The U.S. Air Force is actively testing a new class of armed unmanned aircraft designed to fly alongside crewed fighters as autonomous wingmen—and for the first time, these aircraft are firing live weapons. In July 2026, an Anduril YFQ-44A Fury drone successfully fired a live AMRAAM air-to-air missile while tracking its target, marking a significant milestone in the Collaborative Combat Aircraft (CCA) program. This isn’t theoretical development—it’s operational validation of a fundamental shift in how air combat may be conducted, where humans and machines coordinate at tactical speeds and distances that challenge traditional command structures.
The CCA program represents one of the Air Force’s most ambitious modernization efforts. Rather than building larger, more capable single aircraft, the service is betting on smaller, lower-cost drone wingmen that can extend fighter coverage, share targeting data, and carry air-to-air weapons. Two competitor designs—General Atomics’ YFQ-42A Dark Merlin and Anduril’s YFQ-44A Fury—are locked in parallel testing to prove the concept works at scale. Both aircraft flew for the first time in late 2025, and throughout 2026, they’ve transitioned from basic autonomy demonstrations to actual weapons integration.
Table of Contents
- What Are Collaborative Combat Aircraft and Why Does the Air Force Want Them?
- Weapons Integration Testing: The Bridge Between Autonomy and Live Fire
- The Two Competing Designs: Dark Merlin and Fury
- Integration with Crewed Fighters and Beyond-Visual-Range Command
- The Autonomy Question and Its Limits
- Production Plans and Operational Scale
- Testing Milestones Through 2026 and Beyond
- Frequently Asked Questions
What Are Collaborative Combat Aircraft and Why Does the Air Force Want Them?
Collaborative Combat Aircraft are semi-autonomous drones built to operate under human command but with enough onboard decision-making capability to react to threats faster than a remote pilot can respond. The core appeal is multiplexing: instead of one pilot controlling one multi-million-dollar fighter, that pilot can manage flight lead duties while a drone wingman handles subordinate tasks—defensive scanning, jamming coordination, or stand-in surveillance. The Air Force envisions two CCA drones per next Generation Air Dominance (NGAD) fighter, ultimately procuring roughly 1,000 aircraft to pair with its planned fighter fleet. The operational logic is straightforward but represents a departure from decades of unmanned doctrine.
Traditional drones like the MQ-9 Reaper are controlled from a ground station and excel at persistent surveillance and strike missions in low-threat environments. CCAs operate in contested airspace where real-time communication may be jammed, requiring the drone to make tactical decisions independently—whether to break formation, when to launch a missile, how to defend itself. This semiautonomous layer is what distinguishes the CCA from both traditional remotely piloted aircraft and fully autonomous platforms. The Air Force began weapons integration testing with inert munitions in February 2026, a necessary step to validate that carrying live ordnance doesn’t compromise airworthiness or safety in formation flight.
Weapons Integration Testing: The Bridge Between Autonomy and Live Fire
Before an armed drone can operate tactically, the Air Force must prove three things: that it can carry weapons safely, that it can employ them reliably, and that integration with crewed aircraft command chains works as designed. The weapons integration process started with inert (training) munitions to isolate airframe, power system, and software risks before introducing live ordnance. General Atomics and Anduril each conducted these captive-carry tests through early 2026, gathering data on weight distribution, fuel consumption, and autopilot stability with weapons mounted. The leap to live fire is where the real testing begins, but it also carries genuine risk.
An armed drone operating with semiautonomous decision logic in close formation with manned aircraft introduces liability questions that have no clear historical precedent. If a CCA’s guidance system malfunctions and it launches a missile at an unintended target—or at a friendly aircraft—the consequences are severe. The Anduril Fury’s July 2026 live AMRAAM shot was therefore conducted under tightly controlled conditions with extensive range safety validation. Such testing happens at military ranges where target drones can be pre-positioned and all other aircraft are kept at safe distances. Real operational scenarios will demand far more autonomy than a range can safely simulate.
The Two Competing Designs: Dark Merlin and Fury
General Atomics’ Dark Merlin completed its first semiautonomous flight in February 2026 alongside crewed fighter escorts, demonstrating that the airframe could maneuver in formation with minimal pilot input. The Dark Merlin draws on General Atomics’ decades of experience with the MQ-9 Reaper line, suggesting a design prioritizing operational maturity and integration with existing Air Force logistics. Anduril’s YFQ-44A Fury, meanwhile, represents a newer entrant to the unmanned aircraft market—the company was founded in 2017—but it has opted for a more aggressive design philosophy centered on rapid autonomy iteration and cloud-connected mission planning. The two aircraft embody different engineering philosophies that matter operationally.
The Dark Merlin is expected to leverage proven engines, airframes, and sensor packages that the Air Force already understands; this reduces technical risk but may limit raw autonomy capability. The Fury, by contrast, is purpose-designed for collaborative autonomy from the ground up, incorporating advanced AI mission planning and distributed control logic. Both fired inert weapons during captive-carry testing and passed basic formation handling checks by mid-2026. The fact that Anduril’s Fury achieved the first live fire milestone—the July AMRAAM shot—suggests its autonomy and control architecture may be ahead of schedule, though one successful test is far from proof of operational readiness.
Integration with Crewed Fighters and Beyond-Visual-Range Command
A CCA is only useful if crewed aircraft can command it reliably, even when radio signals degrade or adversary jamming begins. Proof-of-concept for this occurred when an F-35 Lightning II successfully linked with an MQ-20 Avenger unmanned combat aircraft for a beyond-visual-range command and control demonstration. The MQ-20 is a different aircraft than the two primary CCA contenders, but the test validated a critical assumption: legacy crewed fighters can establish a command channel with advanced drones, pass targeting data in real time, and maintain formation without ground station intervention.
This integration test is significant because it sidesteps one of the CCA program’s biggest unknowns—whether existing fighter cockpits can handle the workload of directing a drone wingman during a high-threat engagement. The F-35’s integrated avionics and helmet-mounted display system proved capable of commanding a drone’s weapons employment and maneuvers. However, the demonstration was conducted under controlled conditions; real air combat is noisier, faster, and more ambiguous than a test range allows. The Air Force will need to refine pilot training and crew procedures throughout 2026 and beyond before CCAs are declared operationally ready.
The Autonomy Question and Its Limits
Autonomy is both the CCA’s greatest strength and its most contentious vulnerability. A drone that can independently track targets and navigate in a degraded communication environment enables tactics that a tethered remotely piloted aircraft cannot attempt. During the Fury’s live missile test, the aircraft was actively tracking the target throughout the engagement, meaning its sensors and autopilot were maintaining target solution without constant ground guidance. This level of autonomy allows a wingman drone to break formation and intercept threats while the crewed flight lead coordinates from a safer position. Yet autonomy also introduces ethical and technical unknowns.
Artificial intelligence systems that decide whether to fire on a target—even under a human’s pre-approval—operate in a legal and moral gray zone that military law has not fully clarified. The Air Force’s current doctrine requires human-in-the-loop weapons employment for CCAs, meaning a pilot must explicitly authorize weapons release, but the line between human control and autonomous decision-making will blur as engagements accelerate. Technically, every autonomous system tested to date has failure modes that engineers cannot fully predict or prevent. The CCA program’s risk mitigation strategy relies on extensive simulation, range testing, and incremental integration with crewed aircraft before any drone operates in a real contested environment. This phased approach is sound, but it also means CCAs will not be combat-deployed for at least several years.
Production Plans and Operational Scale
The Air Force has already awarded initial production contracts for approximately 100 CCA aircraft in 2026, with competitive testing still underway to select a final winner or potentially both designs. The procurement plan is ambitious: two CCAs per 500 planned Next Generation Air Dominance fighters translates to roughly 1,000 drone wingmen in the long-term inventory. This procurement scale would make CCAs the most numerous tactical aircraft type in the Air Force by sheer numbers, even if each individual aircraft costs far less than a crewed fighter. The industrial base impact of this scale is substantial.
General Atomics and Anduril will need to scale manufacturing, supply chains, and training infrastructure to support sustained production. Pilot training, maintenance technician certification, and depot-level repair networks must be built in parallel with aircraft production. The Air Force has learned from past programs that unmanned aircraft logistics often consume more resources than originally budgeted; every CCA deployed forward will need spares, maintenance crews, and pilots trained to manage its autonomy settings. Initial production in 2026 will likely focus on reducing technical risk and validating manufacturing processes rather than achieving full-rate production immediately.
Testing Milestones Through 2026 and Beyond
The CCA testing timeline compressed significantly during 2026, accelerated by Air Force pressure to field some capability by decade’s end. Both aircraft flew their first semiautonomous sorties in early 2026, moved to weapons integration by mid-year, and achieved live fire by July—a compressed schedule that reflects the service’s sense of urgency regarding near-peer adversary capabilities. However, the jump from a single successful live missile test to operational deployment is substantial and will require sustained testing through 2026 and into 2027.
The tests ahead include formation combat maneuvering with multiple CCAs and crewed fighters, electronic warfare engagement scenarios, and coordinated strike packages mimicking real campaign structures. Each milestone carries technical and political weight; a catastrophic failure during testing could delay the program years or trigger a shift to a different technical approach. Conversely, each successful integration point—formation flying with weapons, automated threat response, multi-aircraft coordinated maneuvers—validates the architectural choices and builds confidence in the platform. The July 2026 Fury missile shot represents progress, but it is a single data point in what will be years of iterative testing and refinement before collaborative combat aircraft become routine in Air Force operations.
Frequently Asked Questions
What is the difference between a CCA and a traditional remotely piloted drone like the MQ-9 Reaper?
Traditional remotely piloted aircraft are flown by a pilot at a ground station and excel at surveillance and strike missions in low-threat environments where communication is reliable. CCAs operate in contested airspace with degraded communications, requiring semiautonomous decision-making. They’re designed to fly in formation with crewed fighters and react to threats faster than a remote pilot can respond.
Why is the Air Force building two competing designs instead of selecting a winner?
Parallel development reduces technical risk by validating competing approaches and maintaining industrial competition. The Air Force can learn from both programs and potentially select one or both designs based on testing results. Competition also spurs innovation and prevents over-investment in a single platform that might have unforeseen flaws.
How does a pilot control a drone wingman in a high-threat environment?
CCAs are designed with semiautonomous mission planning, allowing a pilot to set objectives and decision rules that the drone executes independently. For weapons employment, current doctrine requires explicit pilot authorization, but the drone can autonomously track targets and navigate through degraded communications. Real-time command links are used when available, but the drone does not depend on them for every tactical decision.
What does “beyond-visual-range command and control” mean for CCAs?
It means a crewed fighter can command and coordinate a drone wingman at distances far beyond line-of-sight, using radio data links and integrated sensor networks. The F-35 test validated that legacy crewed fighters can establish this command relationship with advanced drones without ground station support.
When will CCAs be deployed operationally?
The Air Force is conducting intensive testing through 2026 and beyond. Operational deployment is unlikely before 2028 at the earliest, pending successful integration testing, pilot training development, and formal combat readiness declarations. Initial production contracts cover approximately 100 aircraft in 2026, but full-rate production scales up over years.
