FAA Prepares $875 Million AI System to Help Manage Air Traffic Congestion

The Federal Aviation Administration is preparing to deploy an artificial-intelligence system that would advise air traffic controllers and aviation operators on congestion in the Washington, DC, area. The planned launch could come as soon as September 21, 2026, according to government and industry officials cited in reporting published September 18.

Known as SMART, the system is intended to forecast traffic flows, identify potential conflicts, and give the FAA, airlines, and aircraft operators a shared picture of developing conditions. The Washington deployment would be a limited first step toward a proposed nationwide rollout across the roughly 29 million square miles of US national airspace managed by the FAA.

The agency’s stated goals include reducing fuel consumption, improving flight punctuality, and helping the aviation system recover more quickly from weather disruptions and periods of heavy traffic. The limited initial deployment would allow the FAA to introduce the system in a smaller area before considering a nationwide rollout.

What SMART is designed to do

The FAA describes SMART as an AI-based system that analyzes several factors affecting air traffic, including:

• Airline schedules
• Weather
• Airport capacity
• Airspace conditions
• Expected traffic flows
• Potential conflicts between operational plans

The system would produce forecasts and recommendations rather than directly take control of aircraft or replace air traffic controllers. Those recommendations are intended to help aviation stakeholders coordinate decisions such as preferred routes and departure times.

SMART’s proposed role is a shared planning and decision-support tool. The FAA reportedly told airlines that the initial system would not alter existing procedures for controllers or carriers. Instead, it would generate alternative recommendations that operators could evaluate within the current system.

SMART’s proposed function	What the available material does not indicate
Forecast air traffic flows	That it will autonomously control aircraft
Identify possible conflicts	That it will replace air traffic controllers
Help coordinate routes and departure times	That it will introduce new operating procedures at launch
Provide a common view of conditions	That its recommendations must automatically be followed

Why begin in the Washington area?

The system is expected to debut around the three major airports in the Washington, DC, area. Starting there would give the FAA a limited setting in which to evaluate SMART before considering a nationwide expansion.

A smaller deployment could make it easier to:

• Compare the system’s forecasts with actual traffic conditions
• Identify situations in which recommendations are unclear or impractical
• Monitor how controllers and airlines use the information
• Correct problems before expanding the system

Philip Mann, a principal consultant at Vector Strategic Consulting who previously held several FAA roles over 17 years, described the limited rollout as a sensible approach. He said the concern is not necessarily one isolated prediction but the broader uncertainty created when AI components are introduced into a system operating at national scale.

Limiting the system’s initial scope also limits the number of unknowns that must be managed at once.

What the FAA hopes to improve

SMART is intended to help the FAA, airlines, and aircraft operators coordinate responses to changing traffic conditions.

A weather event or capacity restriction can create a chain of delays. Airlines may adjust schedules, airports may face too many arrivals at once, and controllers may need to manage additional holding or rerouting. A shared forecasting tool could help stakeholders agree earlier on routes and departure times.

The expected benefits include:

Lower fuel use

More efficient routes and fewer unnecessary delays could reduce the time aircraft spend holding, taxiing, or taking indirect paths. The available material describes this as a goal of SMART, not as a demonstrated result of the planned deployment.

Better schedule performance

Earlier coordination of departure times and routes could reduce some avoidable delays. That would not eliminate disruptions caused by severe weather, mechanical problems, staffing constraints, or airport limitations.

Faster recovery after disruptions

When congestion or bad weather affects the network, a common operational picture could help the FAA and airlines agree on a recovery plan more quickly. The value would depend on the quality and timeliness of the system’s forecasts and on whether operators find its recommendations usable.

The main risks are broader than prediction accuracy

An AI system can produce a technically reasonable forecast and still create operational problems. Aviation decisions depend on timing, communication, accountability, and how people interpret recommendations under pressure.

Questions for the FAA as SMART moves into operational use include:

• How will the system signal uncertainty?
• What happens when its recommendation conflicts with a controller’s judgment?
• How will users distinguish a high-confidence forecast from a fragile one?
• How quickly can the system respond when weather or airport conditions change?
• How will the FAA monitor errors, unexpected behavior, and unusual cases?
• What safeguards will prevent users from treating the system’s output as automatically correct?

The available material does not provide detailed answers to those questions. It does indicate that the FAA is presenting the first deployment as advisory and limited in scope.

Why airline confusion matters

The airline industry was reportedly uncertain for weeks about the FAA’s plans. That confusion eased after the agency clarified that SMART would not change procedures for airlines or controllers at launch.

A tool designed to coordinate the aviation system depends on users understanding:

• What information it provides
• When recommendations will be issued
• Whether users are expected to act on them
• Who has final authority
• How disagreements will be resolved
• What changes, if any, will follow a recommendation

If one group treats SMART as a forecast, another treats it as an instruction, and a third treats it as an optional planning aid, the system could create inconsistency rather than reduce it.

That question would become more significant if the FAA expands SMART beyond the Washington area. A nationwide system would involve a far larger range of airports, weather patterns, traffic flows, and operating practices.

A national rollout would be a much larger test

The planned expansion would cover the FAA-managed national airspace system, where conditions vary widely by region and time of day. A model that performs well around Washington may not behave equally well in every setting.

A nationwide deployment would have to account for differences in:

• Airport size and capacity
• Traffic density
• Regional weather patterns
• Airspace structure
• Airline schedules
• Local operating practices
• Disruption types and recovery options

A successful regional launch would provide evidence about the system’s usefulness, reliability, and integration with human decision-making. It would not, by itself, prove that SMART is ready for every part of the national network.

The FAA’s AI tool is described as an $875 million system. The available source material does not break down that figure or explain how much is allocated to software, infrastructure, implementation, oversight, training, or future expansion.

What to watch after launch

Because the initial deployment is expected rather than confirmed in the available material, the first question is whether SMART begins operating on the reported timetable. After that, evaluation should look beyond whether the system can make forecasts.

Useful indicators would include:

1. Operational reliability: Does the system remain available and responsive during busy periods and disruptions?
2. Forecast quality: Are its predictions accurate enough to improve planning, and does it communicate uncertainty clearly?
3. Human use: Do controllers and airlines understand the recommendations and use them consistently?
4. Safety and accountability: Is responsibility clear when a recommendation is rejected, modified, or followed?
5. Measurable outcomes: Does the system contribute to lower fuel burn, better punctuality, or faster recovery rather than merely producing more information?
6. Scalability: Does performance hold as the system is exposed to more airports, users, and operating conditions?

Those questions should be answered through documented operational experience rather than through the size of the investment or the use of the term “AI.”

What SMART does—and does not—promise

SMART is being presented as a tool to help aviation professionals see congestion developing earlier and coordinate a response. It is not described as a replacement for controllers, an autonomous traffic-management system, or an immediate overhaul of airline procedures.

The Washington-area launch, if it proceeds as reported, would test whether AI-supported forecasting can improve coordination in a complex airspace environment. Its broader success will depend not only on prediction accuracy but also on transparency, human oversight, operator trust, and the FAA’s ability to learn from a limited deployment before attempting a national expansion.

Sources

• Ars Technica, “FAA tees up $875M AI tool to help manage air traffic congestion,” September 18, 2026: https://arstechnica.com/ai/2026/09/faa-tees-up-875m-ai-tool-to-help-manage-air-traffic-congestion/