One router, 20,000 disrupted flights: L3Harris maintenance blamed for the FAA’s five-hour ground stop

How did one router stop every departure in America?

On Friday 13 November, the United States stopped flying. At about 15:54 Eastern time, the Federal Aviation Administration ordered a nationwide ground stop, every departure held, after its air traffic communications began failing across the country. The halt lasted nearly five hours, until about 20:15, and it was only the second nationwide stop since the attacks of 11 September 2001.

The scale became clear within hours: more than 20,000 flights disrupted since Friday, according to Reuters, including more than 8,200 cancellations on Friday alone, over 31 percent of the day’s schedule, per Cirium, and more than 7,700 cancellations and 8,000 delays, per FlightAware.

On Saturday, Transportation Secretary Sean Duffy named the suspected trigger: routine maintenance by contractor L3Harris, replacing old copper wire with fibre at a remote FAA location. A newly installed router, officials said, began flooding other routers with more than 1,000 connection attempts every four to five seconds, overloading communications at 17 of the FAA’s 22 en route centres and corrupting radar feeds and controller communications. That sequence turned a local upgrade into the nationwide halt reported on Friday morning.

“This was an FAA system that didn’t meet the moment,” Duffy said on Saturday, adding the department would rethink when such work is done at smaller facilities.

What the Saturday briefing established; and what it did not

What is established is the sequence: L3Harris was switching telecommunication lines from copper to fibre when the failure began, and the new equipment started flooding the network. The FAA said communications did not fail outright, but the data at the centres became unstable; once officials isolated the work, the systems came back online.

What is established, too, is who runs the network: L3Harris, the FAA’s prime contractor for its telecommunications infrastructure, supporting about 45,000 daily flights and connecting more than 4,400 facilities. The firm says it has completed more than 70 percent of the planned copper-to-fibre conversion. The routers being replaced are about 25 years old, Duffy said, and were due for replacement in 2006.

What is not established is the fault itself. “We’re working to determine whether we had incompetence, meaning was it user error or installer error from L3Harris, or did they install a faulty router? We haven’t determined what the fault was,” Duffy said. Officials told Reuters there may have been problems with routers installed in both Arkansas and Texas.

What officials say they have ruled out, for now, is a cyberattack. “Once we were able to isolate the work being done at this location, all of our systems came back online,” Duffy said. “So, we don’t believe this was cyber at all.” FAA Administrator Bryan Bedford added: “We see zero evidence of anything nefarious, and zero evidence of anything AI-related.”

What investigators still cannot say about the router

The central unknown is the one Duffy named on Saturday: faulty equipment or faulty installation? Until the FAA’s investigation answers that, L3Harris is named as the contractor whose work was under way when the system failed; but not found at fault. The distinction matters: a bad router is a procurement question; a bad installation, a competence question.

A second unknown is geography. Officials told Reuters there may have been problems with routers in both Arkansas and Texas, but the department has not confirmed how many sites were involved, or whether the flood began at one location and spread, or at two at once. The FAA’s acting chief technology officer, Rebecca Guy, said only that when the network began showing instability, “we started to look at what was the cause.”

A third unknown is accountability. Duffy was blunt, “We’re going to hold people accountable”, and blunt about the dependency: “They own this network. They’re our prime contractor. I can’t go somewhere else. They own it.” What form accountability takes, and against whom, has not been decided. Nor has the full cost been counted: cancellations were still being tallied on Saturday.

Why a maintenance window became a national failure

A single router should not be able to stop a continent’s aviation. That it could is the story: the FAA’s backbone still runs on copper laid down a generation ago, maintained at about 1 billion dollars a year. Congress has granted 12.5 billion dollars for the upgrade, and Duffy has asked for 50 billion dollars more for air traffic control and airports. The ground stop is now the most vivid exhibit in that request.

The pattern is not new: a 2023 nationwide ground stop that lasted 90 minutes and disrupted 11,000 flights was caused by a government contractor unintentionally deleting files; the first such stop since 11 September 2001. In September, thousands of Northeast flights were delayed after a circuit failed in Philadelphia and a crew severed a fibre-optic cable in New Jersey. “The system will fail again until it is properly upgraded,” officials told Reuters.

The deeper problem is the upgrade trap: the FAA cannot switch its network off to replace it, so every step happens while the old system carries live traffic; new routers talking to 25-year-old ones, fibre spliced into copper, at remote sites, on weekends. The FAA only last week said it would replace by the end of 2028 the floppy disks still used at more than 400 facilities; controllers still use paper flight strips, a change due at 49 airports only in 2028. The network is being rebuilt in flight, and on Friday the rebuild bit back.

The upgrade trap: replacing the system while it flies the planes

The FAA’s dilemma has no clean solution. A network carrying 45,000 flights a day cannot be taken offline for a rebuild, so the rebuild must happen inside the live system; new equipment introduced site by site, weekend by weekend. Friday showed what happens when one small risk is not small: a router that talks too much can drown the network it was installed to save.

L3Harris’s position makes the politics harder. The company is not one vendor among many; it manages the FAA’s telecommunications infrastructure, and as Duffy put it, “I can’t go somewhere else. They own it.” A prime contractor that owns the network is both indispensable and difficult to discipline, and the investigation must apportion fault between the contractor’s work and the FAA’s own management of a modernisation it has run for decades. L3Harris says it takes the issue seriously and is supporting the investigation; the department says it will hold people accountable.

There is also the timing: the stop came a day after the FAA announced an AI tool for air traffic controllers; a new layer on top of the old one, with the controllers’ union saying it was not consulted. The contrast is American aviation infrastructure in miniature: new capabilities at the top, 25-year-old routers and floppy disks underneath. The American profile in our Factbook tracks the same gap between the country’s technological reach and the age of the systems that carry it.

What a five-hour halt does to the week after

The first second-order effect is mechanical. “We have crews and aircraft that are out of position for 12,000 flights that were lost,” Bedford said on Saturday. “These crews, pilots and flight attendants need to get to their points of reconnection with their schedules.” The FAA expects about 4,000 to 5,000 of the 41,000 flights scheduled on Saturday to be cancelled, and full schedule integrity only by the end of Sunday. The backlog does not clear when the ground stop lifts; it clears when every crew and every aircraft is back where the schedule expects them.

The second effect is financial, and it lands unevenly: passengers first, in missed connections and lost weekends; airlines next, in rebooking costs and compensation; the contractor last, in its negotiating position with its only customer. The third effect is political: Duffy’s 50-billion-dollar request now has a five-hour exhibit attached to it, and the next maintenance window will be watched as no maintenance window has been watched before.

When the answers arrive; and what Duffy has promised

The next fact in this story is the root-cause analysis, which Bedford said will be shared with the public when it is available. That report will have to answer the question Duffy left open on Saturday: faulty router, or faulty installation; and whether the problem sat in one state or two. Until it lands, the suspected sequence stands as the official account, not the confirmed one.

Watch, too, for three things: whether Duffy’s promise of accountability produces a named consequence for the contractor or for the FAA’s own oversight; whether Congress moves on the 50-billion-dollar request, and on what terms; and whether the FAA changes how maintenance is done at remote sites. The next upgrade window is the first test of whether Friday changed anything, or whether the system simply goes back to being rebuilt in flight.