Next Generation 911: Rebuilding America’s Emergency Communications System for the Digital Age

A modern smartphone can determine its location, transmit high-resolution images and live video, detect a severe vehicle crash, exchange data with satellites and continuously communicate with applications, vehicles, wearables and sensors. Yet the United States is still completing the migration of 911 from infrastructure whose basic design assumptions came from the telephone era.

Next Generation 911 — NG911 — is the effort to close that gap. The National 911 Program describes NG911 as a digital, Internet Protocol-based system intended to replace legacy analog 911 infrastructure. It is designed to move voice, text, photographs, video and other data more seamlessly while improving resiliency and location-based call handling.^[1]

That definition sounds simple. The engineering is not. NG911 changes how emergency traffic is transported, how location is validated, how a call is routed, how one emergency communications center can transfer information to another, how data is protected, and how a national network of locally governed systems can interoperate.

NG911 is not “911 in the cloud,” and it is not merely Text-to-911 or video-to-911. It is a standards-based, location-aware emergency communications architecture.

From a telephone call to a digital public-safety network

Traditional 911 and Enhanced 911 were built around telephone calls. Enhanced 911 added capabilities such as automatic telephone-number and location information and selective routing. These systems became extraordinarily dependable, but they were built around circuit-switched communications and databases whose logic predated smartphones, mobile broadband, cloud computing, connected vehicles and modern geospatial platforms.

NG911 moves toward secure IP networking and standardized interfaces. The objective is not simply to accept more media. The objective is to create a nationwide system of systems in which independently governed emergency networks can exchange calls and associated data using common standards.^[1][5]

That national end state is important because 911 in the United States is highly decentralized. States, counties, municipalities, regional authorities, tribal governments and federal organizations operate different emergency communications environments. NG911 therefore has to solve both a technology problem and a governance problem.

The technical core: ESInet + Next Generation Core Services

At the transport layer is the Emergency Services IP Network, or ESInet. It is not simply the public Internet. An ESInet is a managed IP network used to carry emergency-service traffic among authorized systems and emergency communications centers.

Running across that network are Next Generation Core Services, or NGCS. These services provide much of the intelligence required to route and manage emergency communications. NENA’s i3 standard defines the detailed functional and interface specifications for a post-transition IP-based multimedia 911 environment and the legacy gateways needed during migration.^[5][6]

A simplified call flow looks like this:

1. A person or device initiates an emergency communication.
2. The originating service provider delivers the traffic toward the appropriate NG911 environment using standards-based IP signaling when the transition phase supports it.
3. Location information is associated with the communication.
4. NG911 core functions use routing policy and authoritative geographic data to determine where the emergency belongs.
5. The emergency communications center receives the call or message along with the information available to that jurisdiction and system.
6. The telecommunicator validates the situation, enters or updates the incident in CAD, and dispatches or transfers the response.

During the national transition, legacy and NG911 systems have to coexist. That coexistence is one reason the migration is difficult: jurisdictions can incur the complexity, cost and risk of operating both old and new pathways at the same time.^[2]

GIS is no longer just a map on the dispatcher’s screen

One of the least understood NG911 changes may be one of the most consequential: Geographic Information System data becomes operational routing infrastructure.

In the i3 environment, road centerlines, address points, PSAP boundaries, police boundaries, fire boundaries, EMS boundaries and other authoritative layers can feed location validation and emergency call routing. NENA’s 2026 NG9-1-1 GIS Data Model defines the data structures used by NGCS, and New York’s statewide plan explicitly identifies GIS as essential to the spatial functions of NG911.^[7][13]

This changes the significance of geospatial data quality. A misplaced address point or an incorrect service boundary is no longer merely a cartographic defect. If that layer participates in routing, a data defect can become a public-safety defect.

That means NG911 requires continuous GIS stewardship: authoritative ownership, change control, quality assurance, edge matching with neighboring jurisdictions, version management and processes for rapidly correcting discrepancies. The map becomes part of the network.

Text, photographs, video and connected-device data

The NG911 architecture is designed to support emergency communications beyond traditional voice, including text and richer multimedia.^[1] That does not mean every PSAP can accept every data type today. Deployment maturity varies widely, and an agency can support Text-to-911 without having reached an end-state NG911 architecture.

This distinction matters. A jurisdiction may have upgraded call-handling equipment, implemented an ESInet, deployed text service or connected to supplemental data while still using legacy gateways or non-i3 components elsewhere. The National 911 Program’s maturity model therefore distinguishes multiple stages from legacy through national end state.^[1]

Over time, the information entering an emergency center can come from much more than a caller’s voice:

• smartphone location and device information;
• vehicle crash notifications and telematics;
• wearable-device alerts;
• building alarms and security systems;
• text, photographs and live video;
• medical information voluntarily associated with emergency services;
• cameras, drones and other situational-awareness systems;
• future sensor and Internet-of-Things sources.

Google’s Android Emergency Location Service illustrates the endpoint side of this evolution. ELS can compute supplemental emergency location on the device using GPS, cellular, Wi-Fi and other sensor information and send it directly to participating emergency-service endpoints; Google states that it does not receive that emergency location itself.^[27] Apple likewise offers safety features such as Crash Detection and Emergency SOS via satellite on supported devices.^[28]

These technologies are not NG911 themselves. They are endpoint capabilities capable of feeding a more capable emergency communications ecosystem.

Where AI is connected to NG911 — and where it is not

Artificial intelligence needs a precise place in this discussion because vendors increasingly market AI and NG911 together.

AI is not a defining element of the NENA i3 core architecture. The core emergency network must route and protect life-safety traffic using standards, authoritative location, policy and deterministic network functions. An Emergency Call Routing Function should not have to ask a generative model where it “thinks” the emergency belongs.

AI is better understood as an assistance and intelligence layer that can sit before, beside or after the NG911 call path. NTIA’s 2025 national study of AI in 911 described real-world use cases and emphasized the need for thoughtful implementation, governance and public trust.^[11]

AI before the call reaches a human

Algorithms may detect that something is wrong before a person can fully describe it. A vehicle may identify a severe crash. A watch may detect a fall. A security system may identify an alarm condition. Those systems can generate structured data that can eventually accompany or augment an emergency request.

AI during the call

This is where AI is becoming commercially visible now. Practical functions include:

• real-time speech transcription;
• language identification and translation;
• automatic extraction of names, addresses, vehicle descriptions and incident details;
• call summaries;
• duplicate-incident correlation;
• keyword or risk alerts;
• quality-assurance support;
• automated handling of appropriate non-emergency workload;
• organizing supplemental information from devices, vehicles, maps and cameras.

Motorola Solutions describes VESTA NXT as providing real-time AI transcription and translation, AI review of transcripts and supplemental information, and concise data packages for dispatch.^[22] Intrado markets real-time translation and AI-assisted 911 workflows within its PSAP portfolio.^[20] Axon describes Prepared AI as an assistive intelligence layer for transcription, translation, key-detail extraction, summaries, CAD handoff and quality assurance.^[26]

AI as the antidote to data overload

NG911 creates a paradox. A traditional call may present one dominant information stream: voice. A future incident could arrive with voice, text, device location, indoor-location context, video, crash telemetry, medical information, building data, alarm data and several external camera feeds.

More information is not automatically more situational awareness. It can become cognitive overload.

The highest-value AI role may therefore be less about “making the decision” and more about finding the six facts the telecommunicator needs now from the thousands of data points that may be available.

But AI errors become life-safety errors

An incorrect movie recommendation is trivial. An incorrect emergency translation, omitted weapon description, bad address extraction or misleading incident summary is not. Public-safety AI therefore needs stronger controls than ordinary consumer AI: traceability to the original source, visible uncertainty, human review, audit logs, cybersecurity controls, data-retention rules, model-performance monitoring and explicit fallback procedures.

AI may transcribe, translate, summarize, flag and recommend. Accountable humans and independently reliable emergency infrastructure still need authority over life-safety decisions.

Cybersecurity: connectivity creates both resilience and attack surface

Moving from isolated legacy systems to interconnected IP networks creates major benefits: redundancy, remote operations, alternate routing, multimedia and easier interconnection. It also creates familiar cyber risks on infrastructure that cannot tolerate ordinary levels of failure.

CISA warns that greater NG911 interconnectivity creates additional vectors for disruption, including denial-of-service attacks, Telephony Denial of Service, unauthorized network access, data breaches, malware, ransomware, phishing, spoofing and malicious applications. A disruption can create call backlogs, delayed emergency response or a forced transfer to manual or alternate operations.^[9]

NENA’s NG-SEC standard establishes minimum security requirements for NG911 entities and assets, with companion audit and NIST Cybersecurity Framework crosswalk resources.^[8]

Security therefore cannot be an add-on purchased after deployment. It has to be part of architecture and operations: segmentation, certificate and identity management, encryption, patching, monitoring, logging, backup power, network diversity, incident response, vendor governance, continuity planning and regular testing.

The fundamental design problem is subtle: NG911 must be interoperable without becoming indiscriminately interconnected.

The federal transition framework changed the timetable

For years, states and local authorities could build NG911 infrastructure yet remain dependent on originating service providers to deliver traffic through legacy interfaces. The FCC’s 2024 NG911 Transition Order created a national two-phase framework intended to reduce that deadlock.^[2]

Under Phase 1, a qualifying 911 Authority can request that covered originating service providers deliver 911 traffic in the requested SIP format, connect through the authority’s ESInet and NGCS, route to designated NG911 delivery points and complete connectivity testing. Phase 2 advances location and additional i3-aligned requirements after Phase 1 is complete or already satisfied.^[4]

The FCC established default implementation windows that vary by provider type. Nationwide wireless carriers and several major provider categories generally have shorter default periods than rural wireline and non-nationwide wireless providers; parties may negotiate different arrangements.^[2]

Then in 2026 the FCC modernized 911 reliability and interoperability rules for IP-based NG911. The rules were published July 10, 2026 and became effective August 10, 2026, while several new compliance obligations are being phased in after additional FCC public notices and information-collection approvals.^[3]

The 2026 framework expands reliability expectations to providers that control critical NG911 pathways and components, brings IP-network best practices into reliability benchmarks, and creates interoperability reporting obligations for ESInet and NGCS providers. In other words, the transition is no longer just about getting onto IP. It is increasingly about proving that the IP-based life-safety system is reliable and interoperable.

How much remains to be built?

A new 2026 NTIA cost study estimates the remaining nationwide investment required to complete the transition at approximately $5.8 billion to $9.27 billion, depending on the post-implementation period included in the seven-year 2026-2032 analysis. NTIA notes that this is lower than the 2018 estimate in part because states have already invested and the market has shifted from equipment-heavy deployments toward software-driven and service-based models.^[10]

That shift matters. Cloud and “as-a-service” architectures can reduce upfront capital expenditure, but they also move more cost into recurring operating expenditure. NG911 should therefore be budgeted as an ongoing critical digital service, not as a one-time replacement of telephone equipment.

Recurring cost categories include network services, cybersecurity, software licensing, GIS maintenance, certificates, monitoring, call-handling systems, testing, staff training, data storage, disaster recovery and vendor support.

New York is a case study in why NG911 is difficult

New York is especially instructive because New York City and New York State are not at the same stage of migration.

New York City: advanced implementation

New York City began developing its NG911 strategy in 2014. The City’s 2025 annual report describes its 911 operation as the largest and most complex emergency communications system in the country, receiving approximately 9 million calls per year.^[17]

NYC divided the modernization into major workstreams: Class 1 ESInet and Core Services; Class 2 Logging and Recording; Class 3 GIS; and a new Call Handling System. By the 2025 report, major ESInet/core, logging and GIS milestones had moved deep into production, including the NextGen Location Database. The remaining major focus was the call-handling migration, which the City said it aimed to complete by summer 2026.^[17]

It would be premature, however, to state that the entire NYC transition is definitively complete without an official completion declaration. City procurement notices in 2026 continued to seek NG911 QA and GIS specialists through November 30, 2026 to support configuration, integration and testing for Class 1/2/3 and call-handling upgrades.^[18]

That makes the most defensible September 2026 description: New York City is in a very advanced implementation and transition stage, substantially ahead of the statewide New York program.

New York State: statewide architecture and procurement

The State’s path has been slower. A June 2025 New York State Comptroller audit—covering activity through December 2024—found that statewide planning had lagged and that 22 of 36 responding counties said they had received no specific NG911 guidance from the State. Importantly, that audit predates the finalized 2025 Statewide 911 Plan, so it should be read as a diagnosis of the earlier planning period rather than the State’s current end point.^[14]

The 2025 Statewide 911 Plan now places DHSES’s Office of Interoperable and Emergency Communications in the lead role for developing and implementing the statewide NG911 system. It calls for a statewide ESInet and NGCS foundation with GIS-based routing, location validation, security, interconnection and failover. The plan identifies 133 primary PSAPs, 34 secondary PSAPs and 25 ancillary PSAPs for consideration in the statewide effort.^[13]

It also acknowledges a major operational problem: CAD systems across New York are heterogeneous, and even jurisdictions using the same vendor may run different versions or customizations. The plan states that some CAD systems are incompatible and cannot transfer incident data without custom interfaces.^[13]

This is a critical distinction. NG911 may make it easier to transfer an emergency call, but moving the entire incident record between incompatible CAD environments is a separate interoperability challenge.

The State must connect to the City

The Statewide 911 Plan explicitly recognizes New York City’s separate NG911 deployment and states that the statewide system will need to interface with the NYC ESInet and NGCS for data sharing and transfer of 911 calls to and from New York City, other New York PSAPs and neighboring jurisdictions.^[13]

That gives New York a real-world illustration of the national NG911 idea:

Not one monolithic network. Interoperable networks.

New York State procurement is now real, not theoretical

At publication time, the New York State Contract Reporter listed a statewide procurement titled Next Generation 911 System Development and Independent Verification, issued June 9, 2026 with proposals due September 14, 2026. The State classifies it as a statewide telecommunications procurement through DHSES OIEC.^[16]

Meanwhile, the combined 2024-2026 ESInet Readiness Grant makes $40 million available to counties and New York City to improve hardened network connectivity and prepare for connection to the State ESInet.^[15]

New York therefore illustrates the entire NG911 problem in one state: a very large city building its own advanced system, dozens of locally operated PSAPs at different technology levels, statewide GIS and network requirements, regional infrastructure that may be reused, incompatible CAD environments, and a need to connect everything without interrupting 911.

The commercial ecosystem: no single company “owns” NG911

NG911 is standards-driven but commercially implemented. Telecommunications carriers, network providers, call-handling companies, GIS specialists, public-safety software vendors, cloud providers, device manufacturers and data platforms occupy different layers.

The figures below are vendor-reported unless otherwise stated. They demonstrate market scale, not independent proof of nationwide NG911 completion.

This vendor convergence is significant. Historically, 911 routing, call handling, CAD, field communications, body cameras and evidence management were distinct technology markets. They are beginning to overlap.

The risk is vendor lock-in. The opportunity is a more continuous information chain. Standards and procurement discipline determine which outcome wins.

NG911 and FirstNet are complementary, not interchangeable

NG911 and FirstNet are often discussed together but perform different functions. NG911 primarily modernizes how emergency communications move from the public into and among emergency communications centers. FirstNet is the nationwide public-safety broadband network serving authorized responders and agencies. The National 911 Program explicitly treats them as separate but complementary systems.^[12]

As integration improves, information captured during the emergency call can potentially travel farther downstream: caller video, vehicle telemetry, floor plans, building information or structured incident data may become available to responders before arrival—subject to policy, privacy, security and operational design.

What could still go wrong?

The technology is impressive, but NG911 is not automatically successful simply because a jurisdiction buys new equipment.

1. Interoperability without operational interoperability

Two networks may technically exchange a call while their CAD systems, incident taxonomies, radio systems or operating procedures still cannot exchange the information needed to act on it.

2. Data overload

A telecommunicator with more cameras, maps, alerts and device feeds can become slower rather than faster unless systems prioritize information intelligently.

3. Cyber dependence

IP networks increase flexibility but also increase dependency on software, credentials, certificates, cloud services, third-party transport, monitoring and supply chains.

4. GIS governance

Local governments must continuously maintain authoritative address and service-boundary data. There is no “finish” date for geographic accuracy.

5. Funding mismatch

Traditional 911 funding structures were often built around telephone-service assumptions. NG911 requires recurring software, cybersecurity and network spending.

6. Records, privacy and evidence

Voice calls already create public records and evidentiary issues. Video, photographs, medical data, device data and AI-generated summaries create much larger questions about retention, access, redaction, discovery and chain of custody.

7. Human factors

Technology must reduce workload rather than merely add screens. Training, staffing, quality assurance and psychological workload remain central to system performance.

What NG911 is likely to become

The next stage of emergency communications will probably be defined by the convergence of several trends:

• more SIP/IP delivery from originating service providers under the FCC transition framework;
• broader i3 implementation and interstate network interoperability;
• GIS becoming authoritative routing infrastructure;
• native multimedia and richer supplemental data;
• more cloud-native and service-based 911 platforms;
• AI-assisted transcription, translation, summarization, triage and quality assurance;
• stronger cybersecurity, reliability and interoperability requirements;
• more structured exchange of incident data between ECCs, CAD systems and responder networks;
• closer integration between 911 intake, real-time operations, field response and digital evidence.

NENA’s standards continue to evolve. In 2026, ANSI approved revisions including the NG9-1-1 GIS Data Model and a Transition to i3 PSAP Standard, underscoring that implementation is still moving while the standards themselves mature.^[7]