Equipment can fail.
Passenger demand fluctuates.
An airport that looked comfortable at 9 a.m. can become severely constrained by noon.
The FAA has long been working toward a more predictive model through its Trajectory Based Operations, or TBO, initiative.
The agency describes TBO as a way to improve strategic planning of aircraft flows and reduce imbalances between demand and available capacity. Instead of considering an aircraft's position only as a point on a map, TBO considers a trajectory in four dimensions: latitude, longitude, altitude and time.
SMART fits naturally into that broader philosophy.
If the FAA knows where aircraft are expected to be and when they are expected to be there, it can begin asking what the entire network will look like several hours into the future.
That creates a critical opportunity.
A problem is much easier to solve before it becomes a problem.
Moving one departure by a few minutes may be enough to prevent a major traffic buildup.
Rerouting several aircraft around a developing weather system may be easier before hundreds more flights are committed to the same airspace.
Adjusting traffic flows early may prevent the need for more disruptive restrictions later.
The purpose is not to eliminate every delay.
No system can do that.
The purpose is to make the overall system more predictable and less reactive.
The Human Being Remains in Charge
The phrase “AI controlling air traffic” can easily create the wrong impression.
That is not what SMART is designed to do.
The FAA's own documentation explicitly says that air-traffic controllers remain responsible for keeping aircraft safely separated. SMART is intended to help them perform that responsibility by giving them better information about what is likely to happen.
The distinction between decision support and control matters enormously in aviation.
A controller can receive a recommendation from a computer and decide whether that recommendation makes sense.
The system can predict that congestion is likely to develop.
The human operator can consider the prediction alongside information the software may not fully capture.
The system can suggest an alternate route.
The people responsible for the operation can determine whether that route is appropriate.
This structure also provides an important layer of human judgment.
Artificial intelligence is very good at processing large quantities of information and identifying patterns.
But aviation involves unusual events, incomplete information and rapidly changing circumstances.
The FAA therefore describes SMART as an enhancement to existing air-traffic-management systems rather than a replacement for them.
Why Controllers Matter
Air-traffic controllers already operate in an environment where seconds can matter.
They have to maintain separation between aircraft while simultaneously managing departures, arrivals, weather, runway configurations and changing traffic patterns.
The FAA has said that controllers currently manage multiple screens of information and that the broader modernization effort is intended to bring more of that information together.
FAA Administrator Bryan Bedford described another component of the modernization effort, the Flight Management Data System, as a way to create a single real-time picture of the airspace. SMART then uses that information to predict congestion and conflicts earlier and recommend adjustments to routes and timing.
That suggests SMART should be understood as one component of a much larger modernization program.
The ultimate objective is not simply to introduce AI.
It is to modernize how information moves through the National Airspace System.
Why Airlines Wanted a Slower Start
The FAA originally envisioned a much broader deployment.
But airlines urged a more cautious approach.
The concern was not necessarily that artificial intelligence should have no role in aviation.
The concern was how quickly such a system should be introduced into a complex operational environment.
According to reporting around the launch, airlines pushed for a “crawl, walk, run” approach, beginning with a limited deployment before expanding the system nationwide.
That caution makes practical sense.
Airlines operate their own highly interconnected systems.
A change to departure timing can affect aircraft rotations.
A route adjustment can influence fuel planning.
A delay can affect crews.
A change in arrival time can affect gates, baggage handling and connecting passengers.
In other words, an air-traffic decision rarely affects only one aircraft.
It can have consequences throughout an airline's network.
That means an AI recommendation has to be evaluated not merely on whether it looks logical in isolation, but on how it interacts with thousands of other operational decisions.
The Washington trial gives airlines and the FAA an opportunity to learn those lessons before the technology is expanded.
The $875 Million Question
In June 2026, Air Space Intelligence was awarded a major FAA contract valued at approximately $875 million to modernize elements of the National Airspace System.
The agreement is structured over 12 years.
The company specializes in operational artificial intelligence and software designed to help organizations make decisions in complex, rapidly changing environments. Air Space Intelligence confirmed the FAA contract in June.
The scale of the contract demonstrates how seriously the FAA is taking predictive technology.
This is not a short-term experiment involving a small research team.
It is part of a long-term attempt to change the architecture of U.S. air-traffic management.
That also explains why the initial deployment is being watched so closely.
A system that eventually reaches airports and airspace across the country could influence how airlines plan flights, how traffic managers balance demand and capacity, and how quickly the system responds to disruptions.
The 90-day Washington trial is therefore much more than a technology demonstration.
It is an early operational test of a much larger vision.
What a Successful SMART System Could Mean for Passengers
Passengers are unlikely to see SMART directly.