F-22 Raptor vs. Chengdu J-20: What the Public Evidence Actually Shows

These two aircraft have never met. No gun camera footage, no wreckage, no after-action report. The most argued-about matchup in modern aviation is a fight that has never happened, judged mostly by people who have never seen the file.

That has not slowed anybody down.

So let's do something harder than picking a winner. Let's separate what the open record actually supports from what the internet has been repeating to itself since 2011.

Nobody arguing about this has the real numbers

Every figure that would settle the question is classified by somebody. Radar returns at specific angles. How far each radar sees a low-observable target. How well the jamming works. How fast a missile bleeds energy at 40,000 feet.

What leaks out instead is a patchy trail. Satellite photos of factory rooftops. Tail numbers shot through a fence. A scale model on a convention floor. A general who says one sentence more than he planned to.

It is real evidence. It is just not the evidence most comparison videos pretend to have. So hold every number here loosely, including the famous ones.

And notice something before we start: one of these programs is audited in public. American fleet size, readiness rates, upgrade costs and retirement fights are argued out in congressional testimony and watchdog reports, in the open, in embarrassing detail. The other is reconstructed by outsiders from serial numbers and commercial imagery. That asymmetry shapes the entire debate, and it consistently flatters the side that publishes nothing.

The Raptor: the airplane nobody has been willing to fight

The F-22 is a Cold War answer that arrived after the Cold War ended, and it got billed accordingly. The production line closed in 2012 after 195 aircraft including test jets, down from a planned 750.

The fleet math is softer than the headline suggests, and the Air Force says so out loud. There are roughly 185 Raptors on the books but only 143 are combat-coded, and 32 of those are training-only Block 20s that fly the overwhelming majority of initial F-22 pilot training. The service has tried for years to retire them. Congress keeps refusing, and upgrading them instead would run at least $3.3 billion over roughly fifteen years (TWZ, 19FortyFive).

Keeping it healthy has been expensive. Mission capable rates fell to around 40 percent in 2024. A stealth coating is not paint. It is a maintenance contract with wings.

Now the part that makes 2026 interesting. The Raptor is being rebuilt in mid-life, and the list is essentially every capability it was supposed to have in 1997 and never got:

• Stealth-shaped external tanks it can carry into a fight instead of jettisoning at the door, adding roughly 850 nautical miles of reach
• Infrared search and track pods under the wings, letting it hunt heat signatures without radiating
• A helmet-mounted display, finally
• Link 16 that can transmit instead of only listen
• A new long-range air-to-air missile

The clearest public images of the new configuration were photographed in March 2026, and the overall package has been reported at around $11 billion (Army Recognition, SOFREP). On the missile, exact range is classified, but Aviation Week's Steve Trimble noted that the test range circle for the AIM-260 is roughly double the AMRAAM's.

Then there is the record. In twenty years of service the Raptor's only air-to-air shot was a single AIM-9X fired at a Chinese surveillance balloon off South Carolina in February 2023, from 58,000 feet.

People use that as a punchline. Read it the other way. The F-22 has flown over Syria, over Iraq, along the edges of Russian and Chinese airspace for two decades, and no air force has chosen to test it. Syrian fighters broke off when Raptors lit them up. That is not an empty resume. That is the entire point of the airplane, achieved without firing.

The Mighty Dragon: thirty years to solve the engine

The J-20 first flew in January 2011 and entered service in 2017. For most of that run it flew on engines that were never meant for it, starting with Russian AL-31 variants and then the interim domestic WS-10C.

The engine story is the one that gets skipped, and it is the most revealing part of the program. The WS-15 was started in the early 1990s. It reached serial production roughly three decades later, years after the airframe it was designed for was already in squadron service. Chengdu released images of a J-20A flying on WS-15s in January 2026, with the engine estimated in the 161 to 180 kN class against the F-22's F119 in the lower 156 kN range (Army Recognition, 19FortyFive).

On paper that closes a long-standing gap. In practice, a brand-new high-performance turbofan is at the beginning of its reliability curve, not the end. The F119 has three decades of fleet hours, hot-section data and failure modes behind it. The WS-15 has a press release and a production line. Those are different things, and every air force that has ever fielded a new engine learned the difference the hard way.

The genuinely novel item is the two-seater. The J-20S puts a second crew member in back for sensor management, electronic warfare and control of unmanned aircraft, and airframes wearing front-line unit serials appeared in 2025 (TWZ, Japan's National Institute for Defense Studies). Fielding an aircraft is not the same as fielding a working concept, and nobody outside the PLAAF knows how much of that rear cockpit is operational versus aspirational. But it is in the air, and that deserves to be said plainly.

About those production numbers

This is where most articles drive off a cliff, so be careful.

Nobody outside Chengdu knows how many J-20s exist. The figures in circulation are reconstructions built from identified units, serial numbers and assumed industrial rates. The last firmly documented threshold was around 300 aircraft in late 2025. Estimates running toward 500 are analyst extrapolations, and some of those airframes may be test articles, replacements, or built but not yet operational.

The production rate is the number that actually matters, and it has moved fast. Satellite imagery analysis presented at the Air and Space Forces Association symposium supports roughly 100 aircraft a year, with RUSI putting 2025 output near 120, against public estimates of 40 to 50 a year as recently as 2024 (19FortyFive).

When estimates move that far that quickly, treat the direction as the signal and the absolute number as noise. Capacity is also not the same thing as output, and a parade flyby is a political event before it is a capability disclosure.

The canard argument that will outlive all of us

This is the oldest technical fight about the J-20, and it is worth understanding rather than repeating.

The case against those forward wings is circumstantial and strong. No American stealth aircraft has canards. Even the YF-22 and YF-23 prototypes dropped them despite earlier stealth studies including them, which suggests US designers tested the idea and rejected it. Canards sit out front with nothing shadowing them, throwing returns in directions a tail tucked behind a wing never would.

The defense is more careful than "China figured it out." The canard leading edge is swept to match the wing, so the resulting spikes get dumped in the same direction instead of scattered, and some shaping analyses put their contribution in the minor range.

The sharpest rebuttal is also the fairest. The F-22 has radar spikes too, because perfect stealth on something that has to fly is impossible. The difference is placement. Published analysis holds that the Raptor's spikes sit where an opponent cannot practically exploit them, while a design with predictable off-center returns has to keep managing its geometry against every emitter in the sky. That is a permanent tax on how you fly the mission.

Where the evidence lands: frontal shaping on the J-20 is credible and taken seriously by serious people. All-aspect performance, particularly from the rear, is where published assessment still favors the F-22, whose frontal radar cross-section Lockheed Martin has publicly characterized at around 0.0001 square meters. Open-source estimates for the J-20 vary by more than an order of magnitude, which tells you exactly how much of it is inference from photographs.

The fight is not where you think it is

Here is where the dogfight framing collapses.

On internal fuel the Raptor's combat radius is commonly put in the 460 to 600 nautical mile band. The J-20 is a substantially bigger airframe with substantially more fuel. That gap is not a bragging right, it is a different theory of war.

The F-22 was designed for a theater with short distances and secure runways. The Pacific offers neither. The explicit purpose of China's long-range missile investment is to push American airpower to operate from farther out, which increases reliance on tankers that can be engaged at ranges no fighter can match, as a February 2026 RUSI paper lays out.

So the J-20 does not need to beat a Raptor. It is widely read as shaped to hunt the aircraft that keep the Raptor airborne. Kill the tanker and you have ended the mission without ever seeing the fighter.

Which is exactly why those new stealth tanks matter more than they look. They are not a range upgrade. They are an attempt to buy back independence from the most vulnerable asset in the American formation, and they are the single most consequential thing on the Raptor 2.0 list.

The missiles stopped being theoretical last year. The export PL-15E saw its first combat use in May 2025, confirmed when debris came down inside India, including a near-intact round recovered for analysis (overview, Army Recognition).

Notice what that proved and what it did not. The weapon flies and the seeker is sophisticated. It did not produce a clean record of kills in that exchange, and the recovered hardware is now being taken apart by engineers who are not friendly to the manufacturer. A missile's brochure and a working kill chain are two separate achievements, and the gap between them is where air wars are decided.

The advantage that does not photograph well

A US Air Force chief of staff framed this better than any spec sheet. Fifth-generation capability stopped being about a platform and became a family of systems.

"It's about a network and that's what gives us an asymmetrical advantage," which is why he treated the head-to-head question as close to irrelevant. (Gen. David Goldfein)

On that measure the scoreboard looks different. Hundreds of F-35s flying alongside the Raptor. Allied air forces operating the same jet off the same data links. Airborne early warning, electronic attack and tanking built over decades of joint operations. Aircrews who have flown real combat sorties, continuously, since 1991.

The PLAAF has not fought an air war in living memory. Its readiness rates, sortie generation, maintenance depth and training hours are entirely unpublished, which means the most flattering possible assumptions get applied to them by default. American readiness problems are public because Americans publish them. That is a feature of the system, not evidence that the other side does better.

The closest thing we have to real data

One moment still anchors every serious conversation about the J-20. In 2022 the commander of Pacific Air Forces described F-35s getting relatively close to J-20s over the East China Sea, and said the US side was "relatively impressed with the command and control associated with the J-20."

Read what he actually praised. Not the airframe. Not the stealth. The command and control. In the same remarks he pointed at China's airborne early warning aircraft as central to their long-range kill chains.

That is a four-star telling you the fighter is the visible tip of a larger machine, and quietly telling you the machine is the part worth studying.

Straight answers

Question	Where the public evidence points
Stealthier overall	F-22, clearly from the rear and sides. Closer head-on than most people assume.
Close-in agility	F-22. Thrust vectoring plus two decades of exercise results.
Raw thrust today	J-20A on estimates, with an engine at the start of its service life.
Unrefueled reach	J-20, and it is not close. This is the Raptor's real weakness.
Sensor fusion and pilot workload	F-22, most likely. Also the largest blind spot in the public record.
Fleet size and build rate	China, by a growing margin, with wide error bars on the total.
Combat-tested weapon	The PL-15 family has flown in anger. The AIM-260 has not.
Drone teaming fielded today	The two-seat J-20S is flying. Effectiveness unknown.
Operational experience and supporting network	United States, and it is the widest gap on this list.

So who wins?

Take the question seriously and the answer is still the Raptor.

Equal pilots, equal support, both aircraft merged and fighting for their lives, and the public evidence favors the F-22 on shaping, agility and maturity. Nothing in fifteen years of J-20 imagery has changed that, and the people who would most enjoy changing it have not managed to.

The problem is that nobody is planning to fight that engagement.

The J-20 was not built to win the duel. It was built to make the duel unnecessary, by reaching past the fighters to the fragile things behind them, in numbers, close to home. Whether it works depends on sensors, jamming and kill-chain resilience, which happen to be the three things the public record cannot see. Anyone answering with total confidence is selling a thumbnail.

The honest conclusion is not that America has been passed. It is that the comfortable margin is gone, and the response is already moving. The first F-47 is in build, targeted for a 2028 first flight, planned to replace the Raptor roughly one for one in the early 2030s (Aviation Week, TWZ). China has had sixth-generation prototypes flying since December 2024.

The argument has moved from airframes to factories, networks and distance. Which is a harder story than who wins the merge, and a more important one.

It also does not make either aircraft less beautiful sitting on a desk.