Breaking the mould: The future of VTOL autonomous fighters in the Indo-Pacific

In short, Shield AI is pitching X-BAT as a “transformative” 4th-generation-class jet that combines VTOL agility, very long range (>2,000 nm), multi-role payloads and rugged autonomous operation.

Origins and development

Shield AI itself is a deep-tech start-up founded in 2015 by former US Navy special forces and Silicon Valley engineers. The company built its reputation on advanced autonomy software (Hivemind) and on Group-3 drones like the V-BAT, which have even seen combat use.

Notable investors include L3Harris and Andreessen Horowitz, while behind the scenes, Shield AI has been assembling a world-class team (even poaching talent from SpaceX) to tackle X-BAT’s challenges. In 2024, it announced the acquisition of Sydney-based Sentient Vision (an AI-enabled ISR specialist) to create a local subsidiary.

In early 2025, its V-BAT drones received certification for Australian civil airspace and began training programs, working with local partners such as Toll Aviation.

All this suggests Shield AI sees Australia as a strategic ally for testing and employing its drones.

X-BAT itself was first teased at the 2024 Paris Air Show and formally revealed in Washington, D.C. Company leaders displayed scale models and concept art – including tail-sitting mock-ups on launch trailers and highlighted aggressive, Silicon Valley-esque-aggressive timelines.

By the company’s account, X-BAT had been under development for only 18 months when unveiled. High-speed ground testing (engines, aerodynamics, radar signature, thermal, etc.) has been underway, and Shield AI’s roadmap targets an initial VTOL demonstrator flight in late 2026 with an all-up flight test program in 2028.

In line with its Silicon Valley roots, the company aims for early production by about 2029, aligning with USAF’s future programs.

From its inception, X-BAT has been pushed as a complete system solution – not just an airframe. Company officials emphasise that nothing in X-BAT’s design is left to chance: all major risks (VTOL control, stealth shaping, engines, sensors) use mature, flight-proven technologies.

For example, X-BAT’s tail-sitter VTOL architecture avoids complex lift fans or tiltrotors; instead, it uses a single F100/F110 class turbojet with afterburner, aligned with the fuselage, to achieve vertical launch. Because many high-performance fighters already have thrust-to-weight >1, the afterburning F100/F110 provides enough push to lift the jet straight up.

A 3-axis thrust-vectoring nozzle (an updated version of 1990s “ACTIVE” technology) will then tilt the exhaust to balance and control the aircraft during hover and transition. In short, no additional booster is needed – X-BAT “lifts off like a rocket” under full reheat, then pitches nose-down at about 1,600ft to fly in wingborne mode, before continuing to climb to its operational altitude.

Landing is essentially the reverse process. Early X-BAT promotional videos show the jet (now nearly empty of fuel) flaring into a vertical descent onto a trailer cradle – a capability evoking the 1950s Bell X-13 tail-sitter experiments. Shield AI has reportedly leveraged its experience from V-BAT and modern guidance/control systems to make this automated landing feasible. (A key advantage is that the system uses throttle control for stability instead of an arrested hook.)

The company plans multiple trailer-launch and -recovery units per aircraft, meaning an X-BAT unit could operate from any prepared site at relatively short notice. This runway-free lift-off and landing mode is central to X-BAT’s strategic promise – it lets jets “go to where the fight is” even if runways are contested or cratered, effectively extending the reach and survivability of airpower.

Incorporating proven tech to mature at pace

At roughly 26ft long with a 39ft wingspan, the X-BAT is about the size of a classic light fighter (comparable to an F-5). Its shape is a flying-wing/cranked-delta planform with smooth control surfaces and no vertical tail – optimised for stealth against radar. Shield AI calls it a “4th-generation” design, and indeed it resembles a sleek, engine-down-the-middle, faceted wing. The powerplant (GE F110-GE-129) and vectoring nozzle are proven hardware, while the rest of the airframe uses advanced composites and shaping to reduce infrared and radar signatures.

Performance goals are ambitious. Shield AI says X-BAT will exceed Mach 1.2 (i.e. supersonic cruise), enabling it to keep pace with modern fighters. Combined with its high thrust-to-weight ratio, this will allow rapid insertion or egress from combat zones. Its reported service ceiling is >50,000ft. Most striking is its range – the company claims over 2,000 nm (3,700km) on internal fuel with a combat payload. At that range, X-BAT’s combat radius could approach 1,000 nm (1,850km), far beyond typical UAVs.

This extraordinary endurance – only possible by combining jet performance with large fuel capacity – is said to allow theatre-scale operations (e.g. Pacific islands), critically without tanker support. Even with external stores, Shield AI suggests that one X-BAT’s range exceeds that of many fighters; for example, its 2,000 nm range is roughly equal to that advertised for Australia’s subsonic Ghost Bat loyal-wingman drone.

Internally, X-BAT is designed to be multi-role. It will include a weapons bay (for air-to-air and air-to-ground munitions) plus four external hardpoints for missiles or bombs. Shield AI’s promotional material shows it carrying standard Western missiles, including AIM-120 AMRAAMs and the new AIM-174B “Gunslinger” very-long-range A2A missile for Navy use.

Concept art shows that it can deploy a wide variety of small, precision weapons. The bay and pylons could also house electronic warfare pods, datalinks or extra fuel tanks as needed.

In practice, X-BAT might carry a mix of air-to-air missiles (for escort or counter-air), bombs or standoff weapons (for strike), and sensors (radar, EO/IR, RWR/EW suites) for ISR or jammer roles. The airframe is intentionally modular – payload units can be swapped for different missions.

Crucially, X-BAT will operate autonomously under Shield AI’s Hivemind software. This AI system is built to cope with jammed or GPS-denied conditions and to fly complex formations and tactics with minimal human input. In practice, a human “mission commander” could task a group of X-BATs with objectives, and the drones would carry out flights, target identification, threat engagement or cover suppression on their own. They can also pair up with a crewed fighter: the fighter acts as a commander/sensor node, delegating tasks to the drone – the so-called “loyal wingman” role.

Because Hivemind evolved from manned-fighter autonomy, even reportedly flying F-16s and MQ-20 Avenger drones during testing, X-BAT’s software is expected to meet high combat standards. The AI also allows flexibility of command – if communications with the mothership drop out, X-BAT continues its mission with built-in contingencies.

A new era for manned-unmanned teaming in Indo-Pacific air combat doctrine

The X-BAT is explicitly conceived for manned-unmanned teaming (MUM-T) and “loyal wingman” operations. In US military doctrine, a loyal wingman is an uncrewed aircraft that accompanies manned fighters into combat, multiplying sensor reach and firepower while taking on high-risk tasks. Australia’s RAAF, for example, is already building the Boeing MQ-28A Ghost Bat as a loyal wingman for its F-35A fleet.

Shield AI pitches X-BAT as the next step in that concept – a high-end wingman that can accompany F-35s or future NGAD-type fighters on deep missions.

In practice, an F-35 or F-22 pilot might direct X-BAT(s) via datalink to perform screen, scout, or suppress missions. For instance, X-BAT could fly ahead or on the flanks of a strike package, using its stealth and AI to locate and engage threats. It could loiter to provide long-range targeting or ECM support.

With its long range, X-BAT could also act independently as a “gunship” from a distant base or carrier, returning after its mission. When operated as a swarm or team, multiple X-BATs can coordinate their actions with or without continuous human oversight.

This fits wider trends. The USAF and Navy have active Collaborative Combat Aircraft (CCA) programs exploring similar concepts, having already chosen demonstrators (e.g. Anduril and GA for Increment 1) and now eyeing more advanced wingman designs. Shield AI plans to compete in CCA Increment 2 with X-BAT. Likewise, Australia, the UK, and other allies are researching CCAs and loyal wingmen. Indeed, an Air & Space Forces report notes that other countries like Australia, the Netherlands, and the UK are also exploring this class of drone.

Australia’s own Ghost Bat is a smaller, subsonic CCA, mainly for sensor escort roles, whereas X-BAT would be a more lethal, supersonic counterpart – potentially of interest to AUKUS partners if proven.

Beyond two-ship teaming, X-BAT points toward future swarm warfare. If hundreds of affordable X-BATs could be produced, they might saturate enemy defences or conduct massed strikes under AI control. Shield AI’s CEO frames this as “airpower without runways” – projecting deterrent mass from many dispersed points.

In a contested Pacific scenario, for example, X-BATs could be dispersed among island outposts or amphibious ships. Losing some to enemy fire would be acceptable if tens remain to accomplish the mission – a departure from risking a single $100+ million crewed jet.

This attritable strategy is a key selling point: the company markets X-BAT as delivering “fighter-class performance at an order-of-magnitude lower” cost and risk. In contrast, 5th-gen fighters and their missiles cost millions per hour or per shot, so adding a layer of cheap, smart drones can multiply the size of the force at modest expense.

Reshaping the strategic calculus

The rise of aircraft like X-BAT could reshape air force structure and operations. Survivability on the modern battlefield increasingly means avoiding static runways and chokepoints. X-BAT’s VTOL capability directly addresses this by allowing operations from remote or quickly constructed sites.

For pilot safety, X-BAT’s fully autonomous mode means human crews stay further from high-threat zones. A human pilot can direct X-BAT to fly point or hunt surface threats, letting the drone face first-pass risks. In the event of heavy enemy air defences or anti-access zones, manned fighters could lurk at standoff while X-BATs press the engagement.

Thus, X-BAT extends the reach of fifth-gen jets without exposing airmen to initial waves of fire.

In terms of force structure, adding attritable wingmen could reduce the number of highly expensive jets needed for certain tasks. A mix of F-35s (or future sixth-generation NGAD) and X-BATs could cover a mission set that once demanded a larger fleet of crewed fighters. The idea is analogous to having “cheap chaff” to distract SAMs while the core of the strike package passes.

The RAAF’s own roadmap is already moving this way: the MQ-28 Ghost Bat (a Boeing-developed CCA) is being acquired to team with F-35s for ISR and escort. X-BAT, while not an Australian program, complements that vision by adding a higher-speed, longer-endurance option. In joint US-Australia exercises or AUKUS integrated force concepts, having interoperable CCAs could allow shared tactics, for example, a US F-35 controlling a US X-BAT and an Australian Ghost Bat in concert.

Finally, on cost-efficiency, Shield AI asserts that X-BAT will be much cheaper to buy and operate than legacy fighters. A recent estimate gave a target unit cost of about $27 million, roughly the Air Force’s original budget for a CCA. By contrast, an F-35A costs about $80–100 million per aircraft, with $100,000+ daily sortie costs.

Even big missiles like the AIM-120 or AIM-9 cost inthe order of $1–2 million. By building X-BATs with commercial practices (rapid digital design, smaller teams, concurrent testing), Shield AI hopes to drive that cost down. (Indeed, War Zone comments note that by leveraging modern CAD and AI tools, it is building X-BAT with far fewer engineers than Cold War jets required.)

The attritable model still has costs, and its jet engine alone is in the million-dollar range, but the idea is to accept losses rather than ensure every aircraft survives at any cost.

Eyes on the future

Despite the hype, experts caution that X-BAT is extremely ambitious. As one analyst put it, “what Shield AI is attempting to pull off … is no small task and there will certainly be sceptics along the way”. Converting a jet fighter into a reliable VTOL drone involves serious aerothermal and control challenges.

Stealth shaping, weapons integration, autonomous tactics, and hardened comms are all complex to get right. Even if the individual technologies exist, integrating them on a tight schedule is risky.

So far, Shield AI has mitigated the risk through extensive early testing (wind-tunnel testing, model RCS, engine runs) and off-the-shelf components where possible. Its team’s pedigree –SpaceX veterans in launch tech, ex-military engineers – also lends confidence.

Another unknown is cost and sustainment. Shield AI claims affordability, but real-world production often reveals hidden costs in the form of engine maintenance, complex GNC gear, and high-end sensors/missiles. The RAAF’s analysis of its own Ghost Bat program emphasises that sensors and missiles remain expensive “even for attritable drones”.

If X-BAT uses large jets and high-end armament, its attritability threshold may be higher than anticipated. On the other hand, Shield’s use of a common fighter engine with existing logistics chains and modular payloads should help keep maintenance more routine than a totally new design.

Critics also note that autonomous aircraft raise ethical and command-control issues, though both US and Australian doctrine currently restricts lethality decisions to humans, so X-BAT would operate under supervision.

Finally, there is a timing challenge. Shield AI’s schedule (first flight ~2026, production by 2029) is very aggressive for a new combat aircraft, even a drone. To succeed, the company will need substantial engineering and funding to meet milestones.

On the other hand, the Pentagon’s urgent drive to field CCAs and allied pressure to counter peer threats (e.g. in the Indo-Pacific) provide strong incentives to push programs like X-BAT forward.

If realised, Shield AI’s X-BAT could significantly enhance joint US–Australian airpower by expanding the “manned-unmanned” battlefield.

For Australia, even partial adoption of X-BAT concepts would dovetail with its MQ-28 Ghost Bat and F-35 fleets to create a layered network of crewed and uncrewed platforms. For the US, X-BAT and similar drones will complement the F-35 and future NGAD fighters, allowing a more flexible, dispersed force posture.

Time and further testing will tell if Shield AI can deliver on this promise, but for now, X-BAT has certainly put the US–Australia alliance’s skies on notice that the age of AI wingmen is near.