They borrowed a thrust-vectoring nozzle from a 1990s experimental F-16 — retrieved, as one official put it, Indiana Jones style — bolted it onto a GE F110 turbofan, and pointed it straight down. That is how you make a fighter jet take off like a helicopter. Shield AI’s X-BAT autonomous combat aircraft showed up at the Sea-Air-Space 2026 exposition near Washington looking nothing like the design the company unveiled last year. The cranked-kite planform is gone. In its place: a dramatically swept arrowhead with a straight leading edge, a half-scale model bristling with design changes driven by real wind-tunnel data, and a timeline that says free-flight VTOL testing before the end of 2026. If it works, the X-BAT will be the first operational combat aircraft that can take off vertically, fly 2,000 nautical miles, drop bombs from internal weapons bays, and land tail-first on a trailer — with no pilot and no runway.

The Redesign

Chief designer Armor Harris was blunt about the changes. The original cranked-kite planform looked elegant on paper, but wind-tunnel data demanded something sharper. The new shape is an arrowhead — more dramatic sweep, a straight leading edge, and an overall profile that prioritises speed, stealth, and aerodynamic efficiency at the expense of the gentler curves of the earlier concept. The half-scale model displayed at Sea-Air-Space showed details that matter: an auxiliary air inlet concealed on the dorsal surface (used only during vertical flight), a custom launch-recovery trailer with an integrated blast deflector, and the AVEN nozzle that makes the entire concept possible. The AVEN — Axisymmetric Vectoring Exhaust Nozzle — has a strange origin story. It was developed in the 1990s for a thrust-vectoring F-16 test programme at Edwards Air Force Base. The original nozzle sat in storage for decades. Shield AI and GE retrieved it, mated it to the F110, and found that it worked remarkably well. New nozzles are now being manufactured with updated digital control systems.

How Vertical Flight Works

The X-BAT’s VTOL sequence is unlike anything currently flying. For takeoff, the F110 enters full afterburner to achieve the thrust-to-weight ratio needed to lift the aircraft vertically off its trailer. The blast deflector built into the trailer directs exhaust away from the engine intake. Landing is the harder problem. The X-BAT approaches its recovery trailer from the side at military power — no afterburner — leaning slightly into the incoming airstream to keep clean air flowing into the intake. It makes contact with the trailer and locks into a securing mechanism. Shield AI officials are refreshingly honest about the risks: they expect to lose at least one prototype during testing, calling it a hardware-rich approach to development. Between takeoff and landing, the X-BAT flies like a conventional combat aircraft. The combat radius is 1,000 nautical miles — comparable to a manned fighter — with a maximum range of over 2,000 nautical miles. Two internal weapons bays can carry any munition that fits inside an F-35, including 2,000-pound-class weapons. The ceiling is approximately 50,000 feet.

The Timeline

Shield AI is running a test programme that would make traditional defence contractors nervous. The current phase has the F110 engine on GE test stands, being adapted for vertical thrust operations. Next comes propulsion integration into the prototype airframe, followed by horizontal and vertical testing while the aircraft remains attached to its launch-recovery trailer. After that: tethered testing using a large crane to hold the aircraft vertically while the engine runs. Then free-flight VTOL operations — targeted before the end of 2026. The structural pathfinder airframe is already in fabrication, and a prototype is planned for construction at Shield AI’s facility in Frisco, Texas, this summer. Mission-capable status is projected for 2028. Production capacity is planned at 150 units per year on a single shift.

Why It Matters

The X-BAT eliminates the runway. In a Pacific conflict, that matters more than any other single capability. China’s missile force can destroy every fixed airfield within the first and second island chains. Every concrete runway is a known target with a known GPS coordinate. Remove the runway requirement, and you remove the most critical vulnerability in American air power. GE’s Steve Russell summed it up simply: put all those things together and put it into a platform that presents a unique dilemma for potential adversaries. Shield AI’s cost model reinforces the point. At roughly one-tenth the lifecycle cost of a manned equivalent, a force of X-BATs could be deployed from dispersed locations — forest clearings, highway strips, ship decks — with no pilot training pipeline, no dedicated tanker support, and no fixed infrastructure to destroy. The X-BAT is not the only autonomous combat aircraft in development. The Air Force’s Collaborative Combat Aircraft programme is pursuing similar concepts. But the X-BAT is the only one that takes off and lands vertically. If Shield AI can make that work — and that remains a substantial if — it will have built something genuinely new.

Sources: The War Zone, Air & Space Forces Magazine, Army Recognition, Shield AI