FAA launches SMART AI system to predict and cut U.S. flight delays

The Federal Aviation Administration has begun limited use of a new AI-supported air traffic management system designed to predict congestion and emerging conflicts before they cascade into widespread flight delays, according to published FAA materials and recent coverage of the rollout.

What the FAA just switched on and where it is starting

Publicly available FAA information shows the tool is called Strategic Management of Airspace, Routes, and Trajectories, or SMART. The agency has described it as an AI-supported capability that synthesizes airline schedules, weather and airport capacity signals, airspace conditions, and operational constraints into a clearer picture of where bottlenecks may form.

Published coverage indicates the FAA began using SMART in a limited mode on September 21, 2026, initially focused on the airspace around Washington, D.C. Early implementation has been tied to operations around the region’s major airports: Reagan Washington National (DCA), Washington Dulles International (IAD), and Baltimore-Washington International Thurgood Marshall (BWI).

The FAA has framed the early phase as a gradual expansion rather than an overnight switch. Recent reporting has also linked the timing to a period of heightened attention on U.S. air traffic reliability, including weather-driven disruptions and equipment-related issues that can strain already tight schedules.

How SMART aims to prevent delays from spreading

In public descriptions, SMART is positioned as an anticipatory planning aid for traffic flow managers, focusing on earlier detection of schedule conflicts and weather impacts. Instead of reacting only after a backlog forms, the system is designed to flag developing constraints and potential conflicts earlier, when reroutes, revised flow strategies, or adjusted demand programs may be less disruptive.

The FAA’s national flow operation is centered at the Air Traffic Control System Command Center, which coordinates initiatives used to manage demand and capacity across the National Airspace System. The agency’s operational playbook already includes tools like ground delay programs and other traffic management initiatives; SMART is being introduced as a new layer intended to sharpen predictions and improve system-wide coordination.

For travelers, that distinction matters because many of the worst travel days are not driven by a single late departure, but by network effects: aircraft and crews out of position, arrival and departure banks collapsing, and downstream cancellations. A tool that helps identify where trouble is likely to form, earlier in the day, could in theory reduce the number of flights that get trapped in that domino effect.

The bigger modernization effort behind the AI rollout

SMART is not described as a standalone experiment. FAA materials indicate it is paired with a broader technology replacement called Flow Management Data and Services (FMDS), a program intended to replace the FAA’s legacy Traffic Flow Management System (TFMS). The FAA has described FMDS as the new technological backbone for its command-center flow operations, reflecting a shift away from aging architecture.

Published FAA announcements indicate the work is being delivered via a long-term contract with Air Space Intelligence, a Boston-based company. Recent coverage has put the contract at $875 million over 12 years, and has described the combined effort as linking a modern data backbone (FMDS) with predictive planning capabilities (SMART).

The modernization also fits within the FAA’s longer-running NextGen roadmap. FAA NextGen materials describe a move from tactical, reactive control toward more strategic, integrated traffic management. That includes time-based tools used in multiple domains, such as Traffic Flow Management System functions, Time Based Flow Management, and Terminal Flight Data Manager, which aim to improve predictability from gate movement through en route metering.

What success could look like and what remains uncertain

Even with an AI-supported system in place, the operational goal is not to eliminate delays entirely. The U.S. network still has hard constraints that no software can erase, including thunderstorms, runway closures, volume surges, and staffing limitations. The practical benchmark for SMART will likely be whether it helps decision-makers act earlier and coordinate better across airlines, airports, and FAA facilities when those constraints appear.

Another key question is how the new capability will integrate into the established collaborative decision-making ecosystem that already exists between the FAA and industry participants. FAA materials describing FMDS reference collaborative decision making as a structured partnership across government, airlines, general aviation, and other stakeholders to address air traffic flow challenges. If SMART’s predictions are trusted and well integrated, they could improve the timeliness and consistency of those collaborative choices.

Published descriptions also emphasize that new automation should complement, not replace, the professional judgment of air traffic controllers. For travelers, that framing suggests the FAA is positioning SMART as decision support rather than an autonomous traffic director, which may influence how quickly it is adopted across regions and how cautious early deployments remain.

For now, the most concrete near-term detail is geographic: the first operational footprint around Washington, D.C., beginning September 21, 2026, with expansion planned over time. How quickly SMART spreads to other congested corridors, and whether measurable improvements follow at the airport level, will be closely watched as the system moves beyond its initial limited mode.