Sixty-nine percent of the copper lines that carry United States air traffic control data now run on fiber and wireless links, and the Federal Aviation Administration put its first five facilities on a new controller display system this month (Department of Transportation, 2026). The agency is working through a $12.5 billion appropriation toward 5,170 new network connections by the end of 2028 (Federal Aviation Administration, 2026).

On Friday, November 13, a router replaced at one remote site in that network flooded the routers around it, and the FAA stopped departures nationwide for more than four hours. FlightAware counted 9,159 cancellations and 7,190 delays that day (FlightAware, 2026), and CNBC reported that weather accounted for a share of them.

I cover applications, security, infrastructure and compute together, and I laid out the reasoning in an earlier post on why I cover the entire stack. Friday is the clearest public example I have seen of that reasoning at work, because a fault at the network layer stopped work at every layer above it.

One router at a remote site reached the whole network

Transportation Secretary Sean Duffy of the Department of Transportation said Saturday that a contractor from L3Harris, the company that runs the FAA's legacy telecommunications network, was moving a line from copper to fiber and replacing a router at a remote location. The replacement began flooding other routers with traffic. Duffy put the volume at 1,000 log-on attempts every four to five seconds, and the flood ran for 45 minutes before the FAA could find the source of the instability (CBS News, 2026). More than 3,500 aircraft were in the air, and all of them landed safely.

Officials have yet to say if the router was faulty or if it was installed or configured wrong, and the investigation is open. L3Harris had not commented when NPR published its report on Saturday. FAA Administrator Bryan Bedford described the swap as routine maintenance, separate from the modernization program itself.

Bedford's description is the part I would bring to your change board. If a swap filed as routine maintenance can reach every facility on the network, the label on the ticket tells you little about the risk, and the review should follow the reach of the device instead of the size of the project.

Every layer above the network inherited the fault

Friday's failure began in telecom and showed up in each of the four layers I track.

• Applications. The FAA has moved 26 towers to electronic flight strips and started its Enterprise Information Display System at five facilities, on a plan for 435 (Department of Transportation, 2026). Each of those tools reaches controllers over the same telecom network, so a new interface is as available as the oldest router in its path.
• Security. Homeland Security Secretary Markwayne Mullin of the Department of Homeland Security said Friday there were no indications of a cyberattack. A thousand log-on attempts every few seconds from a trusted device still has the traffic pattern of a denial-of-service attack, and rate limits and network segmentation contain that pattern whoever sends it.
• Infrastructure. Duffy said the routers in these facilities were manufactured in 2001. With 31 percent of the copper still to replace, the FAA will run two generations of equipment side by side until 2028, and every cutover in that period connects a new device to a neighbor that is 25 years old.
• Compute. A router is a small computer with a fixed processor and memory. Devices built in 2001 have little headroom to absorb a flood of log-on requests, which helps explain how one source overloaded its neighbors. I would ask the same headroom question about every older system that sits next to a new one in your data center.

"Virtually every single institution in America has gone from copper to fiber."

Sean Duffy, Transportation Secretary, speaking to reporters on November 14, 2026, as reported by NPR

Your provider's upgrade record belongs in the buying decision

The airlines had no vote on the router that stopped their departures on Friday. They ran on a network someone else owned and maintained, and they absorbed the cancellations. You have a choice the airlines lacked, because you select the carrier, the cloud region and the colocation site your business runs on.

Picture a manufacturer in a rural town that makes a component its customers cannot build without. The plant may run modern machines and current software and still send every order, shipping notice and machine reading through a local provider whose lines and routers date from the same era as the FAA's. The oldest equipment in that provider's network then sets the plant's uptime, along with the uptime of every customer waiting on its parts.

Rural networks can be new. Roughly 500,000 of India's 640,000 villages now have 5G coverage (Communications Today, 2026), which means a phone user in most Indian villages connects through network equipment installed in the past four years. The FAA routers that overloaded on Friday were built in 2001.

I would choose infrastructure companies that keep upgrading their pipes and can show you the schedule. A provider that postpones upgrades passes the risk to you, and a business that depends on old pipes will struggle to stay profitable through the outages that follow. Put the same question to your critical suppliers, because their provider's network sits in your supply chain too.

Your modernization portfolio has the same meeting points

Enterprise technology teams buy cloud applications, security monitoring, network circuits and compute clusters on separate budgets and separate timelines, and each team can report a healthy project. The FAA's own numbers were moving in the right direction eight days before the ground stop: 69 percent of copper replaced, surface awareness systems at 143 towers, and 226 internet protocol voice switches installed (Department of Transportation, 2026).

Those percentages count what has been replaced. The risk on any given day sits in the share that remains and in the junctions where a new device talks to an old one.

The administration raised its air traffic control funding request to $50 billion late Friday (CBS News, 2026).

More money can shorten the period in which two generations run together, and the junctions stay in service until the last legacy device is retired. Five checks apply to your next cutover, whatever its budget:

• Map each planned swap to the devices one hop away, and record the age and support status of each neighbor.
• Send any device that can pass traffic to the whole network through full change review, whatever the ticket type says.
• Set rate limits at the boundary between new and old equipment before the new device goes live.
• Ask each network provider and each critical supplier for the age of the equipment in your path and the date it is due for replacement.
• Ask your contractor who watches the legacy side during the cutover and how fast that person can isolate one device.