Current AI, drone, and sensor programs are reinventing aerial combat. Looking ahead to 2100: distributed architectures, autonomous swarms, and a new role for pilots.
In summary
Future combat aviation is moving towards distributed architectures where collaborative drones supported by onboard AI operate alongside piloted fighters. Current developments in quantum sensors, adaptive propulsion, and optical links foreshadow the 22nd-century aerial battlefield, dominated by autonomous, fast, and networked swarms. The role of the pilot will evolve into that of a tactical conductor, while the machine will manage maneuvers and part of the combat.
The most sensitive issue remains that of lethal autonomy: drones already have the technical capabilities to identify and strike a target, but certification, traceability, and the legal framework are hindering their use without human intervention. In an environment where the speed of engagements is increasing, operational pressure could lead to delegating firing to the machine in very specific situations, redefining the relationship between humans, AI, and the decision to use force.
The dividing line between AI autonomy and “traditional” drones
Traditional drones rely on pre-programmed flight plans, radio links, and deterministic algorithms. Next-generation systems incorporate learning models, embedded representations of the environment, and emergent behaviors. Flight demonstrations have proven that AI agents can conduct close combat maneuvers, under safety constraints, against human pilots. These tests validate “certifiable” tactical autonomy: AI makes decisions within the flight envelope, while humans set the intent and remain the ultimate authority. This shift anchors the pilot-in-the-loop concept: humans are no longer “the ones who fly,” they are “the ones who decide.”
Distributed architecture replaces the “queen platform”
The core of air superiority is shifting from the airframe to the network. Sensors, effectors, and relays become “nodes” within a mesh network with a low probability of interception. Very high-speed (1 Gbit/s) optical links between aircraft and satellites pave the way for near-instantaneous tactical status sharing at altitudes of up to 500 km. At the same time, laser power transmission has exceeded 8.6 km for 800 W, proving that in the future it will be possible to recharge air-to-ground relay stations or stratospheric drones. The “kill web” is replacing the traditional targeting chain: distributed detection, sensor fusion, effect attribution, and short-loop retargeting.
The role of the “quarterback” fighter pilot
By the 22nd century, pilots will no longer be overwhelmed by the mechanics of flight. They will orchestrate a cluster of collaborative drones from a highly automated cockpit: prioritizing effects, authorizing strikes, repositioning relays, managing electromagnetic deception, and cognitive warfare. AI “co-pilot” assistants translate intentions into action plans, detect inconsistencies, prevent disorientation, and filter sensory overload. Training is based on realistic synthetic environments, reusing AI agents that already operate in real flight.
American Collaborative Combat Aircraft
The United States is industrializing the CCA concept: expendable, stealthy drones with modular sensors, specialized by role (penetrating ISR, jamming, decoy, strike, air defense). Iconic demonstrations—automated air combat, multi-platform meshing—have shown the maturity of robust tactical autonomy. On the industrial side, open software architectures (autonomy OS, containers, APIs) make it possible to load a mission agent in a matter of minutes and orchestrate a swarm from a 5th/6th generation fighter. The operational objective: to stretch the enemy’s denial bubble, saturate defenses, and shift the cost of attrition to vectors that are less expensive than piloted fighters.
European programs: from Wingman to Remote Carriers
In Europe, several lines are converging. The Future Combat Air System (FCAS/SCAF) envisions Remote Carriers cooperating with combat aircraft via a real-time combat cloud. Germany is pushing for a “Wingman” associated with the Eurofighter, with a modular payload (reconnaissance, disruption, decoys). The United Kingdom, Italy, and Japan (GCAP) are designing robust escort drones to work in tandem with a new-generation fighter. European AI software, developed by specialized players, is injecting cognitive jamming and multi-sensor fusion into the architecture.
Chinese advances: Stealth UCAVs and loyal wingmen
China is accelerating the deployment of stealth UCAVs and “loyal wingmen” designed to fly with air superiority fighters. Flying wing platforms intended for penetration and attack of surface-to-air systems operate with internal bays and reduced signatures. Long-endurance escort drones, presented at air shows, are geared towards escort, early warning, deception, and swarm combat. The appearance of naval variants suggests operations from aircraft carriers, which extends the depth of naval aviation maneuvers.
The Israeli and Japanese hubs: swarms and MUM-T
Israel has operationalized tactical swarms, with loitering munitions and micro-drones that are interoperable at the section/company level. The logic is vertical integration: sensors, C2, target detection, and short-loop firing, designed for urban environments and anti-access. Japan, for its part, is structuring “Combat Support Unmanned Aircraft” dedicated to MUM-T, with increasing budgets to triple its drone efforts. The objective is to support unmanned team members with F-35s and future GCAPs, and to build a “shield” of aerial, surface, and underwater drones around the archipelagos.
The technological building blocks are already there
Adaptive cycle propulsion increases useful thrust and thermal management; although rapid integration on certain fighters has been ruled out for reasons of cost and compatibility, the technology is powering next-generation engines. CMC (ceramic matrix composites) materials make hot parts lighter, raise permissible temperatures, and reduce cooling air flow, thereby reducing consumption. On the sensor side, gallium nitride AESA stabilizes high power densities; IRST optronics are back in favor in the face of stealth. Quantum sensors (cold atom accelerometers/gyroscopes) extend inertial holdover by several orders of magnitude, which is key in contested GPS environments. Free-space optics offer discrete multi-gigabit links, complementing LPI/LPD RF networks. Finally, active aerodynamics and boundary layer control (“fly-by-wire-by-AI” avionics) open up more agile flight envelopes without mass penalties.
Emerging weapons and effects
Airborne self-protection lasers face power and cooling bottlenecks on fighters, but directed energy weapons are advancing rapidly on land and at sea. High-power microwaves are becoming the weapon of choice for anti-drone warfare. On the kinetic side, hypersonic scramjet missiles and stealthy cruise missiles with two-way links are expanding the range of options. The “effect” is no longer limited to destruction: active decoys, misleading emissions, electronic attack pods, consumable decoy sensors, and mini-drones launched from cargo holds create confusion, impair enemy decision-making, and open up firing windows.
Economic metrics: the battle of costs
The markets are setting the trend: the “military AI” segment could almost double by 2030 (≈ $19 billion), while the military drone market could exceed $80 billion by that time. At the platform level, CCA unit targets are “in the tens of millions of dollars”; some existing expendable vectors have proven unit costs in the range of a few million. Compared to manned fighters costing more than $80–100 million per unit, the economics of attrition shift to the other side: we accept losing drones to preserve the architecture and operational effect.
Risks, ethics, and certification
Standardizing safeguards is essential: clear boundaries between autonomous engagement and firing delegation, model traceability, “red team” test benches, and “knock-it-off” procedures codifying the termination of an agent. Recent flight tests have established a set of safety rules (minimum altitudes, separations, rules of engagement, supervision by pilots), which will become doctrine. The pilot-in-the-loop principle remains: AI proposes, humans dispose—but at machine speed.
Lethal autonomy: technological horizon and ethical dilemma
The question is no longer whether autonomous drones can fly or maneuver on their own, but whether they will be able to open fire without human intervention. From a purely technical standpoint, several building blocks are already mature: automatic target recognition, GPS-free navigation, threat detection, and real-time calculation of the probability of collateral damage. The integration of embedded AI capable of merging sensors, prioritizing targets, and triggering kinetic or non-lethal weapons is therefore becoming possible.
However, technological feasibility does not guarantee acceptability. Today, pilot-in-the-loop or human-on-the-loop remains the rule: an operator must authorize the shot, even if the machine proposes the decision. Recent experiments have shown that AI can shoot down a target in simulation with reaction times in the order of milliseconds, but the United States, Europe, Japan, and Israel impose doctrinal and legal safeguards.
Three conditions must be met before lethal autonomy can be considered:
- Certifiable reliability: error rate below the thresholds accepted for the use of conventional weapons, demonstrated by thousands of hours of testing.
- Traceability and auditability: continuous recording of algorithmic decisions and the ability to review them after the action to account for them.
- Clear political and legal framework: appropriate rules of engagement and international law, incorporating command responsibility.
The debate is therefore no longer about what is possible, but about what is permissible. Until these conditions are met, autonomous firing will remain confined to experimentation. But as the speed of combat and the density of threats increase, operational pressure will push for certain lethal functions to be delegated to machines, at least in limited environments—for example, close defense against incoming swarms.
Concrete milestones towards the 22nd century
By the 2030s, penetrating CCAs will escort NGAD/GCAP/FCAS, communicating via optical links and piloted by collaborative drones with specialized roles. Fighters will become cognitive hubs, capable of reconfiguring their swarms in flight through software updates. By 2050-2070, quantum sensor inertial navigation will be standardized; solar-powered stratospheric relays, refueled by beams, will provide weeks of persistence. In the 22nd century, the “air force” will resemble a planetary system: a digital skin, multi-orbit and multi-environment, where humans retain strategic initiative, machines retain tactical initiative, and where permanent software and hardware innovation becomes the weapon.
The race is on: whoever masters software, data, and integration wins “time”—the only truly non-renewable resource in combat.
War Wings Daily is an independant magazine.