The FAA is betting $12.5 billion on AI to fix America's broken air traffic system
The FAA is building an AI-powered air traffic management system called SMART — Strategic Management of Airspace Routing Trajectories — and Congress has already allocated $12.5 billion toward the effort. The agency's stated goal is to extend flight conflict prediction from a narrow 15-minute window to a full two hours ahead, giving controllers time to reroute planes before problems snowball. Transportation Secretary Sean Duffy confirmed work is underway, with a target operational date inside 2026.
Three bidders, one big contract
Three vendors are competing for the work, per The Air Current: Palantir, French defense and ATM giant Thales SA, and startup Air Space Intelligence. Each brings a different pedigree. Palantir has prior FAA data contracts — including a sole-source justification issued in April — and deep government relationships. Thales has 85 years of air traffic management history and a strong European footprint. Air Space Intelligence's Flyways platform is already handling routing decisions for roughly 40% of US air traffic through an Alaska Airlines deployment active since 2021, making it the only bidder with real-world, in-production US experience.
No single prime contractor has been named yet. The compressed 2026 timeline — which some sources describe as a demonstration rather than full production deployment — will likely define which vendor has the structural advantage.
Pre-planning, not autopilot
SMART is designed as a pre-planning tool, not a real-time separation system. The idea is to identify bottlenecks days or weeks out and suggest adjustments — shifting a departure by five or ten minutes before the schedule is even published. Human controllers keep final authority over every decision. The National Air Traffic Controllers Association (NATCA) has backed the concept on those terms.
The push is partly driven by a genuine crisis. The FAA is still running infrastructure that relies on floppy disks and paper strips, a LaGuardia ground-stop incident earlier this year exposed the fragility of the current setup, and controller retirements are outpacing new hires. The pressure to deploy something quickly is real.
History says be skeptical
The FAA's last big modernization effort, NextGen, cost $36 billion over roughly two decades and delivered only 16% of its promised benefits. SMART's total program budget is pegged at $32.5 billion. Experts also flag a specific technical concern: modern predictive AI models can produce confident-sounding but wrong outputs — so-called hallucinations — which is an uncomfortable characteristic for safety-critical infrastructure where a miscalculation by a few minutes or miles has real consequences.
Whether SMART avoids NextGen's fate depends on execution, contractor selection, and whether the 2026 deadline reflects genuine readiness or political pressure. The competitive analysis at The Next Web lays out exactly what each bidder stands to gain — and lose — if the contract goes the wrong way.