Best of LinkedIn: Defense Tech CW 36/ 37

00:00:00: This episode is provided by Thomas Allgaier and Frennus, based on the most relevant LinkedIn posts about defense tech from CW-ThirtysixandThirtySeven.

00:00:09: Frenness is a B to B market research company that supports ICTU providers with a structured defense market entry framework designed to move them from European defense opportunity landscape to qualified ministry conversations.

00:00:21: within six to eight weeks.

00:00:23: you can find more info in

00:00:29: a decade and literally billions of dollars developing this state-of the art AI targeted tank.

00:00:35: Right, The Ultimate Weapon?

00:00:36: Exactly

00:00:37: Only for it to be completely blinded by like fifty bucks worth hardware store mirrors And a few cans of zebra striped paint.

00:00:43: I mean It sounds totally absurd when you say that out loud.

00:00:45: Yeah But that is, uh... That's just the reality of modern conflict right now.

00:00:48: If you work in digital transformation or enterprise IT You probably still picture defense as this really slow heavy metal industry

00:00:55: For sure like decades-long procurement cycles for aircraft carriers and stuff Like that.

00:01:00: Right

00:01:00: but that era was effectively over.

00:01:01: Yeah It really is.

00:01:02: So with this deep dive we're looking at most relevant defense tech trends from recent weeks based on these curated insights And were seeing a massive pivot

00:01:12: Huge Pivot.

00:01:13: The landscape is shifting to software-defined, AI driven high volume systems that frankly iterate at the speed of Silicon Valley.

00:01:22: Which is incredibly fast!

00:01:23: Right.

00:01:24: so we are going to unpack how this is happening why it actually matters to you as a civilian tech professional and where the true bottlenecks real are.

00:01:33: And honestly The absolute catalyst forcing this entire paradigm shift Is the drone.

00:01:38: Yeah It's cheap It's literally everywhere, and it has completely inverted the cost curve of conflict.

00:01:45: Angus Bean actually highlighted this shift beautifully in his post.

00:01:48: What

00:01:48: did he say?

00:01:49: He noted that because of the sheer pace of drone development, response cycles and defense just can no longer be measured in years.

00:01:54: Oh

00:01:55: absolutely not!

00:01:55: They

00:01:55: have to measure weeks

00:01:56: Which makes total sense from like an agile development standpoint.

00:01:59: You can't spend five years locked in a waterfall development cycle for a defense system when the adversary is pushing software updates to their attack drones, what every three weeks?

00:02:10: Exactly!

00:02:11: The adaptation loop has just shrunk drastically Yeah

00:02:14: and Sir Steer Aitha brought up that NATO was actually backing this shift with real capital Forty billion dollars over the next five years specifically aimed at counter drone capabilities.

00:02:25: That level of funding is a serious validation huge.

00:02:28: it means counter UASO unmanned aerial systems Is no longer just this niche capability?

00:02:34: It has formally recognized as the front line of deterrence.

00:02:37: right, but you know A major hurdle here is procurement speed.

00:02:41: Oh, we see this in enterprise IT all the time.

00:02:43: Right?

00:02:43: You want to buy a new software problem today but procurement takes six months right.

00:02:48: But in defense that lag isn't just annoying it's a systemic vulnerability

00:02:52: A glaring one.

00:02:53: yeah Michael Chandler analyzed This really alarming data set from Australia.

00:02:58: over a single year There were a hundred and forty seven unauthorized drone incursions Over domestic defense sites.

00:03:04: Wow And that includes Highly sensitive locations like RAF base Williamtown where they literally house f- Thirty five fighters.

00:03:12: That's

00:03:12: wild and the trend is going up.

00:03:14: Oh

00:03:14: steeply.

00:03:15: it was twenty one incursions a few years ago then sixty, and now hundred forty seven.

00:03:21: The frustrating part, though is that the technology to stop this already exists.

00:03:25: Like Australian companies are building systems right now That protect airspace and active conflict zones in Europe.

00:03:31: But their own domestic procurement systems Are so bogged down by legacy processes.

00:03:36: Yeah

00:03:36: they just can't onboard it fast enough Exactly!

00:03:38: They can't protect their own bases.

00:03:39: It's a total adoption failure not an engineering one.

00:03:42: but

00:03:42: I mean when the engineering does break through To deployment...it is phenomenal.

00:03:46: Gregory Rumsey detailed This system from Lockheed Martin called the Mortius X-Rotor.

00:03:50: Okay

00:03:50: And this isn't a traditional kinetic interceptor that just, you know rams into a target or explodes near it.

00:03:56: It actually carries high power microwave payload.

00:03:59: Wait

00:03:59: let's dig in to the physics of that.

00:04:01: for second High Power Microwaves How does that translate in the sky?

00:04:04: Is generating like an electromagnetic pulse to fry the circuits?

00:04:08: Effectively yes!

00:04:10: It operates like a localized directional EMP.

00:04:13: That is crazy.

00:04:14: So instead of relying on a physical collision or fragmentation, it emits a burst of microwave energy that overloads and permanently disables the electronics on incoming drones.

00:04:24: And because its non-kinetic ,it can neutralize upwards to fifty drones in this single flight!

00:04:29: what they call a sortie, and then just return to base recharge and go back

00:04:33: up.

00:04:33: That changes the unit economics completely.

00:04:36: Oh totally.

00:04:36: And there is another approach to this swarm problem too.

00:04:39: Jonathan Wells pointed to CIRI Swarm Interceptor which also from Lockheed.

00:04:45: This system links with a sensor suite called Sanctum To build an autonomous drone wall.

00:04:50: The interceptors essentially communicate peer-to-peer to physically block incoming threats

00:04:56: Like a self-organizing mesh network of physical interceptors.

00:04:59: It is a fascinating concept

00:05:01: it, but I hear that and Let me push back a bit with the cybersecurity analogy.

00:05:07: okay go

00:05:07: for

00:05:07: if we view this as building physical firewalls For The Sky We know That Digital Firewalls Have A Fundamental Weakness right If an attacker hits you With Overwhelming Volume like a distributed denial of service or DDoS attack, the firewall just drops packets and fails.

00:05:26: So if an adversary sends one hundred thousand drones at this autonomous physical wall want it suffer?

00:05:34: Well, that is a really compelling analogy.

00:05:36: A physical wall could absolutely be overwhelmed if you view it in total isolation right?

00:05:41: I mean the physical drone walls are bit more resilient than a digital firewall because physical space has constraints.

00:05:45: that bandwidth doesn't

00:05:47: yeah

00:05:48: but your core premise is totally correct volume defeats us.

00:05:51: single point of failure.

00:05:53: Mark K looked into the Houthi Drone strikes on Saudi energy infrastructure and his analysis proved that relying on a single layer of interceptors no matter how advanced they is just a losing strategy.

00:06:04: So the physical firewall alone is insufficient?

00:06:07: Exactly, you need multi-layered architecture.

00:06:10: You need electronic warfare to jam signals radar and optical sensors for early detection And of course rapid command software To coordinate it.

00:06:19: all

00:06:20: Right

00:06:20: Christian Emmaus reinforced this with data from NATO's Baltic Trust.

00:06:25: twenty six trials.

00:06:26: The key finding there was that effectiveness doesn't come From single magic bullet interceptor.

00:06:32: It never does.

00:06:33: No, it comes down to sensor integration and maintaining a common operational picture.

00:06:39: disparate sensors literally must synthesize data instantly which

00:06:43: perfectly bridges us into the software side of this.

00:06:45: we're really moving from A hardware problem to a data synthesis problems completely.

00:06:50: you can't just build physical walls because The incoming drones are no longer these dumb remote controlled planes.

00:06:55: they Are moving toward full autonomy?

00:06:58: Yeah This is the race for last mile targeting.

00:07:00: yeah Natalya Mikulska brought up a stunning figure on this.

00:07:02: What was it?

00:07:03: There are currently around two hundred Ukrainian companies actively developing this specific capability.

00:07:09: Two hundred companies tackling one specific feature that is massive.

00:07:13: It really shows how critical it is, right?

00:07:15: Let's clarify why.

00:07:16: the last mile is The Holy Grail here for anyone listening.

00:07:19: Usually a drone requires a continuous radio data link to a human operator.

00:07:24: Right Yep?

00:07:24: So if you use electronic warfare or EW You can just blast radio frequency noise at the drone sever That connection and the drone falls out of the sky.

00:07:34: That

00:07:34: Is the traditional jamming defense.

00:07:36: but If the drone has an on-board AI vision model, it only needs general GPS coordinates to get close.

00:07:44: Right

00:07:44: Once its in final approach at last mile It visually locks onto target using its own internal camera and processor doesn't need to talk to the operator anymore.

00:07:54: You can jam the radio frequencies all day long, but you can't jam an offline processor that is visually guiding

00:07:59: itself.".

00:08:00: Autonomy basically renders traditional electronic warfare obsolete in those final moments... And this acceleration of AI isn't just happening physical space either!

00:08:09: Tom S highlighted a parallel shift in the cyber domain with Funtyr AI.

00:08:13: Oh right, The Cyber Stuff!

00:08:14: Yeah we're seeing penetration testing timelines which traditionally took human red teams two full years to complete being compressed down

00:08:24: Three weeks.

00:08:25: Just to define Frontier AI for a second, we are talking about the absolute bleeding edge of machine learning

00:08:32: models.

00:08:32: Smartest stuff out there!

00:08:34: Right.

00:08:34: Models capable complex multi-step reasoning in a cyber context.

00:08:39: The AI isn't just running like an automated vulnerability scan.

00:08:43: It's way past that.

00:08:44: it is actively chaining exploits together writing custom scripts and pivoting through networks at machine speed.

00:08:53: Which

00:08:53: provides a massive asymmetric advantage for cyber defenders, I mean they can identify and patch vulnerabilities before human adversaries even find them?

00:09:01: Yep!

00:09:02: By the way if you want to keep up with how AI is compressing timelines like this across enterprise software in cloud architectures make sure you subscribe to catch our future deep dives.

00:09:11: because

00:09:15: But I think we do need to temper this enthusiasm slightly.

00:09:19: There's a tendency to view military AI as this flawless, invincible digital brain.

00:09:25: Yeah the AI hype cycle and defense is just as pervasive As it is in the commercial sector?

00:09:31: Look at these autonomous targeting systems And i wonder aren't they incredibly fragile when subjected To the messy unpredictable physical world?

00:09:40: Oh totally like.

00:09:41: can you Just defeat an advanced computer vision model with Basic optical illusions you

00:09:47: can.

00:09:47: and it's happening on the battlefield as we speak.

00:09:49: Robert F shared a brilliant example of this exact vulnerability.

00:09:51: Oh,

00:09:52: the mirrors.

00:09:53: Yes Russian forces have been observed bolting cheap building supply mirror panels onto their vehicles along with painting zebra-like stripes over The armor.

00:10:01: okay mirrors in Zebra paint explain the mechanics Of why that actually works against a multi million dollar AI?

00:10:07: It exploits the very nature of how these models are trained.

00:10:10: so a computer vision model is trained on thousands of images what a tank looks like.

00:10:15: A tank has specific shape, edge geometry and thermal signature because the massive engine runs hot.

00:10:23: But reflective mirror surface reflects ambient environment around it meaning much cooler to sensor fundamentally altering that thermal signature.

00:10:32: And zebra stripes I assume break up the edge geometry?

00:10:35: Precisely!

00:10:37: It is called Dazzle Camouflage.

00:10:39: The AI look at and doesn't see the continuous edge of a tank, it sees fragmented visual noise.

00:10:46: That's hilarious actually!

00:10:47: So the confidence score for AI just drops from ninety-nine percent to forty percent...and it refuses to lock on You have highly sophisticated software stack being entirely defeated by some hardware store supplies.

00:11:00: That is the ultimate low-tech countermeasure.

00:11:03: It honestly reminds me of the vulnerability Justin Nerdrum pointed out regarding physical capture...

00:11:07: Oh right, The Underwater Drone!

00:11:09: Yeah he discussed an incident where Iran recovered a largely intact dive LD which was autonomous underwater vehicle used by US Navy.

00:11:16: This introduces completely different threat vector for digital transformation professionals to really consider because it's hardware itself The whole, the motors might cost around two point five million dollars.

00:11:28: But Nerdrum astutely noted that financial loss of hardware is irrelevant compared to exploitation risk inside it.

00:11:35: Right!

00:11:41: They aren't just looking at the wiring, they are ripping out hard drives.

00:11:44: Exactly!

00:11:45: They reverse engineering decision-making logic and navigation algorithms to fail safes.

00:11:51: It is equivalent of a competitor getting their hands on your company's proprietary source code with massive national security implications.

00:11:59: Yeah when you deploy an uncrewed system You're essentially deploying your intellectual property into hostile territory.

00:12:06: If assumed it will be captured.

00:12:08: And there is another layer of fragility here related to connectivity.

00:12:12: Eva Sula touched on how electronic warfare exposes our deep reliance on GPS and continuous data networks.

00:12:19: Because if the system was designed constantly phone home or sync with a cloud environment, severing that connection just leaves it paralyzed

00:12:26: Or worse It receives spoofed data and starts operating on false premises.

00:12:31: Sula made a critical point.

00:12:32: for any system architect, true resilience is not simply having a backup system mapped out on some architecture diagram

00:12:39: right?

00:12:39: A diagram isn't reality!

00:12:41: A back-up protocol that hasn't been stress tested in the live environment Isn't to strategy.

00:12:46: it's a liability.

00:12:47: Resilience requires practiced fallbacks.

00:12:50: In defense, that means delegating command authority down the chain so human operators can actually make localized decisions when the main network inevitably goes dark.

00:13:01: perfectly to corporate incident response.

00:13:04: If the central servers go down, do local branches have the authority and the analog tools?

00:13:09: To keep operating exactly so we've covered the software The AI the eponymy.

00:13:13: but here is the inescapable physical reality.

00:13:15: whether you are writing code for an autonomous swarm or building those microwave interceptors You still have to manufacture of a physical chassis.

00:13:22: it's doing hardware right

00:13:24: And you have to do at an unprecedented scale.

00:13:26: industrial capacity Is the ultimate bottleneck.

00:13:29: Frank Rose articulated this perfectly when he noted that modern missile wars are fundamentally wars of attrition.

00:13:36: It is basically a math problem, the side that runs out of munitions first loses.

00:13:41: Yeah consumption rates in modern peer conflicts or just astronomical.

00:13:45: This requires not to sustain production but radical flattening cost curve.

00:13:50: We can't afford shoot down ten thousand dollar drones with two million dollars interceptor missiles.

00:13:55: No!

00:13:55: The math doesn't work.

00:13:57: And Iman Khan highlighted a sobering metric regarding Canada's manufacturing base.

00:14:02: Their artillery shell production capacity has been hovering around just five thousand rounds per month.

00:14:07: Five Thousand Rounds A Month!

00:14:09: To frame that for you listening, in the heaviest fighting in Ukraine, five-thousand rounds represents A Western nation's entire monthly output is being burned through in twenty-four hours.

00:14:20: Which

00:14:20: is terrifying when you contrast it with the manufacturing capabilities ramping up elsewhere.

00:14:25: Florian Seibel shared a data point that should be huge wakeup call for the defense sector

00:14:29: The cost of that drone?

00:14:31: Yes, he pointed to German designed Shahed style drone airframe so long range loitering munition That has now been manufactured.

00:14:39: Two hundred and forty-three euros per unit.

00:14:42: Let that sink in, two hundred and fourty three euros?

00:14:44: It's wild!

00:14:45: That is cheaper than a mid tier smartphone.

00:14:48: how is it even possible?

00:14:49: are they just using commercial off the shelf parts?

00:14:51: mostly?

00:14:51: yes They're utilizing advanced polymer manufacturing commercial electronics And highly optimized assembly lines.

00:14:58: The stated manufacturing capacity for that specific airframe Is over one hundred seven thousand units A year

00:15:05: Over a hundred thousand units.

00:15:07: This shift to hyper-stale low cost mass production is happening globally too.

00:15:11: Devon Pardwaj noted that an Indian manufacturer called Inside FPV, Is currently constructing one of the world's largest FPV first person view drone production facilities.

00:15:20: Yeah

00:15:21: India has a very interesting case study here.

00:15:23: But Ritesh Prasad added some necessary nuance To indias push

00:15:26: Right?

00:15:27: He pointed out The critical distinction between mere assembly and true industrialization.

00:15:31: Buying motors from one country Flight controllers From another And screwing them together in warehouse That's just assembly.

00:15:38: It is fragile!

00:15:39: If a supply chain gets cut, your assembly line stops immediately.

00:15:42: Exactly True industrialization means localizing the critical components.

00:15:47: It means building the tooling, maintaining quality assurance at scale and fostering a deep domestic ecosystem of engineers and suppliers.

00:15:55: And that deep integration takes a lot

00:15:57: time.

00:15:58: Decades sometimes

00:15:59: Which brings up massive structural question.

00:16:02: We keep talking about agile startups hitting one hundred thousand unit capacities But historically Western defenses dominated by massive prime contractors.

00:16:11: Their development cycles take ten to fifteen years.

00:16:14: How do the US or Europe scale up quickly enough to meet this new demand without waiting a decade?

00:16:19: To permit and build entirely new defense factories.

00:16:22: Well, This might be my favorite insight from The Batch.

00:16:24: Mark A Whitefeld proposed a historical framework to solve this the World War two Sherman tank model.

00:16:30: Okay I love it.

00:16:31: good historical analogy.

00:16:32: What did they do differently back then

00:16:34: when the allies needed extreme military mass in the nineteen forties They didn't try to scale-up that bespoke Defense industry alone.

00:16:41: They tapped into the massive, pre-existing industrial base.

00:16:44: Right!

00:16:44: The car companies.

00:16:45: Exactly they went to Ford Chrysler General Motors... ...they took factories that built passenger cars and retooled them to churn out tanks aircraft engines and munitions.

00:16:55: You leveraged the commercial manufacturing scale That already exists.

00:16:58: Whitefield argues that Europe's current challenge is identical Attempting to build a parallel defense.

00:17:05: only manufacturing base from scratch Is just too slow And Too expensive.

00:17:09: So what's the move?

00:17:11: The only way to achieve the necessary scale is to convert existing automotive and commercial industrial capacity into defense production.

00:17:18: You have to design the software-defined weapons specifically so they can be mass produced by commercial hardware partners using existing assembly

00:17:26: lines.".

00:17:27: That completely reframes the problem, the hardware basically becomes a commodity mass produce by car manufacturers while the real value in differentiation live in the software stack provided.

00:17:38: But that requires a massive organizational shift.

00:17:41: Dr.

00:17:42: Ulrich Horsman and Alex Mathar discussed Denmark's new Armed Forces AI Center, And they both zeroed in on this reality.

00:17:49: The success of digital defense transformation is rarely A technological problem.

00:17:54: It is fundamentally an organizational and cultural

00:17:57: problem If the procurement office operates On a waterfall methodology from nineteen ninety five.

00:18:02: it doesn't matter if your AI models are bleeding edge.

00:18:05: If you treat an autonomous drone swarm deployment like a classic ten-year IT infrastructure upgrade, the project will be obsolete before the ink on contract dries.

00:18:15: The culture clash between Silicon Valley Speed and Pentagon style bureaucracy is real front line here.

00:18:21: It's profound shift.

00:18:23: We've covered massive amount of ground today from microwave emitters frying drones in midair to AI computer vision being defeated by hardware store mirrors all the way to resurrecting the Sherman tank model for modern automotive factories.

00:18:37: Yeah, it's a lot to take in and I'd like to leave you with a final thought to mull over building directly on that Sherman Tank analogy.

00:18:43: go

00:18:43: forth.

00:18:44: if The future of defense relies on co-opting existing commercial and automotive capacity What does that mean?

00:18:50: For you as a civilian tech digital transformation or manufacturing professional at the firewall between commercial technology and defense technology is practically gone.

00:18:59: It's completely blurring, a sensor designed for self-driving cars is now navigating an autonomous drone

00:19:06: which forces very uncomfortable question will every major tech firm software company and automotive manufacturer eventually be forced to pick aside in global conflicts?

00:19:17: Wow!

00:19:17: As your APIs Your machine learning models and your factory floors become dual use by default.

00:19:23: The concept of remaining a quote-unquote neutral commercial entity in the tech sector might soon become impossible.

00:19:29: A heavy but vital thought for anyone writing code or managing supply chains.

00:19:33: today If you enjoyed this episode, new episodes drop every two weeks.

00:19:37: Also check out our other editions on cloud insights and sovereignty ICT and Tech Insights digital products & services AI and agentic systems Green ICT sustainable AI HealthTech.