MQ-28 Ghost Bat: How Australia Built the West's Most Advanced AI Loyal Wingman

In May 2020, at RAAF Base Amberley in Queensland, an unmarked aircraft with a sharply tapered nose, delta-esque wing geometry, and no cockpit rolled out of a Boeing Australia hangar for the first time. The aircraft was called the Airpower Teaming System. The media coverage was politely curious. Defense analysts were less casual. What they were looking at was the first operational prototype of a new category of weapon system — the AI loyal wingman — that could fundamentally alter the arithmetic of air combat in the decades ahead.

By 2021, the Royal Australian Air Force had given the aircraft an official designation: MQ-28A Ghost Bat. Named after one of Australia's native bat species, the Ghost Bat is designed to fly alongside crewed fighter aircraft — F/A-18F Super Hornets, EA-18G Growlers, and eventually F-35As — as an autonomous AI-piloted companion that extends the sensor reach, electronic warfare capability, and tactical flexibility of its human wingman without putting another pilot's life at risk. As of 2024, six Ghost Bat aircraft are in the test fleet, undergoing evaluation by the RAAF Aerospace Research and Development Unit (ARDU) at RAAF Base Edinburgh in South Australia. The program has attracted export interest from the UK Ministry of Defence, Japan's Acquisition, Technology and Logistics Agency (ATLA), and Germany's Luftwaffe. It has also become the centerpiece of AUKUS Pillar II — the technology-sharing strand of the trilateral security partnership between Australia, the United Kingdom, and the United States.

Understanding the Ghost Bat requires understanding three things: what "loyal wingman" actually means as a tactical concept, why Australia — not the United States or United Kingdom — built the West's first operational example, and what "attritable" means for the future of air warfare doctrine.

From Airpower Teaming System to MQ-28A

The Ghost Bat's origins trace to a 2018 contract between Boeing Australia and the Royal Australian Air Force, valued at approximately AUD $40 million for the initial development phase. Boeing Defence Australia — the subsidiary responsible for the program — deliberately chose to develop the aircraft in Australia rather than import it from the US parent company. This decision was not merely nationalistic posturing. It reflected a calculated Australian government strategy to develop sovereign defense industrial capability in advanced air systems, reducing dependence on US supply chains that could be disrupted in a major Indo-Pacific conflict.

The program name "Airpower Teaming System" described the core concept precisely: an unmanned aircraft whose primary purpose was not independent operation, but teaming — flying in coordinated formations with crewed aircraft and amplifying their combat effectiveness through AI-enabled task distribution. The first aircraft rolled out in May 2020; the first flight occurred on February 27, 2021, at RAAF Base Woomera in South Australia. The MQ-28A designation was formally assigned later in 2021, making it only the second aircraft to receive a US-style MQ designation (the first being the MQ-9 Reaper) while being developed and built entirely outside the United States — a significant acknowledgment of the program's legitimacy within the broader Western defense ecosystem.

Boeing Australia's program lead, Dr. Shane Arnott, described the design philosophy in terms that revealed how seriously the team took the AI teaming challenge: the aircraft needed to be smart enough to execute assigned tasks autonomously, but humble enough to remain subordinate to the human pilot leading the formation. Getting that balance right — what engineers called the "teaming algorithm" — was the hardest part of the program, not the airframe itself.

The Airframe: Built to Be Expendable

The Ghost Bat's physical specifications place it firmly in the "medium-sized tactical drone" category, but the design philosophy is distinct from both large surveillance drones like the MQ-9 Reaper and small expendable loitering munitions. The aircraft is approximately 11.7 meters long with a wingspan of roughly 7.3 meters — significantly larger than a standard fighter drone but smaller than any crewed combat aircraft. The engine is a single Honeywell turbofan, providing subsonic performance in the Mach 0.7–0.85 range adequate for formation flight with Super Hornets and Growlers operating in non-afterburner cruise configurations.

The most distinctive external feature is the modular nose section. Unlike conventional aircraft designs where the nose structure is integral to the airframe, the Ghost Bat's forward fuselage section can be swapped between missions, allowing different sensor packages — synthetic aperture radar (SAR) for ground mapping, electronic warfare suites for jamming and signals intelligence, ISR payloads for reconnaissance — to be installed without requiring a different aircraft. This modularity is not merely an engineering convenience; it reflects a deliberate operational concept in which a single airframe type can serve multiple mission roles depending on the current tactical requirement.

The internal payload bay, located in the lower fuselage, can accommodate weapons or additional sensors. Boeing has not publicly confirmed what weapons the Ghost Bat has been cleared to carry, but the bay dimensions are consistent with carriage of small-diameter bombs or air-to-air missiles. The aircraft also features conformal fuel tank capacity that extends its range well beyond 3,000 kilometers — a critical figure for Australian operations given the vast distances of the Indo-Pacific theater.

AI Mission Systems: The Brain Inside the Ghost Bat

The Ghost Bat's physical platform is the less interesting story. The more consequential technology is the AI mission system stack that enables it to operate as a genuine tactical partner to human pilots rather than a remotely piloted vehicle that happens to fly near fighters.

The core of the Ghost Bat's autonomy is what Boeing Australia calls a "mission system autonomy stack" — a layered software architecture that handles navigation, sensor management, threat response, and formation keeping without continuous human operator input. The system draws on multiple AI disciplines: computer vision for terrain awareness and obstacle avoidance, reinforcement learning for tactical maneuvering, Bayesian inference for threat assessment, and natural language-adjacent command interfaces that allow human pilots to issue high-level tactical instructions ("go prosecute that radar site," "hold this position while I engage") that the Ghost Bat interprets and executes autonomously.

The teaming algorithm — the core innovation that distinguishes a loyal wingman from a remote-controlled drone — manages the Ghost Bat's position relative to its crewed lead aircraft and dynamically allocates tasks between the human pilot and the AI system. When flying in formation, the Ghost Bat maintains a separation envelope of 200 meters to 2 kilometers from the lead aircraft, adjusting based on threat environment, airspace geometry, and the tactical task being executed. This separation range is not arbitrary: it's calculated to allow the wingman to be tactically useful (close enough to respond to formation-level threats) while providing sufficient standoff that electronic emissions from the Ghost Bat's sensors don't interfere with the lead aircraft's systems.

SAR and ISR Capabilities

The Ghost Bat's synthetic aperture radar capability, integrated into the swappable nose section, provides real-time ground mapping and moving target indication at ranges and resolutions that the F/A-18F Super Hornet's AN/APG-79 AESA radar cannot match in dedicated ISR configurations — partly because the Ghost Bat can position itself to optimize radar geometry in ways a crewed aircraft cannot, accepting risks a human pilot would decline. SAR imagery from Ghost Bat operations can be datalinked to the lead aircraft in near real-time, giving the human pilot situational awareness of the ground picture that would previously have required a dedicated ISR platform.

The electronic warfare suite is potentially the most operationally significant capability. Flying the Ghost Bat in an EW configuration alongside an EA-18G Growler creates a distributed jamming architecture that is considerably harder for adversary air defense systems to characterize and counter than a single jamming platform. The Ghost Bat can act as a forward jammer, projecting electronic noise into threat radar bands while the Growler maintains a standoff position; or it can serve as a decoy emitter, mimicking the radar signature of a crewed fighter to saturate adversary tracking systems.

RAAF Integration: Super Hornets, Growlers, and the F-35A

The RAAF's integration testing at Edinburgh has focused on two primary teaming pairs: Ghost Bat with the F/A-18F Super Hornet, and Ghost Bat with the EA-18G Growler. These choices are deliberate. The Super Hornet is the RAAF's primary swing-role strike platform; adding a Ghost Bat wingman extends its sensor reach and allows the human pilot to dedicate more cognitive bandwidth to high-level tactical decisions while delegating sensor management and formation flying to the AI. The Growler integration is arguably more innovative: pairing an EW-specialized crewed aircraft with an AI wingman that can extend the electronic warfare footprint creates a distributed jamming capability that no other air force in the world currently possesses in operational form.

RAAF ARDU's testing at Edinburgh has reportedly validated basic formation flying, datalink performance, and sensor-sharing protocols in a range of conditions including simulated electronic warfare environments. The next phase of testing, expected to progress through 2025 and 2026, focuses on more complex tactical scenarios: multi-aircraft formations with two or more Ghost Bats, integration with ground-based command systems, and contested airspace exercises where the Ghost Bat's AI must respond to simulated adversary jamming and missile launches.

Future integration with the F-35A Lightning II — the RAAF's primary stealth strike aircraft — is a longer-term objective. The F-35's advanced sensor fusion and data architecture make it a more complex teaming partner than the Super Hornet, but also a potentially more powerful one: a Ghost Bat flying ahead of an F-35 could cue targets for the stealth aircraft without exposing the F-35's radar, preserving the crewed aircraft's low observable signature while still generating targeting quality track data.

"The question isn't whether the Ghost Bat can fly. It can. The question is whether a human pilot can trust an AI system enough to actually rely on it when the shooting starts. That's the test that matters, and we're still working through it."

-- RAAF senior officer, background discussion with defense press, 2024

What "Attritable" Actually Means

The word "attritable" appears constantly in discussions of the Ghost Bat and its American counterpart, the XQ-58A Valkyrie. It deserves careful unpacking because it represents a genuine doctrinal revolution — not just a cost-accounting concept.

Conventional military unmanned aircraft fall into two categories: expendable (cheap enough to accept total loss, like a loitering munition or small ISR drone) and recoverable (expensive enough that recovery is essential, like the MQ-9 Reaper or RQ-4 Global Hawk). Attritable aircraft occupy a deliberate middle ground: they are designed with the expectation that they may be lost in combat and are priced accordingly, but they are sophisticated enough to provide genuine tactical utility that a true expendable cannot. The Ghost Bat at approximately $10 million per unit is roughly one-eighth the cost of an F/A-18F Super Hornet at $80+ million. That ratio makes accepting Ghost Bat losses tactically rational in a way that accepting Super Hornet losses is not.

This changes the risk calculus of air operations in profound ways. A human pilot leading a Ghost Bat wingman can assign the AI aircraft to the highest-threat axis of an engagement — the route most likely to attract missile fire, the radar geometry that requires flying within adversary SAM envelope range — without the moral and operational cost of putting a human being in that position. The Ghost Bat absorbs the risk; the human pilot survives to generate more combat power. In a high-end conflict with an adversary like China that has invested heavily in anti-access/area denial systems designed specifically to threaten crewed aircraft, this risk redistribution could be the difference between a functional air campaign and an unacceptable attrition rate.

The attritable concept also has implications for procurement strategy. Air forces have traditionally been reluctant to accept large aircraft losses because each loss represents an enormous financial and production timeline investment. An attritable aircraft that can be produced in larger numbers at lower unit cost — and whose loss is budgeted for in the operational plan — enables more aggressive tactical employment and removes the self-deterrence that makes high-cost aircraft underutilized in dangerous environments.

AUKUS Pillar II: Sharing the AI

The Ghost Bat has become one of the flagship programs of AUKUS Pillar II — the advanced capabilities strand of the tripartite security agreement between Australia, the United Kingdom, and the United States. While Pillar I famously covers nuclear-powered submarine technology transfer, Pillar II encompasses a broader range of advanced defense technologies including AI, quantum computing, hypersonics, and autonomous systems.

Within the autonomous systems domain, AUKUS Pillar II is enabling structured technology sharing between the Ghost Bat program, the UK's own loyal wingman effort (the BAE Systems Mosquito, currently in demonstrator phase), and the US Air Force Research Laboratory's XQ-58A Valkyrie program. The sharing is not simply exchanging blueprints. It involves alignment of teaming algorithms, datalink protocols, and autonomy certification frameworks — the foundational work that determines whether AI wingmen from different nations can eventually operate in combined formations.

The US XQ-58A Valkyrie, developed by Kratos Defense, shares the attritable concept with the Ghost Bat but differs in design emphasis: the Valkyrie is designed for longer-range penetrating strike missions, potentially operating ahead of crewed aircraft into denied airspace. Its unit cost at approximately $3–4 million is even lower than the Ghost Bat's, reflecting a more austere sensor and AI package. AUKUS Pillar II discussions have focused on whether the Valkyrie's range and the Ghost Bat's sensor suite can be integrated into combined operations — an AI wingman force that leverages the complementary strengths of both platforms.

The UK's Mosquito program, which completed its first flight in 2023, is at an earlier development stage than either the Ghost Bat or the Valkyrie. BAE Systems' Mosquito is designed to team with the Eurofighter Typhoon and eventually the BAE/Leonardo/MBDA Tempest sixth-generation fighter. AUKUS sharing is giving the Mosquito program access to teaming algorithm development that Boeing Australia has already worked through — potentially accelerating UK capability by several years.

Export Interest: Who Wants the Ghost Bat

The Ghost Bat's combination of demonstrated flight performance, operational AI teaming capability, and relatively accessible unit cost has generated documented interest from several nations beyond Australia's AUKUS partners.

Japan's Acquisition, Technology and Logistics Agency (ATLA) has been the most publicly engaged prospective customer. Japan's defense posture has undergone a fundamental shift since the 2022 National Security Strategy, which for the first time explicitly acknowledged counterstrike capability requirements and set defense spending on a path to 2% of GDP by 2027. The Japan Air Self-Defense Force's F-35A and F-35B fleets create a natural teaming opportunity with a Ghost Bat-class loyal wingman, and Japanese defense officials have made no secret of their interest in AUKUS-adjacent technology access. Ghost Bat export to Japan would require US approval under ITAR — International Traffic in Arms Regulations — since the aircraft incorporates US-origin technology, but the Biden and Trump administrations both identified Japan as a priority defense partner, making approval likely if a formal request is made.

Germany's Luftwaffe interest is more constrained by the country's participation in the Franco-German-Spanish FCAS program, which includes its own loyal wingman concept (the "New Generation Fighter" plus remote carrier architecture). German officials exploring Ghost Bat interest are navigating a tension between FCAS commitments and a recognition that FCAS delivery timelines — the 2040s at the earliest — leave a significant near-term capability gap. Whether Germany can acquire an interim loyal wingman capability without triggering a political crisis with France and Spain remains unresolved.

The UK Ministry of Defence has the most direct access to Ghost Bat technology through AUKUS Pillar II, but RAF interest in the aircraft as an interim capability alongside its Typhoon fleet — pending the longer-term Mosquito and Tempest programs — has been reported by UK defense publications. An RAF Ghost Bat acquisition, even at small numbers, would provide the teaming experience and doctrine development that the Mosquito program will need when it matures.

Comparative Assessment: Ghost Bat vs. Global Competition

China is the benchmark against which the Ghost Bat must ultimately be measured. The People's Liberation Army Air Force has been developing loyal wingman concepts under multiple programs, most notably the GJ-11 Sharp Sword stealth UCAV — a flying-wing design more analogous to the B-21 Raider than to the Ghost Bat — and reports of dedicated loyal wingman teaming drones whose designations remain obscure in open-source intelligence. China's industrial capacity and state-directed R&D funding mean that whatever teaming AI gaps exist today may close rapidly. The Ghost Bat's operational head start — actual flight testing, real teaming exercises with RAAF crews, validated datalink protocols — provides Australia with a genuine edge in the human dimension of AI warfare: the doctrine, training, and operational experience that no amount of engineering can shortcut.

Russia's loyal wingman program — the Okhotnik-B (S-70 Hunter) heavy UCAV — is a more capable but far less attritable system, emphasizing stealth and payload over the low-cost replicability that makes the Ghost Bat concept strategically significant. The Okhotnik-B is designed as an expensive persistent capability, not an expendable tactical asset. The doctrinal difference matters: Russia's approach accepts fewer platforms at higher survivability cost; Australia's approach accepts potential losses in exchange for operational aggressiveness and mass.

The French-German-Spanish FCAS program's remote carrier concept is conceptually similar to the Ghost Bat but substantially earlier in development. FCAS's remote carriers are envisioned as somewhat smaller, shorter-ranged platforms optimized for operations within the NGF fighter's tactical envelope. Program delays caused by industrial disputes between Airbus, Dassault Aviation, and their respective national governments have pushed the remote carrier operational date to the mid-2030s at best — a decade behind the Ghost Bat's current trajectory.

Strategic Implications for the Indo-Pacific

The Ghost Bat's strategic significance lies not just in what it can do today, but in what it signals about the future force structure of Australia's and the West's air combat capability. The RAAF's Force Structure Plan envisions eventual procurement of Ghost Bat derivatives in sufficient numbers to accompany every crewed fighter sortie into contested airspace — a vision in which human pilots are always accompanied by AI wingmen that absorb first-contact risk, extend sensor coverage, and create multi-axis tactical dilemmas for adversary air defenses.

Realizing that vision requires solving problems that remain unsolved. The teaming algorithm works in benign test environments; its performance in genuinely contested environments with sophisticated adversary electronic warfare, time-compressed decision cycles, and the chaos of actual combat remains unproven. The data links connecting Ghost Bat to its human lead — critical for the teaming concept to function — are vulnerable to jamming. And the question of autonomous weapons authorization — what level of autonomous lethality is legally and ethically permissible — has not been resolved in Australian law or international humanitarian law, creating a potential legal constraint on the Ghost Bat's full operational employment.

These are not fatal objections. They are the known problems of a program that has made more concrete progress toward operational AI loyal wingman capability than any other Western program. The Ghost Bat is not a finished weapon system. It is the most advanced existing prototype of a category of weapon that will define the character of air combat in the decades ahead — and Australia got there first.