to 9.3 billion and no party is assigned to pay it

What is happening?

The National Telecommunications and Information Administration published its Next Generation 9-1-1 Cost Study on 2026-04-29. It puts the remaining cost of finishing the nationwide move from legacy circuit-switched 911 to IP-based NG911 at USD 5,802 million to USD 9,277 million across a seven-year window labelled 2026 through 2032, as of April 2026. The study names no completion date, no responsible party and no federal share.

The three published figures are not an uncertainty band around one scenario. They differ only in how many years of post-implementation recurring cost are counted — USD 5,802 million for two years, USD 7,583 million for three, USD 9,277 million for four, all as of April 2026. The USD 3,475 million distance between the low and the high figure comes from that counting choice rather than from any difference in the scope of work.

What both federal bodies describe as missing is a capability, not only money. NTIA wrote on 2026-04-29 that even the most mature state systems often lack geospatial routing, the ability to transfer calls with accompanying data across jurisdictions, and real-time sharing of incident information with first responders. The Federal Communications Commission wrote in the Second Report and Order it adopted on 2026-06-25 that interoperability sufficient to permit policy routing and seamless transfer of NG911 traffic across jurisdictional boundaries is still the exception rather than the norm, and that cross-jurisdictional dispatch capability is rarer still.

The federal transition rules that became enforceable on 2025-03-25 start a carrier obligation only when a local 911 authority files a valid request. As of June 2026 those authorities had issued more than 190 requests for Phase 1 service, which delivers 911 traffic in IP format, and one request for Phase 2 service, which delivers it in the standards-compliant form carrying the location object. The requests together cover parts or all of twenty-eight states and more than 2,200 public safety answering points.

Whose problem is this?

No party on this list can compel another. The regulator writes rules for carriers, the money is raised one state at a time, and the capability that the whole program exists for sits on a boundary between two neighbors who each answer to a different budget.

Where does this problem end?

The boundary matters here because the money already flows and the rules already exist. What is missing is a party with authority over both of those at the same time.

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

Now

Needs a new measurementthe target state: no source opened here names a date by which the nationwide transition is meant to be complete, a party responsible for the remaining USD 5,802 million to USD 9,277 million, or a federal share of it. NTIA states that decisions about the appropriate federal role are outside the scope of the study, and that cost elements can be attributed across federal, state, local and private-sector entities depending on the nature of the cost, without publishing any split. The study does carry modelling timelines of five, three and two years by state maturity, but a modelling assumption is not a target.

How big is it?

Between 300,000,000 and 334,914,895 people in the United States. The upper bound is the resident population figure of 334,914,895 for 2023-07-01 that the FCC used in its 2024 NG911 order. The lower bound is an over-allowance for the jurisdictions that both agencies concede are exceptions, and it is not measurable from any source opened here.

The band is narrow because the honest answer is close to everyone. NTIA in April 2026 and the FCC in June 2026 both describe complete end-to-end NG911 as an exception rather than as a share, and neither publishes a count of the jurisdictions that have it. So the number is bounded above by a census figure and below by an adjective. This is not a shortfall concentrated in one disadvantaged group; it is a national system that has not been finished anywhere end to end.

This number does not claim that anyone who dials 911 fails to reach help. Most calls complete on legacy infrastructure. The failure is conditional and appears at jurisdiction boundaries, during outages, when a caller cannot speak, and when the delivered location is wrong. No source opened here measures how often those conditions occur.

Annual exposure is a different quantity. In 2022 the states reporting by service type counted 21,194,035 wireline, 157,999,298 wireless and 12,262,577 VoIP calls to 911, a total of 191,455,910. Four jurisdictions did not break out categories, so the real figure is higher. That is calls rather than callers, and one person may call several times while most people never call at all. The Phase 1 and Phase 2 request counts are deliberately not used to size this population, because they measure uptake of a federal trigger rather than deployed capability and fusing the two would break the chain.

Under what conditions does it arise?

1. The regulator can compel carriers and only carriers. The 2024 FCC order places dated obligations on roughly 2,300 originating service providers and sets a default cost rule under which carriers pay to bring traffic to the NG911 delivery point while 911 authorities pay for everything past it. Footnote 462 of that order states that the rules encourage 911 authorities to effectuate the transition but impose no requirements on them, and that the benefits the Commission calculates cannot occur without a transition it does not require.

2. The trigger is local and optional. A carrier obligation begins only when a 911 authority files a valid request, and filing presupposes that the authority has already built and paid for the system that will receive the traffic. Federal deadlines therefore cannot pull anyone forward. Fifteen months after the earliest possible start, the tally stood at more than 190 Phase 1 requests and one Phase 2 request, as of June 2026.

3. Money is raised at the layer with the smallest jurisdiction. 911 is funded by state and local fees. Reported NG911 spending was USD 535,126,846.47 in calendar year 2024 across 42 states plus the District of Columbia, Guam and Puerto Rico, against USD 512,168,670.94 in calendar year 2022 across 44 such jurisdictions. No mechanism moves money from a state that has finished to a state that has not.

4. The defining capability sits on a boundary nobody owns. Handing a call and its data across a county or state line requires two neighboring authorities to hold compatible core services at the same moment. NTIA treats network-to-network interoperability as a secondary requirement priced at an aggregated level and assigns it USD 29 million as of April 2026, unchanged across all three scenarios.

5. Subscription pricing turns a capital raise into a permanent operating line. The study models USD 495,302 of annual recurring cost against USD 200,515 of one-time cost for a 12-position PSAP as of April 2026. Cheaper to start is also harder to ever finish paying for, and it makes the national total sensitive to an accounting choice rather than to engineering progress.

What has been tried?

Two directions have been tried and the third has not. Rules were written for the party that could be regulated, and money was authorized twice for the party that could not. Neither money bill reached a floor vote, and a full-corpus search of the Federal Register for the phrase naming the grant program returns nothing after 2021-08-17.

What was found?

Why is it still unsolved?

Coordination failure — each piece of the transition is held by a different kind of party, and none of them can compel the others.

The first movement is that the regulator regulates the wrong half. The FCC has real authority over carriers and used it: the 2024 order gives roughly 2,300 originating service providers dated duties and a default cost rule, and the Commission priced its own rule at USD 321 million in present value over ten years. Then footnote 462 of that same order sets out the limit. The rules encourage 911 authorities to effectuate the transition but impose no requirements on them, so the Commission excluded their costs from its analysis, while recording that the benefits it claims cannot occur without a transition it does not require. A regulator has written down that the benefit of its rule depends on spending by parties outside its reach. The costed half is USD 321 million. The uncosted half is USD 5,802 million to USD 9,277 million as of April 2026.

The second movement is that the clock runs backwards. Because a carrier obligation starts only when a 911 authority files a valid request, and because filing presupposes the receiving system is already built and paid for, the federal deadline structure cannot pull anyone forward. It can only record who has already moved. Fifteen months after the earliest possible start, more than 190 authorities had asked for Phase 1 and one had asked for Phase 2, as of June 2026. That count should be read as uptake of a federal trigger rather than as a census of capability, because a state whose core services predate the rule has no reason to file at all and the order permits negotiated alternative timelines. Read either way, the phase that delivers the capability the whole program exists for has been requested once.

The third movement is that the capability sits on a boundary while the money is raised inside borders. Handing a call and its data across a jurisdiction line requires two neighbors to hold compatible core services at the same moment, which is precisely the outcome no single authority can produce alone. NTIA prices the entire national interoperability line at USD 29 million as of April 2026 against a total of USD 5,802 million to USD 9,277 million, and at the published unit price of USD 250,000 per interoperation that line buys 116 of them nationwide — a division performed by this dossier, not by NTIA. The FCC, asked in 2026 to mandate interoperability benchmarks and conformance testing, declined both, citing the immaturity of the testing ecosystem, the lack of testing entities and facilities, the absence of standardized procedures and the absence of a commonly accepted standard, against public safety commenters who argued that delay would embed the incompatibilities already in place. What was adopted instead is a definition and one report due within 18 months from a subset of providers.

The pattern holds itself in place because the money is local and annual. Each authority buys what its own budget year can carry, which now means a subscription, and subscriptions convert a one-time raise into a permanent operating line of USD 495,302 a year against USD 200,515 one-time for a 12-position PSAP as of April 2026. That is why the national number now moves with an accounting choice, and why the reported decline of roughly 30 to 40 percent since 2018 cannot be read as progress without a scope crosswalk that no source publishes. What would falsify this reading is straightforward. If the constraint were only that nobody has enough money, the authorities with money would have finished, including across their own borders. Both federal bodies said otherwise in 2026.

What observation would mean it is solved?

Candidates — (a) the FCC publishes a periodic tally showing Phase 2 valid requests and completed Phase 2 deliveries covering the great majority of PSAPs (b) a federal body publishes a national count of jurisdictions that can transfer a call together with its data across a boundary, and that count approaches the total (c) reported annual NG911 expenditure by states reaches the USD 829 million to USD 1,325 million a year that the seven-year schedule implies.

(a) alone counts paperwork. A valid request starts a carrier clock. It does not show that the receiving system works, and the 2024 order lets authorities and carriers negotiate alternative timelines, so a fully capable jurisdiction may never file at all. The count is strong evidence about how little the federal mechanism has been pulled on and weak evidence about capability.

(b) alone is the right measurement and does not exist. No source opened here publishes such a count, so adopting it as the test means adopting a series nobody produces. Until someone produces it, the word both agencies use is exception, which is an adjective standing where a denominator should be. The newest national deployment count available is 2,287 PSAPs across 47 states from 2021 data published in 2023.

(c) alone is money without a destination. The annual requirement of USD 829 million to USD 1,325 million is this dossier dividing the published totals by seven years, and it must travel with its caveat: those totals already include USD 1,538 million to USD 3,138 million of recurring cost on transitions that are complete, part of which is plausibly inside the USD 535,126,846.47 that states reported for 2024. No source publishes the clean subtraction. Beyond that, spending can rise while the boundary problem stays exactly where it is, because a state can buy a complete system inside its own line and still have no compatible neighbor. The three have to be read together, and (b) is the one that would settle it.

What is it connected to?

Fills with researchemergency medical dispatch outcomes, rural broadband and middle-mile availability, the FirstNet public safety broadband network, enforcement against diversion of state 911 fees, accessibility of emergency calling for people who use real-time text or video, and cybersecurity of IP-based emergency networks. Relation type and evidence grade were not confirmed in this round.

What these sources do not say

• The USD 4.5 billion figure has no home. The NTIA blog post announcing the study states that more than USD 4.5 billion has been invested through 911 fees and state and local investment since 2018. A full-text search of the study it announces does not find that figure anywhere, and no method is given for it. The number that establishes what has already been spent, and therefore what the word remaining means, exists only in a blog post with no source line.
• There is no target and no owner. The study carries modelling timelines but no date by which the transition is meant to be done and no named party responsible for the remaining bill. NTIA states that decisions about the appropriate federal role are outside the scope of the study. A number was produced for a decision that the author of the number declines to frame.
• No state can find itself in the number. The estimate is built on six archetypes of state maturity and network structure, and it states that it does not provide individual state cost estimates. So the figure being used to argue for a national program cannot tell any legislature what its own share would be.
• A uniform 15 percent haircut is applied to every state without a derivation. Because the NG911 definition used in the study is broader than the definitions its data sources use, maturity scores are decreased by 15 percent for each state. That single adjustment scales how much work is deemed remaining nationwide, and no basis, range or sensitivity for it is published.
• The comparison with 2018 is presented as like-for-like and never reconciled. The 2026 scope adds artificial intelligence functionality, computer-aided dispatch and over-the-top integration, and removes records management systems, carrier-side location capture and routing, PSAP consolidation, land mobile radio, first-responder equipment and every cost after year seven. No crosswalk between the two scopes is offered, so a reader cannot tell how much of the reported decline is progress and how much is redefinition.
• Nobody counts the harm. No source opened here publishes a count of misrouted 911 calls, of failed transfers between PSAPs, or of calls delivered without location. The benefit estimate in the 2024 FCC order instead runs a proxy chain from a single Salt Lake City study of 73,706 incidents through a linear rescaling to a figure of 16,868 lives a year. The arithmetic is published; no confidence interval, no alternative specification and no measured failure rate stands beside it.
• The territories are inside the spending series and outside the cost estimate. NTIA excluded United States territories from the estimate for lack of data and information available on them. Guam and Puerto Rico report NG911 expenditure to the FCC every year. No source notes the inconsistency.
• Nothing says what happens next. The only new interoperability instrument is a one-time report due 18 months after a notice. The order does not state what threshold would count as failure, what would trigger further action, or when the Commission would look again. The same silence covers the deployment evidence itself, 45 percent of which comes from a voluntary survey that states may decline to answer.

See the evidence

None of the four sources in the blank-URL rows was opened. The four NTIA pages, the four Federal Register documents, the bill text, the NASNA page and the two legislative-status pages were opened directly and carry every figure attributed to them above. One operational caveat has to travel with this table: the four ntia.gov URLs were retrieved by direct HTTP fetch and returned 200, but they fail TLS chain validation under a fetch tool, so an automated re-check will report those rows as dead when they are not. Second hand by necessity are the two state-spending totals and the ESInet count, which reach this document only through Federal Register text quoting reports that returned 403, and the 2018 estimate of USD 9.5 billion to USD 12.7 billion, which is known here only through the 2026 characterization of it and must not be treated as primary. Three disagreements are left visible rather than resolved. The 2024 FCC order gives 2,327 originating service providers in its body and 2,287 in its own footnote, and corrects neither. Inside the cost study, a table places legacy system savings and current funding capacity within the cumulative national estimate while section 5.1 states that the savings have not been deducted from that estimate — the two readings cannot be separated from a text layer alone, so this is recorded as an unreconciled ambiguity and not as a proven error. And reported NG911 expenditure rose 4.48 percent between 2022 and 2024 while the number of reporting jurisdictions fell from 44 to 42, with no source saying whether the two missing jurisdictions spent nothing or simply did not report, so that percentage cannot be read as a change in spending. This is a Path A output, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range 300,000,000–334,914,895 As of 2026 capability statements on a 2023-07-01 population base

Sensitivity The band is narrow because the honest answer is close to the entire population. This is not a shortfall concentrated in one disadvantaged group; it is a national system that has not been finished anywhere end to end, so the figure is bounded above by a census count and below by an adjective. What the number fails to count is harm. It does not claim that any call fails, because most 911 calls complete on legacy infrastructure and the failure is conditional on crossing a jurisdiction boundary, on an outage, on a caller who cannot speak, or on a wrong delivered location, and no source opened in this round measures how often those conditions occur. In the opposite direction the number is also too small and too coarse. It excludes United States territories, which NTIA left out of the cost estimate for lack of data even though Guam and Puerto Rico report NG911 expenditure to the FCC every year, and it cannot separate people served by a PSAP with no ESInet from people served by a PSAP that has one and still cannot transfer a call with its data across a line. The counts of more than 190 Phase 1 and one Phase 2 valid requests are deliberately not used as a population step, because they measure uptake of a federal trigger rather than deployed capability; fusing the two would break the chain. The annual exposure figure of 191,455,910 voice calls to 911 in 2022 is calls rather than callers and covers only the states that reported by service type, so it is neither a substitute for this range nor a subset of it.

Regional breakdown No source opened in this round publishes a state-level or jurisdiction-level count of end-to-end NG911 capability. The NTIA cost study is built on six archetypes and states that it does not provide individual state cost estimates. The FCC reports that Phase 1 requests cover 28 states but does not publish the list of which ones. The newest national PSAP-level deployment count available is 2,287 PSAPs across 47 states, from 2021 data published in 2023. Splitting the national figure by population share was rejected rather than attempted.