First, these are intended to be "loyal wingman". They'll be commanded (but not really remotely controlled) from manned fighters nearbyish. Presumably, the "shoot authorization" will be delegated down to the pilots.
Secondly, the actual unmanned platform (the Kratos Valkyrie) is also part of a program of record for the USMC (US Marine Corps) to act as a partner SEAD (suppression of air defence) vehicle.
Thirdly, the "MARS" system chattered about looks to be Airbus' open architecture /system of systems pitch that they were developing for FCAS (the European 6th generation fighter program). MARS and all pitches like it are about ways to make individual platforms as software defined as possible, and to get different platforms/instances to really data/function share as much as possible.
If this program goes well, it shows that Airbus' MARS has the flexibility and capability required to just... layer into/ontop of some random other vendor's hardware/software and then "just work". I think it would be major demonstration/validation of the work.
Like when Blackberry tried to make BlackBerry Storm after iPhone and Blockbuster tried to make Blockbuster Online after Netflix.
Technology shifts rarely wait for these stodgy middle ground transitionary products to find a market.
The real question is basically - is full autonomy both technically possible and culturally/politically acceptable within 5, 10, or 20 years? Because full autonomy isn't really ready now (or else we wouldn't need hundreds to thousands of drone operators in the Ukraine war). And at least the USAF doesn't think remote control will let them do what they need (which is to fly from Japan to Korea or Taiwan, or Philippines to Taiwan, and contest/control the skies in the face of a basically peer adversary).
Because no one knows that answer, everyone (governments, militaries, manufacturers) is hedging, and CCA is part of that hedge.
>or else we wouldn't need hundreds to thousands of drone operators in the Ukraine war
I don't think this is the reason the systems are not fully autonomous right now ("fully autonomous" here meaning that they can complete the kill chain independently, no HITL). Even if we assume it true that the drones are not "good enough" to be at parity with a human operator, if you had an essentially limitless amount of them, would you really waste the manpower on operating them in FPV mode? You would not, you would completely saturate the battlefield with them. Thus, as it was in the beforetimes and ever shall be, logistics wins wars.
Disrupting the signal for a normal-sized aircraft is much harder. If you're flying at 10s of thousands of feet and have a line of sight to multiple satellites it's going to take some serious weaponry to disrupt that.
The G forces are another thing. I wonder why they aren't stsrting wth missle platforms instead.
Sure, winged flight has uses, but taking a missle platform, adding small munitions instead of a big bang?
Only if every mission is absolutely critical. If disruptions are rare then you don't need autonomy.
I mean, they wouldn't think that, would they? It would put their pilots out of a job. But most flying has been done by autopilot long before AI, and even if/when you need a human in the loop, why would you want to put that human in the cockpit rather than safely in Virginia?
If you can outsource the radar on a jet it is not a huge leap in logic to put the very hot missiles onto a unmanned aircraft. All of these concepts where written up 20 years ago by both china and the US
'loyal wingman' gives the kill / no kill decision to an Air Force officer. And having the decision maker geographically close eliminates jamming, delays, and the requirements to have a satellite infrastructure (like is required for Predator UAV's).
i hope we never assign a piece of code, AI or not, to be the decision maker.
This is already a past station, just not at Airbus.
Airbus has publicized that it is working on a Project Maven style project with France's DGA [0][1].
Thales also publicly launched and demonstrated SkyDefender a couple days ago [2].
Mistral AI also announced in January 2026 that it is working with the DGA to productionize it's models for military applications [3] - ironically similar in manner to how the DoD was using Claude but is now using Gemini and GPT.
No country is going to leave networked, autonomous offensive and defensive capabilities on the table.
[0] - https://www.reuters.com/business/aerospace-defense/airbus-wi...
[1] - https://www.janes.com/osint-insights/defence-news/defence/ai...
[2] - https://www.janes.com/osint-insights/defence-news/air/thales...
[3] - https://www.linkedin.com/posts/marjorietoucas_were-happy-to-...
https://www.vp4association.com/aircraft-information-2/32-2/m...
https://aviation.stackexchange.com/questions/6312/why-are-ai...
"Airplanes are riveted, not screwed because they are the product of engineers, not lawyers."
This is actually an important part of what makes a stealth airframe "stealthy", along with other stuff.
Ukraine and Iran are both showing it quickly becomes a war of attrition and fancy weapons get very expensive very fast, or run out very fast.
Land warfare is next on the list: https://time.com/article/2026/03/09/ai-robots-soldiers-war/
I have more trust in Airbus than the PayPal mafia though.
So this is Skynet v0.1?
These drones are "helpers" for fighter jets. It's a type of role that is still in development (no one has an operational collaborative combat aircraft as far as I understand), both technically and in concept.
But the basic idea is that you'll have drones that can somewhat keep up with your fighter jets and help it do stuff that might be too risky. Maybe fly ahead, or be the one with the active emissions or sensors or whatever. Or maybe it's just a way to increase the amount of ordnance/sensors you can fly per sortie / generate from a given amount of training/flight hours in a year.
* The Kratos Valkyrie with the USMC in a SEAD role
* Anduril (YFQ-44) and General Atomics (YFQ-42) are battling it out for the USAF's CCA Increment 1 contract (we're apparently supposed to get a decision on that this year) - with Increment 2 probably getting spun up pretty soon
* USN has the Boeing MQ-25 as an drone tanker... once that gets up the speed, I'm fairly certain it's going to morph into something strike capable
Elsewhere, Boeing Australia's Ghost Bat seems to be doing well as well.
Also, let’s see how the 43% holds up when European and gulf states do their next round of procurement.
Wars are always bad news and robot wars are very bad news. Many countries will fall into an endless war economy.
There were more wars before any type of mechanisation of warfare, with the only slow down really happening after nuclear weapons were developed.
yes but they weren't comparable. With the exception of ancient Chinese wars which are a bit of an odd case given the population sizes and that they kept sending farmers to the front until everyone starved, European pre-modern wars consisted of small armies and relatively low civilian casualty ratios.
It's this and the late 20th century that saw civilian death ratios climb up to 80-90% in mass bombing campaigns and urban warfare environments. People like to use 'medieval' as an insult but the medieval age was quite constrained compared to Gaza. And if you take the pilots out of the equation and fully automate this, that's probably only a taste of what people will do to civilian populations.
Because a picture says more than words, this is the kind of thing you can probably look forward to:
Cities were routinely razed and famines and disease killed scores of people in historical warfare as well - we have the accounts, we know it happened. The "difficulty" of implementing any of this was enormous given the lack of modern logistics or simple things like refrigeration to keep armies resupplied.
How does this support your argument though? World War 1 increased the level of danger and destruction of warfare and...then we had World War 2. If the hypothesis was that making war easy leads to more wars, then no example presented shows that because WW1 was at the time the most destructive war in history and simply set the stage for an even more destructive war.
(I'm pretty sure Musk could make them reusable. /s)
- Even if the XQ-58 has a low radar cross section, a swarm of four drones flying in formation with a non-stealthy Eurofighter significantly increases the aggregate probability of detection. Unless these drones are performing active electronic countermeasures or "blinking" to spoof radar returns, they’re essentially a giant "here we are" sign for any modern radar. I wonder if they've compensated via the flight software to manage formation geometry to minimize the group's total observable signature?
- Anti-air systems will prioritize the command aircraft (the Eurofighter) immediately. If the C2 link is severed (kinetic kill, high-power jamming) what is the state-machine logic for the subordinates? Do they revert to a fail-passive (return to base) or -active (continue last assigned strike) mode? Without a human-in-the-loop, rules of engagement issues are abound. (I'm not even accounting for the fact that the drones probably rely on calculations from the command craft, so edge-computing will factor in as well.)
- They're calling these "attritable," but at $4M a pop plus the cost of the sensors, they aren't exactly disposable. Is the cost-per-kill for an adversary’s interceptor missile actually higher than the cost of the drone it's hitting?
(2) Remains to be seen.
(3) Individual Patriot missiles are around that price point, with S300/S400 anywhere from 500k-2M depending on capability. One would think that cost-per-kill would be favorable considering the increased capability granted.
You're right about them both costing about the same, so the real leverage only comes if these drones can stay out of the engagement envelope while sending cheaper submunitions (likely using something like these Ragnaroks (~$150k) https://www.kratosdefense.com/newsroom/kratos-unveils-revolu...) to do the actual baiting.
When you're high up you can have pretty long 'line of sight' so it's not unreasonable that these could fly way way ahead. 100 miles and way more is not unreasonable.
You basically get 'double standoff'.
I can see this as being almost as effective as manned stealth and if they are cost effective they could very plausibly defeat f22 scenarios.
Once you add in the fact that risk is completely different (no human), then payload, manoeuvrability, g-force recovery safety, all that goes out the window and you have something very crazy.
3 typhoons with 2-3 'suicidal AI wingmen' each way out ahead is going to dust them up pretty good at minimum. It's really hard to say for sure obviously it depends on all the other context as well.
If the unmanned aircraft are the ones flying far ahead, taking the risk, and extending the standoff envelope, then why is the human still sitting in the forward fighter rather than supervising from a safer node further back?
At that point it seems like the architecture is optimizing for tactical latency and current doctrine, not necessarily for the cleanest end-state.
Hard to win at jamming, when you're further away and the opponents are frequency agile.
1. They can use directionality more effectively to their advantage
2. Inverse square law works against you (unlike e.g. jamming GPS where it works for you).
3. They can be frequency agile, strongly rejecting everything outside of the 20MHz slice they're using "right now"-- and have choices of hundreds of those slices.
Fighters already have radars that they expect to "win" with despite that being inverse fourth power, a longer range, and countermeasures. They can send communications-ish signals anywhere over a couple GHz span up near X-band. Peak EIRP that they put out isn't measured in kilowatts, but tens of megawatts.
My concern is less total link loss than what happens under degraded or intermittent connectivity. If the wingman still depends on the manned aircraft for tasking or weapons authority, then the interesting question is how it behaves when the link is noisy rather than gone.
That feels like the real hinge in the concept.
https://en.wikipedia.org/wiki/Multifunction_Advanced_Data_Li...
