Saab achieves AI milestone with Gripen E | Hacker News Reader

Saab achieves AI milestone with Gripen E

saab.com

77 pointsfnordsensei57 comments

Saab also work with Hopsworks - the only EU owned/based AI platform for developing and operating AI systems at scale. Saab is Palantir-free.

Refernece: https://youtu.be/A7FPenztrcw?si=-_s-kQ_D-nCUWJWh&t=2276 Disclaimer: i work on Hopsworks

What kind of models do these run? Vision? What “predictions” are being made?

Multimodal most likely. They are able to predict time to fire accurately and are able to do complex manouvres in the air. Sounds extremely impressive to me.

Maybe in the future wars will involve sending multimodal LLMs in charge of fighter jets to strike enemy country's infrastructure targets with minimal loss to human life.

The risk being that your enemy cannot reasonably steal all your fighter jet airmen but the can absolute steal a transformer model.

LLMs? Sounds backwards to me. It’s not like there is a lot of text to process up there.

LLMs work on tokens. There's no reason those tokens can't be something other than text.

> and are able to do complex manouvres in the air

If there’s something pilots really don’t like, it’s when someone else does things without them knowing exactly what, when and why. There’s a reason you separate pilot flying from pilot monitoring.

Automated systems are obviously crucial but they need to be so reliable and predictable that the pilot can form a mental model of how they work. If not, even simple systems like MCAS can wreak havoc. It’s like having a silent copilot fiddling on the controls - even the ”right” decision can cause problems. This is similar in AVs - when you have L3 autonomy you end up in awkward inbetween split-brain in-between states, that can result in new failure modes.

> Maybe in the future wars will involve sending multimodal LLMs in charge of fighter jets

Probably not LLMs but yeah, it feels like this is the type of tech that belongs in a UAV without life support systems. We are probably in the last generation or two with humans in the cockpit anyway.

There is slightly more information on Helsing website: https://helsing.ai/centaur

They should replace some of the panels with machined metamaterial and give it some RF absorbing paint. Would sell like hotcakes in the current political climate.

I don't think they have much issues with their order backlog currently. Sweden just joined NATO and every NATO member is trying to modernize their military and is increasing spending big time.

This demonstration is interesting. Basically they are trialing the concept of having planes without pilots. Which of course turns them into big drones. Lots of people talking about this but not a lot of companies demonstrating they can do it. I think that might stimulate sales a lot more than some panel work.

With the Embraer partnership[1] and the additional production capacity _and_ demand it generates, a hotcake situation is looking increasingly more realistic for Saab.

1: https://embraer.com/global/en/news?slug=1207196-gripen-e-pro...

What about blockchain though? Does it have that?

I really hope the Gripen stays the way it is - just solid engineering with a solid scope for a solid price point, and doesn't turn into the feature-creep, over engineered , money-gobbling monstrosity that is the F-35.

I really wish that Germany would have gone with something like the Gripen instead of buying the very few white elephants that won't actually make a difference should it come to a real war with Russia. Bonus points with the friendly Scandinavians, too, and no dependence on the madman in the white house.

These purchases was made in another time.

However, given the current political climate mixed with nato spending that will likely go up to 5% will likely not be bad for European military suppliers (and their supply chains).

Yes. Threatening to block export of parts to pressure other countries into buying f35 guarantees no one will ever pick american made engines again.

Unfortunately the Gripen also uses a lot of US parts.

Unlike the Rafale ! That’s a true European jet

> buying the very few white elephants that won't actually make a difference

There’s a lot to criticize in the f 35 program but there seems to be enough there there for it to succeed markedly in the export market - beyond Germany there is Poland, Czech Republic, Italy, Denmark, Finland, Norway, Netherlands, Belgium, Switzerland, Greece, Romania and UK — and that’s just in Europe. This includes several countries that did not historically equip with US kit (and excludes Turkey, which ordered it before it got sanctioned). You’re saying all those countries have made a huge mistake?

They made a political purchase at a time when they didn't expect to have any real use for those planes, so the actual capability price ratio didn't really matter.

In that case, would Gripen have been a wise order either? Germany already has Typhoon. There is a history of Typhoon equipped countries ordering f35 because it’s a generation ahead. But Gripen is same generation as Typhoon.

Probably not, you're right, the logical choice, from a capability perspective, would have been a modernization of the Eurofighter and/or and the pursuite of a Next Generation European Fighter program. (Which could have been much less expensive than the f-35 program if given less exotic requirements).

The Eurofighter was not a logical choice because Germany couldn't afford to get it certified for participation in the NATO nuclear sharing program. That would have driven the cost way higher than the F-35A.

The ability to deploy US nukes has been useless since the end of the cold war, and now that it's starting to get relevant again, it's not reliable anymore.

And how is Germany supposed to integrate French nukes on the F-35 if Scholz wants to pursue “strategic independence” now?

Also, it was likely not a matter of cost, as the US has nothing to gain from allowing Germany to use a European plane to carry their nukes in the first place when they can sell them US planes instead.

Cost was absolutely a factor. The Panavia Tornado was previously certified to carry US nuclear bombs. The same could have been done with the Eurofighter but Germany would have had to pay 100% of the program cost because no other NATO member was interested.

I can't see any scenario where France would agree to share nuclear weapons with Germany. The only approach that might work would be to forward deploy French nuclear forces on German territory, with Germany picking up the cost.

> The Panavia Tornado was previously certified to carry US nuclear bombs

During the cold war. That's a very important bit.

> The only approach that might work would be to forward deploy French nuclear forces on German territor

Are you sure you know how the current program works? Because that's how it does. The US aren't sharing nukes right now (that would violate NPT), the plan would be to hand off the weapons to its allies in the event of an invasion.

Ah interesting, I didn’t know about the next generation, link for anyone else interested -

https://en.m.wikipedia.org/wiki/Future_Combat_Air_System

FWIW I think the US should be more willing to buy euro fighters, because the number of US suppliers is getting smaller and they have too much leverage over US government.

Gripen uses an engine subject to US export controls, so unfortunately not totally free from dependence.

Which is why Gripen hasn't been sent to Ukraine. (US blocked it.)

Gripen is more expensive and less capable than F-35.

Define "less capable".

No stealth, and a questionable sensor suite - which cripples its SEAD and BVR capabilities.

Notably, those are the capabilities that modern near-peer warfare stresses most heavily.

"Cripples" might be stressing things a bit hard. Modern near-peer warfare is network-centric - you don't need your radar to be perfect if your AWACs can share a 300km lock over Link 16. Similarly, active radar SEAD is just one option, and passive homing/home-on-jam should also be fine for anti-radiation strikes (though I still wouldn't want to send a Gripen on a Wild Weasel mission). It isn't a Marine or Navy plane, so it really doesn't have to fill the Joint Strike Fighter role in it's entirety.

I'm a huge F-35 fan myself, so I'm pretty hard-pressed to undersell the plane compared to other single-engine fighters. The Gripen isn't equipped to handle contested airspace whatsoever, the F-35 is absolutely king in that department. That said, a Gripen armed with Meteors is a mean payload even compared to an F-15 lugging around AMRAAMs. Unless you have expeditionary/naval operation roles to fill, the Gripen isn't lacking much that a normal customer would want.

Is there a source on the price that is not speculative? There seems to be several opinions about the pricing of the F-35 actually going down or not.

As for the long run I've read that the Gripen is cheaper to run.

Parent is very mistaken. The F-35 is famous for taking a long time to turn around, for long repair times, and being non-mission capable frequently.

It's also absurdly expensive to maintain. it requires special hangers and maintenance bays, all up to US requirements.

Even the US military has to use Lockheed to do all the work.

The F-35 is a pork trough, designed to provide employment and jobs more than defense.

And anyone buying it is locked into wonderous maintenance costs, all benefitting US employees.

One telling sign of this is that Gripens are designed to be maintained anywhere. You know, like when you're in a thing called a "war", and your bases have been blown up?

You can turn around some Gripens in 30 minutes with them landing on a highway.

Try that with a princess like the F-35.

I prefer to have planes which are designed to be in the air, instead of to provide jobs.

The latest report on the F-35 by the Director, Operational Test and Evaluation [1] (typically referenced as DOT&E) describes an aircraft that struggles to meet its operational goals, with under 50% of the fleet fully mission capable at any given time. In short, the F-35, while a very capable military aircraft when fully operational, is also a "hangar queen".

1. https://www.dote.osd.mil/Portals/97/pub/reports/FY2024/dod/2...

Or even better:

https://www.pogo.org/analysis/f-35-the-part-time-fighter-jet

"The F-35 fleet can only perform the full range of its combat roles 30% of the time. This unreliability renders the entire program ineffective."

"The services consider an aircraft as mission capable if it can perform at least one of the program’s assigned missions. Such a threshold may be appropriate for a program like the C-17 transport, which has essentially a single mission. For a multi-role program like the F-35, however, a different standard should be used. Because the F-35 is designed to perform many missions, from delivering nuclear weapons to supporting troops on the ground, program officials aren’t even using the right yardstick to measure the aircraft’s performance.

Fortunately, such a yardstick does exist. It is the full mission capable rate, or the percentage of aircraft available to perform all the assigned missions. The testing director said the full mission capable rate standard is “a better evaluation of combat readiness” for the F-35 program. When this higher standard is applied to the F-35 fleet, the magnitude of the program’s failure becomes clear: DOT&E reports the full mission capable rate for the F-35 fleet was 30% in 2023."

Essentially, if a plane they approach cannot perform the role, they don't mark it as bad and move on until they find a plane capable, as long as the F-35 can perform a role.

Not very helpful if you need all planes for combat, or troop support immediately. Not very helpful if, unlike the US, you only have 70 planes spread across 5 bases.

F-35s shot down: 1

Gripens shot down: 0

Q.E.D. /s

This comes across as really uniformed. Gripen is a pinnacle of aerospace engineering, providing abilities to handle unique infrastructural and geographical scenarios present in Sweden and northern Europe. It covers all abilities you would otherwise get with an F-35 at a third of the cost. It is simply remarkable how genius it is.

It lacks the range and, most significantly, low-observability of the F-35.

Gripen is a 1980s design updated with the latest software, just like Typhoon and Rafale. The F-35 is a 2000s design, there's just no way to close the stealth gap with it.

Stealth gets a lot less effective when either bistatic or passive radar is used. I'm fairly certain that we'll find out in the long run that it's been countered by our adversaries.

This vulnerability is why KrakenSDR had to pull one of their git repos a few years ago.

Stealth might help at long ranges, but think of it like locks on doors, only effective at keeping honest people out.

This kind of thing is why ECM and stealth are not substitutes for each other. They work best when they work together.

Every number I find shows that the "flyaway" unit cost for the Gripen E is about 4% higher than that of the F-35A (the closest comparable model), but its "cost per flight-hour" is a mere fraction.

Germany wanted to continue participating in the NATO nuclear sharing agreement so that made the F-35A the only practical option. They considered purchasing the Eurofighter Typhoon instead but found that modifying it for the nuclear strike mission and then going through the certification process would be too expensive. When you have to rely on another country for your nuclear deterrent that cuts off your options.

The Gripen is fine for what it is. But it's very small with limited thrust so performance goes to crap when you load it up with a bunch of external stores for a long range strike mission.

The F-35A is only like 40% more expensive per-unit than the Gripen and is arguably at least 40% more capable (the payload alone beats that). It's true that the development program was a boondoggle (though Saab isn't completely clean here either), but that cost isn't for the most part borne by the export customers.

Economy of scale continues to rule all, basically, and there are already 3x as many Money-Gobbling Monstrosities in the air as there are Gripens.

Per unit purchase cost is the wrong metric for something which will be used for decades. Compare operating costs and availability, and the Gripen gives you drastically more bang for your buck.

It sounds like it has some kind of ML based prediction system. It’s not like they put ChatGPT in there.

Bummer, I was really looking forward to that.

“Oh you are correct, it is a school, I see it now »

Helsing has an ex McKinsey (which ruined the German army) CEO. McKinsey is pushing AI:

https://www.mckinsey.com/industries/aerospace-and-defense/ou...

Until we hear from captains who actually fly the aircraft, we will not know if this technology even works. Of course they will not be allowed to speak freely.

In Germany, the clueless Blackrock chancellor will of course back up all of this.

I'm not judging the guy, but usually [recent] Series C level companies [0] (i.e. fast-growing companies) have good leadership in place, I would think their leadership is still good.

[0] https://techcrunch.com/2024/07/11/defence-ai-startup-helsing...

Was expecting a new player with level 3 self driving cars.

Was not expecting planes.

Saab Automobile AB went bankrupt in 2012.

NEVS (who bought the assets in 2012) still built Saab branded vehicles until 2014, and there was continued hope that Saab AB would again allow NEVS a license after they revoked it at that time. But NEVS effectively dissolved in 2023. Though there's no reason why Saab AB couldn't start using the name again themselves, or issue a license to someone else.

As a proud ex-owner of a Saab 900, let's hope they never do.

Nah, I had an old 900 Turbo, best car I ever owned (at least subjectively, obviously fewer features than modern cars).

That's what I mean! I adored my Saab. If 'they' make something now can you imagine what it'd look like.

My favourite car of all time is the Volvo 850 T-5R. In cream yellow. ;-) And now look at Volvo: the same bland SUVs as everyone else.

May Saab rest in peace with its integrity unblemished.

Ah, I gotcha. You're right, marketing people would just release something boring, even enough I think a new quirky 900 would do well with modest expectations.

Which has nothing to do with SAAB AB, the defence company.

SAAB Automobile was split out from SAAB AB in 1990 with General Motors (GM) taking a 51% stake, and was fully a part of GM ten years later. GM then tried to build SAABs on GM platforms which meant the quality tanked and tanked the company too. And as another posted, What was left became Nevs.

Wikipedia has a good writeup on SAAB, with its many divisions. It's a bit like Volvo. Both companies have had divisions that makes automobiles, heavy vehicles and other types of products. Volvo Cars was sold off from Volvo AB to Ford. The heavy vehicle division of SAAB (Scania) is now part of Volkswagen

https://en.wikipedia.org/wiki/Saab_AB

Reply on news.ycombinator.com