When the Federal Aviation Administration issued a nationwide ground stop on November 13, 2026, the immediate explanation was familiar: aging infrastructure, a telecommunications failure and a modernization project that had gone wrong. The harder question is more important than the identity of any single failed device.
Why could a network change at a remote site destabilize communications and radar systems across much of the national air-traffic-control system?
The answer remains under investigation. The available evidence does not establish that a particular Juniper router, a specific configuration or L3Harris alone caused the outage. It does show that the FAA was changing a safety-critical network while legacy systems remained in service—and that the transition did not contain the resulting disruption quickly enough.
A local change with a national blast radius
The FAA grounded departures for approximately four hours and 45 minutes on November 13 after telecommunications instability affected air-traffic-control communications and radar systems. More than 3,500 aircraft were already airborne, and 17 of the FAA’s 22 en-route control centers were reportedly affected, according to a Congressional Research Service summary.
The Department of Transportation’s account said the initial investigation focused on work converting older copper and time-division-multiplexing systems to fiber and Internet Protocol systems at a remote FAA site. L3Harris operates the FAA telecommunications network.
That description identifies the setting, not necessarily the cause. A router can function normally in isolation and still interact badly with older switches, routing policies, timing systems, monitoring tools or traffic assumptions elsewhere in the network. The technical failure may have been a device problem, a configuration problem, an interoperability problem or several of those at once.
What matters operationally is that an abnormal condition apparently traveled beyond the site where the work was being performed. A well-designed network does not need every component to be perfect. It does need failures to remain inside a defined boundary.
Modernization creates dependencies before it removes them
The FAA’s telecommunications effort is large enough to create precisely the kind of dependency problem that ordinary equipment-replacement language tends to hide. The program involves replacing 5,170 copper lines, transitioning nearly 30,000 services to IP, replacing legacy voice switches and digitizing radio systems, according to Government Accountability Office reporting.
During that work, the FAA must operate several technical systems at once:
- legacy equipment still carrying production traffic;
- new routers and fiber connections being introduced incrementally;
- different protocols and interfaces operating together;
- contractors changing systems that cannot simply be taken offline; and
- communications, surveillance and automation services that may depend on common network paths.
This is normal for a large infrastructure migration. It is also where the system becomes difficult to understand. The new system is not yet independent, and the old system is no longer the only system anyone has to understand.
The relevant risk is not simply old equipment versus new equipment. It is dependency density: how many services rely on the same routing layer, timing source, management system, vendor, configuration process or recovery path.
The device is not the same as the failure
Congressional investigators said information provided to the Senate Commerce Committee indicated that the conversion involves approximately 3,648 routers, with about 400 installed and 69 operational when the committee wrote to the FAA and DOT on November 16. The committee requested a pause in further deployment until the FAA could demonstrate stronger operational testing, independent validation, verified failover and rollback procedures. The committee’s letter does not, however, establish that the routers caused the ground stop.
That distinction is not technical hair-splitting. The device identified during an investigation may be the trigger, the first visible symptom or merely one component in a larger configuration failure.
The investigation needs to establish what generated the abnormal traffic, which network boundaries it crossed and why the condition reached so many remote facilities. It also needs to determine whether radar and communications systems were sufficiently separated, whether rate limits constrained abnormal behavior and whether operators could disable affected sites without destabilizing other services.
Until those questions are answered, the strongest conclusion is that the FAA experienced a failure of interaction between systems during a migration—not that fiber is unsafe, that every new router is defective or that modernization itself should stop.
Testing must resemble the network that actually exists
Safety-critical infrastructure cannot be validated only in a laboratory or against an idealized network diagram. New equipment must be tested against the legacy devices, traffic patterns, timing assumptions, management tools and failure conditions it will encounter in production.
That is especially important when a contractor performs work on an operational system. The FAA and L3Harris need clearly assigned responsibility for the change window, configuration approval, acceptance testing, monitoring and rollback. If those responsibilities are divided without a single operational owner, each organization may possess part of the explanation while neither can immediately restore the whole system.
The FAA also needs to demonstrate that its monitoring can do more than announce that service has been lost. Operators must be able to identify the initiating condition, isolate the affected segment and understand whether a recovery action is reducing or spreading the problem.
Earlier GAO findings on aviation communications cybersecurity identified weaknesses involving monitoring, risk assessment and coordination. Those findings do not prove that the November 13 outage was a cyberattack. They do reinforce a broader point: availability failures and security failures often expose the same organizational weakness—an incomplete understanding of how systems connect and who is responsible for them.
A schedule is part of the safety case
GAO has also reported that the FAA modernization effort contained more than 11,000 individual project schedules that were not integrated into a single master schedule at the time of its review. That is not merely a project-management concern.
When communications, voice, radar, automation and weather systems are being changed in overlapping programs, schedule integration becomes a technical control. It tells the agency which systems are changing simultaneously, which contractor activities create shared risk and when a temporary configuration will remain in place longer than expected.
The FAA’s modernization case will be stronger if it can show that the program now has an integrated schedule, system-level dependency map, independent acceptance testing and recovery exercises based on realistic operating conditions. A larger appropriation without those controls would not necessarily produce a safer network. It could simply expand the number of components being changed without improving the agency’s ability to see how they interact.
What the FAA should prove before work resumes
The agency does not need to promise that another failure will never occur. Complex systems will fail. It needs to demonstrate that the next failure will be smaller, visible sooner and reversible without a national ground stop.
That means publishing or providing oversight bodies with evidence of:
- logical segmentation that limits the spread of abnormal traffic;
- independent communications and operational paths for essential services;
- monitoring that identifies initiating conditions rather than only downstream outages;
- field testing against the actual legacy equipment and traffic patterns in use;
- clear change-approval and configuration-control records;
- failover procedures tested under realistic conditions;
- a rollback path that can restore the previous topology without losing operational awareness; and
- defined responsibility among the FAA, L3Harris and equipment suppliers.
Those are not optional extras added after modernization. They are part of modernization.
The transition is itself an operational system
The FAA outage did not show that replacing copper lines with fiber or moving from older telecommunications protocols to IP is inherently unsafe. It showed that the transition between old and new equipment is itself a system that must be engineered, monitored and tested.
In an office, replacing a network router may cause inconvenience. In a national air-traffic network, it is more like replacing a junction in a railway system while trains are still running. The new junction may be better, but the transition requires signals, isolation, traffic limits and a reliable way to restore the old route if the new one behaves unexpectedly.
Modernization remains necessary. But a modern network is not resilient merely because its components are newer. Resilience comes from containment, observability, independent paths, tested fallback procedures and the ability to reverse a change before a local problem becomes a national event.
The FAA has restored service. It has not yet publicly demonstrated that it can safely manage the next transition. That is the technical question the investigation—and any future modernization funding—must answer.













