The FAA Runs 4,600 Facilities on a Network Built for Phone Calls. The Deadline Is 2028.

Nothing about a ground stop looks technological from seat 4A. The jet bridge stays attached, the crew stops giving updates, and eventually a gate agent says the word Washington, which passengers have learned to translate as an unspecified delay of unspecified length. What is rarely said out loud is that a meaningful share of those pauses trace back to the physical layer underneath the whole system, which is to say the wires, switches and radar heads that carry a controller's voice and a target's position from one building to another.

The federal government is now spending more than $31 billion to replace that layer, and it has given itself a date. The Brand New Air Traffic Control program carries a December 2028 operational deadline, and the telecommunications backbone that everything else depends on is supposed to be finished by the end of the same year.

On August 13, 2026, the Federal Aviation Administration named AT&T lead project manager and prime contractor on the FAA Enterprise Network Services program, opening with a $74.3 million task order against what is expected to become a multibillion-dollar award running as long as fifteen years, as first reported by Washington Technology. The scope covers program management, engineering, network design and transition planning for a network that connects roughly 4,600 facilities running continuously. Verizon, selected as FENS prime in the spring of 2023, moves into a supporting role providing core network functions.

What is actually being torn out

The technical substance of the change is one sentence long and it explains most of the schedule pressure. The FAA is migrating from time division multiplexing, a circuit-switched approach designed for voice telephony, to an Internet Protocol network. Time division multiplexing reserves a fixed slice of capacity for each connection whether or not anything is moving through it, which was a sound engineering choice when the traffic was analog voice and the alternative was nothing.

It is a poor fit for surveillance data, digital flight strips, weather feeds and the machine-to-machine traffic that a modern airspace system generates, and it is served in many places by copper. Transportation Secretary Sean Duffy told reporters in April 2026 that almost half of the 1960s-era copper in the air traffic telecommunications system had been replaced, alongside digital voice switches installed at 40 locations, surface awareness systems at 54 airports, and 17 towers converted to electronic flight strips, according to Nextgov reporting on his remarks.

The FAA outlined the August task order in terms of that clock. "Authorizing this initial work now helps maintain the schedule to complete the program by the end of 2028 and ensure a smooth transition from the current network while the FAA and AT&T negotiate and finalize the long-term FENS contract," the agency said.

The radar side has its own ledger

Running in parallel is the Facility Replacement and Radar Modernization effort, which the agency scopes at 377 critical radar systems and more than 20 air traffic control facilities. The funding structure proposed for it is instructive about how the agency expects the work to unfold: roughly $1 billion a year in the first two fiscal years, then $2 billion a year for three more, a shape that reflects design and procurement giving way to construction and installation.

Radar heads are the least glamorous item on any modernization list and the hardest to schedule, because a surveillance asset cannot simply be switched off for eighteen months while a replacement is built beside it. Coverage has to be maintained through the transition, which typically means running both systems in parallel and accepting the cost of doing so.

Why the deadline is the interesting number

Duffy has been explicit that the timeline is political as well as operational, telling lawmakers that "for the most part, we're on track to have this project completed before President Trump leaves office," and pressing for money beyond the $12.5 billion Congress has already appropriated against a program the department prices above $31 billion. FAA Administrator Bryan Bedford described inheriting a 680-page modernization document after his confirmation and deciding with Duffy to scrap it because it "represented more of the same."

Aviation infrastructure programs have a long history of slipping past their announced completion dates, and the useful posture for a traveler is neither cynicism nor confidence but attention to intermediate markers. The copper replacement percentage, the count of towers running electronic flight strips, and the number of digital voice switches installed are all published figures that move independently of any speech, and they are the ones that will indicate whether the December 2028 date holds.

What a passenger can reasonably expect to feel

Very little of this will be visible in a terminal, and that is worth stating plainly because modernization announcements tend to promise experiences they do not deliver. An IP backbone does not shorten a security line, add a gate, or create a controller. What it changes is the failure behavior of the system underneath, and failure behavior is where the passenger experience actually lives.

A circuit-switched network degrades in a particular way, which is that when a link goes down there is often no alternate path and the affected facility loses a capability until the link is restored. A properly designed IP network reroutes, which converts a hard outage into a slower connection. If the program works, the visible result is fewer of the single-point failures that cascade into a national ground stop on a Tuesday afternoon, and more of the small unnoticed degradations that never make the news.

Digital flight strips, surface awareness tooling and the broader push toward automated conflict detection all assume bandwidth that a time division multiplexing circuit cannot supply, which is why the network contract precedes the applications rather than following them. That sequencing also determines how quickly the biometric and automation layers now spreading through terminals can extend into the operational side of the airfield, a subject we examined in our look at where artificial intelligence is genuinely changing aviation.

The stress test already happened

The summer of 2026 delivered an unusually clean natural experiment. A World Cup spread across North American host cities pushed record volumes through airports that had not been rebuilt for them, and the system's behavior under that load told a more honest story about capacity than any planning document, which is what we tracked in our reporting on how the United States aviation network handled the tournament.

What that summer demonstrated is that the binding constraint is rarely runway pavement. It is staffing, it is the sequencing tools controllers have available, and it is the reliability of the links between facilities when several regions are simultaneously saturated. Two of those three are what FENS and the radar program are meant to address.

The practical read for anyone who flies a lot

Book the first departure bank when the itinerary matters, because equipment and staffing failures accumulate through the day and the morning aircraft is already at the gate. Treat connections through facilities that have not yet been modernized with more schedule padding than the airline's own minimum connection time suggests. And read the intermediate metrics rather than the announcements, because a program of this size will be judged in 2029 on how many of the 377 radars were actually replaced.

The copper being pulled out of these buildings was strung when the busiest airport in the country handled a fraction of today's traffic. Replacing it will not make any individual flight faster. It determines how the system behaves on the day something breaks, which, for anyone who has spent an afternoon at a gate listening for the word Washington, is the part that has always mattered.