FAA launches $875 million AI air traffic tool SMART in Washington region

The Federal Aviation Administration has started using artificial intelligence to help manage some of the busiest airspace in the United States. On September 21, 2026, the agency began the first operational deployment of SMART, short for Strategic Management of Airspace, Routes and Trajectories, a cloud-based system that predicts air traffic flows and flags potential conflicts before they turn into delays. The first site is the Washington, DC region, where the tool will advise controllers handling flights at Ronald Reagan Washington National Airport, Washington Dulles International Airport and Baltimore/Washington International Thurgood Marshall Airport.

SMART is not a controller replacement. The FAA says the system works alongside existing air traffic systems, offering what it calls alternative route information and a shared view of conditions so that the agency, airlines and other operators can agree on more efficient routes and departure times. The stated goals are lower fuel burn, better on-time performance and faster recovery when weather or congestion disrupts schedules.

The software comes from Air Space Intelligence, a Boston-based company that won a 12-year contract worth $875 million in June 2026. The award, announced on June 22, 2026, covers two pieces: Flow Management Data and Services, or FMDS, which will replace the FAA's aging Traffic Flow Management System, and SMART, an AI-enabled planning layer inside FMDS. FMDS is slated to become the technical backbone of the FAA's Air Traffic Control System Command Center in Virginia.

The launch follows weeks of anxiety in the airline industry, a limited-scope decision by the FAA, and a broader push by the Department of Transportation to bring AI into air traffic management. Transportation Secretary Sean Duffy has asked Congress for between $6 billion and $10 billion for that integration, on top of $12.5 billion already approved for upgrades to ground-based infrastructure, with a goal of completing the wider overhaul by the end of 2028.

Key Facts

Ars Technica reported on September 18 that an AI tool is set to start advising air traffic controllers in the congested airspace above Washington, DC, as the first step toward a nationwide rollout covering the 29 million square miles of US national airspace overseen by the FAA. The FAA describes SMART as using AI models to predict traffic flows and identify potential conflicts based on airline schedules, weather, airport capacity and airspace conditions.

MLQ.ai reported on September 18 that the FAA awarded Boston-based Air Space Intelligence a 12-year contract valued at $875 million to develop software that forecasts air-traffic congestion and helps officials coordinate flight schedules before aircraft leave the ground. The award was announced on June 22, 2026. Washington Technology reported that 12 companies bid, citing federal procurement data.

The Washington Times reported on September 20 that the cloud-based tool is aimed at enhancing existing traffic management systems, and that it is not meant to replace air traffic controllers or solve the significant shortage of those personnel. The report confirmed that the first deployment covers the three major commercial airports serving the US capital region.

Business Insider reported on September 19 that airline industry officials finally stopped panicking about the soon-to-launch AI tool after the agency narrowed the rollout. Weeks of confusion eased when FAA officials said they would use SMART in just a limited way at first. Administrator Bryan Bedford said the roughly 90-day first phase will focus on testing the system and building airlines' confidence in its predictions. The FAA convened several meetings, including one on September 10 between major airline CEOs and Bedford; the CEOs of United, American, Delta and Southwest attended.

La Voce di New York reported on September 21 that the FAA is beginning operational deployment in the Washington, DC area of a system designed to predict air traffic patterns and reduce disruptions for passengers. According to technical documents reviewed by TechCrunch, SMART is a cloud-based architecture designed to integrate with existing air traffic control systems, and the algorithm identifies potential operational conflicts on the ground and along flight routes, letting officials recommend small adjustments to departure times before congestion becomes disruptive. Today the FAA manages an average of more than 45,000 flights and over 2.9 million passengers each day. According to the Department of Transportation, predictive models could eventually let officials identify and address disruptions as much as 45 days in advance.

Analysis

What this really means is that the FAA is treating AI as a decision-support layer rather than an autopilot for the skies. The agency's own fact sheet says SMART will work alongside existing air traffic systems and that controllers remain responsible for safely separating aircraft. That distinction is the difference between a tool that suggests and a system that commands, and it explains why the first deployment is deliberately small: three airports, one region, and a 90-day test phase before any wider expansion.

The scope decision answered the airline industry's loudest complaint. Before the FAA narrowed the rollout, the biggest worry among airline officials was how extensively the AI would be integrated into day-to-day air traffic management. One industry official told Politico that this was what people were freaking out about. The FAA responded by convening meetings, including one on September 10 between major airline CEOs and Bedford that the CEOs of United, American, Delta and Southwest attended. A fourth airline official described the reception as exceptionally favorable and supportive.

The bigger picture here is that SMART is one piece of a much larger modernization effort that carries more unknowns than anything the FAA has fielded before. Philip Mann of Vector Strategic Consulting, who advised on aviation safety and AI governance and previously worked at the FAA for 17 years, called the limited launch the right call. He wrote that SMART's risk was never any single prediction, but that it is a national-scale system with AI components, and that every cut in scope shrinks the unknowns. That is a fair reading. A prediction a controller can ignore is a very different risk from a prediction that quietly reshapes a traffic flow.

What remains unresolved is governance. Public materials do not provide a detailed breakdown of the $875 million across development, operations, maintenance, options or performance incentives, and they do not spell out how SMART's predictive models will be validated, monitored for performance drift or handled when recommendations conflict with controller judgment. Those gaps are not proof of failure. They are, however, the questions that will decide whether a 90-day test produces durable trust or another round of industry anxiety.

Why It Matters

The stakes are measured in passengers and minutes. The FAA handles more than 45,000 flights and about 2.9 million passengers each day, and it oversees 29 million square miles of airspace. Even small improvements in how departure times are sequenced during bad weather can ripple across the network, reducing fuel burn and helping airlines recover faster. That is the operational case for SMART, and it is why the agency frames the tool as a way to see problems coming and act before they do.

There is also a workforce argument. The decision to accelerate investment addresses a long-standing problem at the agency: a persistent shortage of air traffic controllers. SMART is not meant to solve that shortage, and the FAA has been careful to say so. But a system that predicts congestion and proposes adjustments could ease the cognitive load on controllers who are already stretched, at least at the margins.

Finally, SMART is a test of public trust in AI for safety-critical work. Duffy told CBS News that AI is a tool but that humans are not replaced in how the airspace is managed. If the Washington trial confirms that framing, it strengthens the case for the $6 billion to $10 billion the department wants for AI integration. If it does not, the same trial becomes evidence for caution.

Next Up

Over the next 90 days, the FAA will test SMART in the Washington region and work on building airlines' confidence in its predictions. After that test, the agency hopes to roll the tool out on a wider scale, with a nationwide expansion projected for 2028. Air Space Intelligence expected the two systems, FMDS and SMART, to roll out over 12 to 24 months.

In the background, the FAA selected Peraton in December 2025 as prime integrator for the Brand New Air Traffic Control System program, which includes 13 Phase 1 programs, while Air Space Intelligence holds the separate FMDS award. How those efforts interlock, and how quickly SMART moves beyond three airports, will determine whether the Washington deployment is remembered as a cautious first step or the start of a genuinely different way to manage US skies.