Federal Aviation Administration tests AI system to reduce flight delays

The Federal Aviation Administration is betting that artificial intelligence can solve one of modern travel’s most persistent headaches: the cascading flight delay. The agency’s new program, called Strategic Management of Airspace Routing Trajectories, or SMART, aims to predict air traffic congestion and potential conflicts up to two hours before they materialize.

The initiative is part of a $12.5 billion modernization push for the nation’s air traffic control infrastructure, and it has three vendors competing to build the first operational prototypes by late 2026.

How SMART actually works

The AI system is designed to spot potential conflicts 1.5 to 2 hours in advance, giving controllers enough lead time to adjust flight paths and schedules proactively rather than reactively.

Major hubs like Atlanta and Chicago have struggled with only 20-25% on-time performance, with delays exacerbated by staffing shortages and reliance on outdated technology. SMART identifies where pressure points are likely to develop and suggests adjustments before congestion begins compounding.

Transportation Secretary Sean Duffy has highlighted the system’s ability to notify controllers of potential conflicts well before they become operational emergencies. FAA Administrator Bryan Bedford has framed the project as a cornerstone of a broader redesign of how American airspace operates.

The vendor lineup and what’s at stake

The three vendors tasked with building competing prototypes are Palantir Technologies, Thales, and Air Space Intelligence.

Palantir is known primarily for its data analytics work with defense and intelligence agencies. Thales, the French multinational, already provides air traffic management systems in multiple countries. Air Space Intelligence is a San Francisco-based startup that focuses exclusively on AI-driven flight routing and already works with several major airlines.

The FAA has requested an additional $20 billion for future upgrades beyond the $12.5 billion already allocated for ATC modernization, signaling a sustained, multi-decade investment in overhauling how American skies are managed.

One critical detail worth noting: the FAA has been explicit that human controllers will retain decision-making authority. The AI suggests; humans decide. The system’s effectiveness will depend not just on the quality of its predictions but on how seamlessly those predictions integrate into existing controller workflows.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.