FAA Picks Palantir, Thales, and Airspace Intelligence to Build AI That Predicts Flight Conflicts Two Hours Out

The Federal Aviation Administration is quietly developing an AI-powered air traffic management tool that would let controllers deconflict flight paths up to two hours before a collision risk emerges, according to reporting from The Air Current published April 17, 2026. The program is called Strategic Management of Airspace Routing Trajectories, or SMART. Three companies are competing to build it: Palantir Technologies, Thales, and Airspace Intelligence.

FAA Administrator Bryan Bedford, who was confirmed by the Senate in July 2025, is driving the effort personally. Transportation Secretary Sean Duffy first acknowledged the project publicly at a Semafor event on April 17, saying controllers would get a notice to adjust a flight path “an hour and a half or two hours before the conflict even happens.” That is a massive jump from today’s roughly 15-minute planning window. DOT and FAA officials are expected to share more details at a press event on April 21.

SMART Would Shift Air Traffic Control From Reactive to Predictive

SMART is designed to model the entire National Airspace System in four dimensions and flag schedule conflicts before an aircraft even leaves the gate, according to The Air Current. Today’s system is human-centric and reactive. Controllers separate aircraft in the moments before a conflict would occur. SMART pushes that decision window back by an order of magnitude, letting controllers nudge a flight path by a few miles hours before any collision risk materializes.

The FAA told Bloomberg the project sits inside a broader modernization push that has already secured $12.5 billion in congressional funding, with roughly $20 billion more needed to finish the overhaul. Per The Next Web, the total program budget is around $32.5 billion and includes replacing 612 aging radar systems and hiring 1,200 new controllers in fiscal 2026.

Three Very Different Bidders Are Fighting for the Contract

Palantir brings the deepest federal relationships of the three. The company signed a 10-year, $10 billion enterprise contract with the U.S. Army in August 2025 and has expanded aggressively into the drone sector through investments in Shield AI and partnerships with Red Cat and Anduril. Thales is a French aerospace giant with decades of air traffic management experience across Europe and Asia. Airspace Intelligence is the smallest and most interesting of the three. The Silicon Valley startup built Flyways AI, the platform Alaska Airlines has used since 2021 to optimize routes and cut fuel burn. Flyways reportedly found optimization opportunities on 55 percent of Alaska’s flights and saved the airline 1.2 million gallons of fuel in 2023.

In short: Palantir has the government access. Thales has the ATM heritage. ASI has the only real-world deployment of AI-driven flight path optimization at a major U.S. carrier.

SMART Is the Infrastructure the Drone Industry Has Been Waiting For

SMART is a manned-aviation tool, but the downstream effect on drones is enormous. Predictive flight path management at the national level is exactly the kind of infrastructure that makes routine BVLOS operations defensible at scale. The FAA’s Part 108 BVLOS rule is due as a final rule by February 1, 2026, per the June 2025 executive order, and one of the unresolved questions is how drone flight plans will interact with the manned traffic layer. A predictive ATC system that already knows where every airliner will be two hours from now is a much easier system to hand a drone corridor request.

Separately, NASA’s work on Unmanned Aircraft System Traffic Management (UTM) has always assumed an eventual handshake with the manned ATC layer. Whichever vendor wins SMART becomes the default counterparty for UTM-to-ATC integration. That is the architecture decision that will shape U.S. drone delivery, advanced air mobility, and eVTOL operations for the next decade.

DroneXL’s Take

I’ve been covering FAA modernization since founding DroneXL, and the pattern is depressingly consistent: big program, big budget, slow delivery. NextGen took two decades and is still not finished. What makes SMART different is that two of the three bidders already have working software. Palantir’s Foundry platform runs production workloads for the Army. ASI’s Flyways has been nudging Alaska Airlines dispatchers for five years. The FAA isn’t asking these companies to invent the technology. It’s asking them to adapt it.

That matters for the drone industry for one reason. The whole BVLOS integration story (Part 108, UTM, low-altitude corridors, eVTOL) has been blocked by the manned side of the house. ATC cannot absorb thousands of drone flight plans a day on a system that sees 15 minutes ahead. A system that sees two hours ahead can. SMART is, functionally, the missing infrastructure layer between drones and everyone else sharing the sky.

Here is my call. By the end of Q3 2026, the FAA will announce a single prime contractor for SMART, not a split award, and it will be either Palantir or ASI. Thales is the establishment pick and the safest on paper, but Bedford’s pattern so far has been to break from the traditional FAA procurement playbook. Whoever wins will also, within 18 months of that award, be handed the UTM-to-ATC integration scope. One vendor, one airspace, one throat to choke. That is the bet this administration is making, and for once, it might be the right one.

DroneXL uses automated tools to support research and source retrieval. All reporting and editorial perspectives are by Haye Kesteloo.