No, this isn’t the plot of the next Top Gun sequel. The U.S. Air Force really is
planning to field around 1,000 “robot jet fighters” autonomous or semi-autonomous
combat drones designed to fly alongside human pilots. And it has already tapped two
companies to compete for the job: defense startup Anduril Industries and veteran drone
maker General Atomics Aeronautical Systems.

These aircraft, formally known as Collaborative Combat Aircraft (CCA) or
“loyal wingman” drones, are meant to act as robotic sidekicks for front-line fighters
like the F-35 and the Air Force’s future Next Generation Air Dominance (NGAD) jets. The
Air Force’s long-term plan calls for at least 1,000 CCAs in the first wave alone and
senior leaders have hinted they could eventually buy far more.

The stakes are massive: billions of dollars in contracts, a reshaped defense industrial
base, and a fundamental shift in how air wars are fought. Let’s unpack what this robot
fighter competition really is, who’s in the running, and why the number
“1,000 robot jet fighters” is suddenly serious Pentagon math instead of sci-fi fanfic.

From Sci-Fi Concept to Real Air Force Plan

The Air Force has been public about its goal for several years: pair each of roughly
500 advanced manned fighters (about 200 NGAD and 300 F-35A jets) with at least two
unmanned CCAs. Do the math and you get around 1,000 robot wingmen with the option to
grow that fleet even larger if the concept works.

These drones aren’t intended to replace human pilots entirely. Instead, they act as
force multipliers: carrying extra missiles, jamming enemy radars, gathering
surveillance data, or even flying ahead into dangerous airspace so that the human pilot
doesn’t have to. Think of a star quarterback suddenly getting two extremely fast,
expendable wide receivers who never get tired and never complain about the play call.

To turn this vision into reality, the Air Force funded five companies to propose CCA
designs: Boeing, Lockheed Martin, Northrop Grumman, Anduril, and General Atomics. After
initial studies and prototyping, service leaders made a big call in April 2024:
Anduril and General Atomics would move on to the next phase detailed
design, test articles, and flight testing.

The goal is to choose at least one winner around 2026 and start receiving operational
aircraft before the decade closes, with ambitions to field the first tranche of 1,000
CCAs by around 2028.

What Exactly Is a “Robot Jet Fighter”?

“Robot jet fighter” is catchy, but the Pentagon’s preferred label is
Collaborative Combat Aircraft. That phrasing matters: these drones
are meant to collaborate with humans, not operate totally on their own as
free-roaming Skynet minions.

Here are a few key characteristics that define CCAs:

• Jet-powered and fast: The Air Force wants aircraft with turbofan
  engines in the 3,000–8,000 pounds thrust range, pointing to high-subsonic or even
  supersonic performance fast enough to keep up with frontline fighters.
• Relatively affordable: Target unit cost is roughly
  $20–30 million far cheaper than an F-35 (around $80 million) but still capable
  enough to be tactically useful.
• Highly autonomous but supervised: CCAs are being designed to perform
  many tasks on their own maneuvering, basic evasion, and some combat functions
  while human operators set objectives, rules of engagement, and mission plans.
• Manned–unmanned teaming: The buzzword is
  manned-unmanned teaming, where a single pilot may control or coordinate with
  multiple robotic wingmen that extend the reach and flexibility of the crewed jet.

This is not theoretical. The Air Force and DARPA have already run trials where AI
systems controlled virtual fighters and beat human pilots in simulated dogfights.
Those algorithms have since been flight-tested on a modified F-16 known as the
X-62A VISTA, proving that AI can handle real-world flight dynamics and rapidly changing
data, at least in controlled test regimes.

Meet the Finalists: General Atomics vs. Anduril

General Atomics: The Experienced Drone Powerhouse

If you’ve heard of the Predator or Reaper drones, you’ve heard of General Atomics.
The company practically defined the image of armed drones in the 2000s and 2010s,
supplying aircraft used extensively for surveillance and precision strikes in multiple
theaters.

For the CCA competition, General Atomics has proposed a Gambit family
of aircraft variations built around a common core design. Different Gambit models
would focus on missions such as:

• Long-endurance sensing and reconnaissance
• Air-to-air combat (carrying missiles to act as forward shooters)
• Serving as realistic “aggressor” aircraft for training
• Stealthy, long-range “combat reconnaissance” inside enemy air defenses

General Atomics has already flown a related drone, the XQ-67A Off Board Sensing
Station (OBSS), which serves as a stepping stone toward a family of common jet drone
designs. This “common genus” concept aligns closely with the Gambit vision and gives
the company real flight test data instead of just PowerPoint slides.

Anduril: The Software-First Disruptor

Anduril is the upstart in the fight a fast-growing defense tech company founded in
2017 by Oculus VR creator Palmer Luckey. Instead of traditional metal-bending, the
company made its name with AI-heavy software platforms like its Lattice operating
system, which fuses sensor data and helps classify and track targets in real time.

Anduril’s CCA contender, Fury, came through an acquisition of Blue
Force Technologies. Fury is a sleek, jet-powered drone concept initially designed as a
stealthy “adversary” to train fighter pilots. It’s expected to fly at high-subsonic
speeds near Mach 0.95, tolerate high-G maneuvers, and use modular hardware and
software so it can be quickly reconfigured for combat roles.

Anduril’s pitch leans hard on:

• Digital engineering and rapid iteration: heavy use of simulation and
  software updates rather than slow, hardware-centric upgrade cycles.
• Composite manufacturing and automation: aiming for fast, scalable
  production.
• AI-native design: building the drone around autonomy and networked
  operations from day one.

The contrast is striking: General Atomics is the seasoned drone manufacturer with
decades of operational history; Anduril is the software-first newcomer promising
Silicon Valley speed and disruption inside a traditionally cautious industry.

How the Competition Is Structured

The CCA program is unfolding in stages. Initially, five companies received design
contracts. Then, in early 2024, the Air Force narrowed that field to two by exercising
additional options only with Anduril and General Atomics.

In this current phase, the Air Force is paying both firms to:

• Finalize detailed designs
• Build production-representative prototypes
• Conduct ground and flight testing
• Demonstrate how their aircraft integrate with government-furnished autonomy software

By around 2026, the Air Force expects to choose which aircraft (or aircraft family) it
wants for Increment 1 the first major production batch. Future
“increments” could bring in new designs and even foreign partners, so the story won’t
end with this first downselect.

An interesting twist: the aircraft itself and the autonomy core are treated as
separate pieces. The government is developing core autonomy software through programs
like DARPA’s Air Combat Evolution and the Air Force Research Laboratory’s Autonomous
Air Combat Operations. The winning CCA airframe has to play nicely with that stack,
like a smartphone that must run a specific operating system.

Why 1,000 Robot Jet Fighters?

Why this specific scale 1,000 drones and not “a few dozen to see how it goes”?
Two reasons: threats and math.

First, potential adversaries like China and Russia are investing heavily in their own
autonomous combat aircraft and integrated air defenses. To fight effectively in a
high-end conflict, the U.S. needs enough assets to absorb losses, saturate defenses,
and maintain persistent presence, without risking every mission on human-crewed jets
that are expensive, slow to replace, and politically sensitive when lost.

Second, the Air Force’s own budget and fleet plans don’t allow it to simply buy
thousands more manned fighters. CCAs offer a way to increase combat mass at a fraction
of the cost per tail. At $20–30 million per drone, the total bill is still huge, but
it’s cheaper than expanding the fleet only with traditional fighters that cost
two to four times as much per airframe.

In other words, CCAs are a hedge: a way to buy “more capability per dollar” while
spreading risk across both humans and machines.

The Upsides: Why the Air Force Is Betting on Robot Wingmen

Supporters of the CCA push point to several major advantages.

1. More Firepower and Sensors in the Fight

A manned fighter has only so much space for weapons, fuel, and sensors. Add two or
more CCAs and suddenly you can carry extra radar-guided missiles, electronic warfare
pods, decoys, or surveillance payloads without redesigning the crewed airplane.

That means a single F-35, teamed with robot wingmen, could cover a wider area, track
more targets, or present more threats at once making it harder for an adversary to
defend against.

2. Lower Risk to Pilots

One of the stark truths of high-end air combat is that aircraft get shot down.
Replacing a lost drone is painful for the budget office; replacing a pilot is
impossible. CCAs are designed to take on the riskiest parts of a mission penetrating
dense air defenses, scouting dangerous areas, or absorbing the first wave of enemy
fire.

The Air Force hasn’t labeled these drones “expendable,” but officials have acknowledged
that their lower relative cost makes them “more attritable” than manned fighters,
meaning you can afford to lose some if the mission demands it.

3. Faster Reaction Times

In modern air combat, milliseconds count. An AI-assisted CCA can react faster than a
human in narrow situations for example, executing pre-approved defensive maneuvers,
de-conflicting formations, or handling routine navigation tasks. In theory, this frees
pilots to focus on tactics and decision-making while their robotic teammates handle
the “stick and rudder” details and some pre-defined combat actions.

The Hard Questions: Risks, Ethics, and Reality Checks

Of course, it’s not all smooth contrails and sci-fi marketing slides.
Building 1,000 robot jet fighters raises some tough questions.

Who’s Really in Control?

Pentagon policy stresses that humans must remain responsible for the use of lethal
force, but as autonomy improves, keeping that control meaningful not just a rubber
stamp gets more complicated. Researchers and ethicists worry about “automation
bias,” where humans overly trust the machine’s recommended course of action, especially
under time pressure.

That’s one reason the Air Force is moving cautiously: limited field experiments,
carefully framed rules of engagement, and autonomy that’s rolled out in stages, from
basic flight control up to more complex decision-support roles.

Can the Industrial Base Actually Build Them?

Announcing 1,000 CCAs is the easy part; building them in realistic timeframes is
harder. Prime defense contractors, newer firms like Anduril, and even other aerospace
players are racing to modernize their supply chains with composites, 3D printing, and
more automated assembly.

Some observers point out that recent programs have struggled with production ramp-ups.
That makes the CCA effort a key test of whether the U.S. can scale up advanced air
systems quickly enough to matter in a fast-moving crisis.

What About the Big Defense Giants Left Out?

Boeing, Lockheed Martin, and Northrop Grumman normally the A-list in U.S. military
aviation were all cut from the first CCA increment. That doesn’t mean they’re out of
the game forever, but it does signal a shift: the Air Force is willing to entrust
critical combat aircraft to companies outside the traditional handful of primes.

Future increments, Navy and Marine Corps programs, and export opportunities will all
provide chances for those giants to re-enter the arena with new designs or partnerships.

On the Flight Line: Early Experiences with Robot Wingmen

While 1,000 robot jet fighters haven’t appeared on the ramp yet, we already have a
glimpse of what life with CCAs might feel like from related experiments, test flights,
and exercises.

Take the X-62A VISTA program, where test pilots flew with AI agents controlling key
aspects of the aircraft’s behavior. One reported challenge was learning to
trust the machine: pilots described the AI flying “different but safe,”
taking slightly non-intuitive paths that still satisfied mission constraints. That’s a
preview of the human-machine interaction problem that CCA pilots will face:
understanding what the robot is doing and why, even when its logic doesn’t look like
a human’s instinct.

Training environments are also changing. Instead of a simple two-ship formation one
wingman, one lead pilots will train in mixed teams of multiple humans and multiple
unmanned aircraft. Simulators are incorporating virtual CCAs that can be tasked to
scout, flank, or act as decoys. Early feedback suggests that once pilots get over the
novelty, they start thinking about robot wingmen as extra “chess pieces” on the board,
allowing more creative plays and more redundancy when things go wrong.

On the engineering side, program managers talk about CCAs as a software-centric
experience. Instead of waiting years for major hardware upgrades, teams push frequent
software updates new autonomy behaviors, improved sensor fusion, refined
cooperative tactics. That sounds great until you realize it also demands a disciplined
test and safety pipeline. Every change has to be validated not just in a lab, but
across complex mission environments so that a clever new algorithm doesn’t accidentally
cause unintended behavior in the air.

There’s also a cultural factor. Traditional fighter pilots have built a professional
identity around being the “tip of the spear” humans in the cockpit making split-second
life-or-death calls. CCA concepts ask those same pilots to see themselves as
orchestrators, managing a networked formation where some platforms are
uncrewed. Early anecdotal reports from exercises suggest a gradual shift: pilots still
value hands-on flying, but they increasingly view their main advantage as judgment,
adaptability, and the ability to integrate multiple information streams rather than
pure stick-and-rudder skill.

Meanwhile, ground crews and maintainers are getting used to different maintenance
rhythms. CCAs are expected to be simpler and cheaper per aircraft than manned fighters
but potentially produced in greater numbers. That means more airframes to care for,
more modular swap-outs instead of deep depot-level maintenance, and more reliance on
digital diagnostics. Lessons from today’s drone fleets like the MQ-9 Reaper and
other unmanned systems are feeding into how maintainers think about servicing a
future swarm of jet-powered CCAs.

Finally, there’s the international experience. Allies and partners, from Australia’s
Ghost Bat program to European “loyal wingman” efforts, are experimenting with their own
robotic wingmen. Their lessons from how to structure
training, to how to write doctrine that clearly defines when a machine can act are
feeding back into U.S. thinking. The Air Force’s 1,000-drone vision doesn’t exist in a
vacuum; it’s part of a global race to figure out how humans and robots fight together
without losing control of the fight.

In short, the “experience” so far suggests that robot wingmen won’t replace humans, but
they will change what it means to be a pilot, a maintainer, or a commander. The
competition between Anduril and General Atomics is about hardware and contracts but
it’s also about shaping the day-to-day reality of future aircrews who will grow up
assuming that, of course, they have robotic teammates on the wing.

Conclusion: The Future Wingman Might Not Need a Flight Suit

The phrase “1,000 robot jet fighters” sounds dramatic, but it’s really shorthand for a
deeper transformation in airpower. By tapping Anduril and General Atomics to compete
for the first wave of CCAs, the Air Force is betting on a future where humans still
call the shots but increasingly from inside a network of autonomous teammates rather
than as lone knights of the sky.

Whether this gamble pays off will depend on more than just sleek airframes or clever
AI code. Success will hinge on trust, training, industrial capacity, and the messy
realities of budgets and geopolitics. What’s clear already is that the race is on:
nations that master effective human-machine teaming in the air will shape how future
conflicts are deterred or, if deterrence fails, how they are fought.

For now, two very different companies have a rare opportunity: build the world’s first
large fleet of robot jet fighters that real pilots actually want to fly with. The rest
of us will be watching closely to see which vision of the future wins and how long
it takes before “robot wingman” sounds less like a headline and more like everyday
Air Force jargon.