Old design, new era: what to expect from X-BAT
In July 1957, test pilot Peter Girard climbed out of the cockpit of the Ryan X-13 Vertijet along a rope ladder. The machine hung tail-down on a steel mast, hooked to a horizontal cable, like an insect stuck to a wall. A year later, the program was shut down: the pilot couldn't land the jet fighter properly on its tail, twisting his neck over his shoulder. Another 67 years later, California-based Shield AI demonstrated X-BAT, a jet aircraft of exactly the same design. Only without a human in the cockpit, and therefore, according to the developers, it should work this time.
Ryan X-13 Vertijet (#54-1619) with tail assembly, 1955
Pogo, Salmon, and Vertijet: How the US Navy Tried to Land a Fighter Jet on Its Tail
In 1951, the US Navy announced a competition for a fighter capable of vertical takeoff from small escort ships that lacked, and were not intended to have, a flight deck. The idea was simple: the aircraft would sit on its four tail fins, spin its propellers, lift off the deck, land on its belly in the air, and then fly like a regular aircraft. On landing, the reverse was true: the aircraft would raise its nose, slow down, and slowly descend onto its tail.
By 1954, two competing aircraft had taken to the air. The Convair XFY-1 Pogo with turboprop Allison T40 with a power of about 5500 horsepower and coaxial propellers, it managed to complete a full cycle: vertical takeoff, transition to horizontal flight and back. Lockheed XFV-1 Salmon with the same engine it flew only horizontally, with a temporary wheeled chassis: they never dared to take off vertically.
A year later, the Ryan company joined the program with its X-13. VertijetIt was already pure jet tailsitter. Engine Rolls-Royce Avon, with a thrust of approximately 4500 kgf, and landing not on its tail struts, but on a special mast-trolley: the machine was hooked onto a horizontal cable. On July 30, 1957, the Vertijet made a spectacular landing on this mast in front of the Pentagon, and the newspapers were delighted.
The second prototype of the Ryan X-13 Vertijet
The excitement wore off quickly. The pilot, in his reclining seat, couldn't see the landing point: he was looking up, but he needed to land down. He estimated the altitude above the platform by eye, using the ground controller's hand signals to indicate "lower and left." A slight altitude error, and the hook missed the cable. By 1958, the project was abandoned, all three projects. The diagnosis wasn't aerodynamic: the design was, in principle, operational. It was just that there was no use for a human in it.
X-BAT: Same Silhouette, Different Era
In the spring of 2026, Shield AI published the first open data on X-BATThe first vertical flight is promised for the end of 2026, with combat readiness declared for 2029. These timelines are still corporate promises, not contractual obligations, and should be treated accordingly.
According to Shield AI, the vehicle is 7,9 m long, has a wingspan of 11,9 m, and is about 1,4 m high at the tail. It takes up one space in the hangar. F-35, can accommodate three X-BATs, and that's the key figure behind the whole thing. There's no chassis: launch and reception are from a mobile platform, essentially a descendant of the very same Vertijet mast, only now with automation and hydraulics instead of a ground operator with flags.
Engine - General Electric F110, the same one that stands on the later ones F-16 и F-15The choice is understandable: millions of operating hours, clear logistics, and, according to GE Aerospace, approximately 3400 engines of this family will be in operation worldwide by 2024. On the nozzle— Aven (Axisymmetric Vectoring Exhaust Nozzle) with thrust vectoring. The technology is not new: AVEN was tested on experimental F-16s in the 1990s as part of the program F-16 MATV (Multi-Axis Thrust Vectoring) and parallel tests on the research board F-16 VISTABack then, she was looking for a use and couldn't find one. Now she has: it's the thrust vector that makes it possible to land a jet aircraft tail-down with precision. The Vertijet physically lacked this.
Here is the answer to the natural question “why not a lifting fan, as in F-35B"Lifting fan and rotating nozzles Harrier — that's hundreds of kilograms and cubic meters of volume, permanently subtracted from the combat load and fuel. The Tailsitter uses its main engine for both takeoff and cruise; it has no separate lift system at all. The price is landing tail-first on the platform, but that's precisely what the automated system handles.
The internal weapons bay, according to the same Shield AI data, is comparable to the F-35 bay: four missiles AIM-120 AMRAAM Or about 900 kg of other payload in a stealthy configuration. The external components carry the same amount, but without any claims to low observability. The claimed range is over 2000 nautical miles, approximately 3700 km; the ceiling is 50,000 feet, approximately 15 km. These figures are reliable, but there is no way to verify them yet: the aircraft has never flown.
Hivemind: What 'Pilotless' Really Means
The second half stories — software. The Shield AI autonomy system is called Hivemind and appeared long before the X-BAT. It was tested on a light reconnaissance aircraft. V-BAT, which is reportedly Defense News And according to Shield AI itself, it has been operating in Ukraine since 2023, including in areas where GPS is constantly jammed and communications are limited.
Hivemind's principle differs from a conventional autopilot. The car doesn't follow a preset route, but replans it mid-flight, using onboard sensors, without external cues. Application Solution weaponsAccording to the company, the control remains with the human pilot; everything else—navigation, avoiding threats, and choosing the right moment to maneuver—is handled onboard. If contact with the operator is lost, the vehicle does not turn back home but continues operating according to its latest set of rules.
This is where the main difference from the Vertijet lies. The Tailsitter isn't difficult to fly: in flight, it behaves like a regular airplane. It's difficult to land: you need to simultaneously reduce speed, maintain vertical flight, control drift, and see the landing point. A human lacked the neck and eyes for this. The algorithm requires an inertial unit, optics, and precise positioning using visual markers on the platform; all of this costs pennies by aviation industry standards in 2026.
The project has its skeptics, and serious ones at that. A British analyst RUSSIA Justin Bronk, in his publications on unmanned wingman programs, has repeatedly pointed out that mass-scale, low-cost autonomy and resistance to electronic countermeasures are usually at odds with each other: one or the other. There's also a more specific complaint: landing a jet-powered tailsitter in a strong crosswind remains an unsolved problem for the algorithm. DARPA's experiments with small tailsitter UAVs have run into precisely this: at certain angles and wind speeds, the platform simply lacks verticality. Deck trials in the Pacific Ocean in 2027–2028 will reveal the extent to which this problem has been addressed.
Where does this fit in: X-BAT's place in the family of unmanned wingmen
The X-BAT isn't just coming out of nowhere. The US Air Force has already selected it for the program. Collaborative Combat Aircraft (CCA) two machines: YFQ-42A from General Atomics and YFQ-44A From Anduril. Both are classic aircraft designs, both require a runway. XQ-58a Valkyrie Kratos is the same way: he takes off with a rocket booster and lands with a parachute. Everyone's runway is different, but a runway is essential.
The X-BAT wasn't selected for the CCA. It's a Shield AI initiative, not a Pentagon contract—a detail glossed over in press releases, but one that's hard to ignore. The company is carving out a niche for itself with a single selling point its competitors lack: vertical takeoff. According to Shield AI management in public interviews, the price of the aircraft is $20–30 million, with a production run of 150 units per year, operating in a single shift. Both figures are based on the developer's own statements, without external confirmation. The history of the American aviation industry over the past thirty years teaches us to take such figures with a grain of salt.
The problem the X-BAT solves isn't new. In 1951, the US Navy wanted a jet fighter for small ships without a flight deck and didn't get one. In 2026, Shield AI is proposing a jet strike vehicle for amphibious assault ships, small platforms, and undeveloped landing pads—precisely for those carriers that don't have, and aren't expected to have, a runway. The mission statement hasn't changed in 70 years. What has changed is the thrust vectoring engine, the onboard electronics, and, most importantly, the lack of someone in the cockpit looking over their shoulder.
What this will actually mean for the future will become clear by the end of 2026, when the X-BAT will attempt its first vertical lift and landing. If successful, the aviation industry will have its first viable answer to the age-old question of ships without a flight deck since Vertijet. If not, the idea will be shelved for another thirty years, as the next attempt at the same approach will have to wait for a new generation of autonomous flight.
Information