Best of LinkedIn: Defense Tech CW 12/ 13

00:00:00: provided by Thomas Allgeier and Frennis, based on the most relevant posts on LinkedIn about DefenseTech in CW-XII and XIII.

00:00:07: Frenis is a B to B market research company that supports ICT providers with a structured defense market entry framework designed to move them from European defense opportunity landscape to qualified ministry conversations within six to eight weeks.

00:00:20: you can find more info in description.

00:00:23: so today we are diving into the absolute

00:00:26: top defense tech trend seen across linkedin

00:00:28: Exactly.

00:00:29: Top defense tech trends seen across LinkedIn, and it is a fascinating batch of insights this week.

00:00:36: It really is I mean welcome to the deep dive everyone.

00:00:38: We are essentially taking this real-time intelligence feed from industry professionals And just distilling it down.

00:00:44: no fluff Just the core shifts happening right now

00:00:47: Right in reading through these posts.

00:00:48: so that overarching theme Is just this massive structural shift.

00:00:51: we're talking A complete ground up rearchitecting Of how security even operates.

00:00:56: Yeah its wild like for the longest time.

00:00:58: When you and I thought about defense technology, we pictured this classic Swiss watch.

00:01:03: Exquisitely

00:01:04: engineered?

00:01:05: It's exactly taking decades to design incredibly expensive protected at all costs like a stealth bomber.

00:01:11: but looking at the feed recently that pristine Swiss Watch is basically being overwhelmed by a swarm of angry bees.

00:01:20: Honestly, that's the perfect way to visualize it.

00:01:22: It's not an incremental upgrade anymore.

00:01:24: The center of gravity is moving totally away from heavy hardware and shifting into well software data And just overwhelming volume.

00:01:33: let's

00:01:33: start right there actually with a volume aspect specifically drones Because the conversation has fundamentally changed.

00:01:39: Oh entirely

00:01:40: if you follow this base drones aren't experimental novelties any more.

00:01:43: But what strikes me from the sources?

00:01:45: Is that they aren't even just tactical tools for like a single squad to peek over a hill.

00:01:51: Right, they are scaled primary battlefield systems now?

00:01:54: Yes

00:01:54: Tim Desider shared this really compelling analysis about this.

00:01:57: He was looking at reports showing how first-person view or FPV drones have essentially become a horizontal threat layer.

00:02:04: A horizontal thread layer?

00:02:05: That sounds intense!

00:02:06: It

00:02:07: is, we aren't talking about single drone buzzing overhead.

00:02:10: We're talking about persistent swarms operating at ranges up to twenty kilometers.

00:02:15: it creates the constant low altitude blanket of threats.

00:02:19: Wow

00:02:20: But the mechanism is the crazy part.

00:02:22: The Zitter pointed out that groups like Hezbollah are now deploying fiber optic FPV drones?

00:02:28: Wait, I really want to unpack that because trailing twenty kilometers of physical cable behind a drone sounds...I mean it sounds like massive logistical headache

00:02:37: Like World War One dragging a telegraph wire.

00:02:40: Exactly Why on earth would you do this instead just using standard wireless signal

00:02:44: Because of electronic warfare or EW?

00:02:47: See, traditional drones rely on radio frequencies to talk the operator.

00:02:52: So if you're defending against it... You don't even need shoot down!

00:02:54: Just turn an EW jammer, blast white noise across a spectrum and the drone basically goes deaf and drops out of sky.

00:03:02: But fiber optic drone operates in closed loop light inside physical glass tube.

00:03:07: You literally cannot jam light into cable with radio waves.

00:03:11: It's physically impossible.

00:03:12: That is insane.

00:03:14: A school-of thread completely bypasses these state-of-the art air defense networks.

00:03:19: Completely useless against it, It forces militaries to rethink everything.

00:03:24: All those multi million dollar domes are irrelevant Against a twenty thousand dollar piece of plastic

00:03:30: And the targets they're going after Are shifting just as fast.

00:03:33: Dr.

00:03:33: Jonas Singer brought up a really stark example of this.

00:03:36: Oh, the gas prom post?

00:03:37: Yeah

00:03:37: He noted that Ukrainian drones recently targeting Gazprom compressor stations Which for context are the pressure nodes feeding Russia's last gas pipeline to Europe.

00:03:46: Right you don't need to blow up a thousand miles of steel pipe

00:03:49: You just hit the fragile machinery pumping the gas.

00:03:52: And Summers point was that European nations are sitting around debating heavy tank procurement.

00:03:58: But suddenly continental energy security is fundamentally a drone defense

00:04:02: problem which turns a localized tactical weapon into a massive strategic threat.

00:04:06: Exactly, A handful of cheap quadcopters threatens an entire nation's energy supply and if you can't jam them like with the fiber optics we just talked about your only option is to physically shoot them down And

00:04:18: mathematically that is a nightmare.

00:04:20: This brings up what Mark K calls The Cost Exchange

00:04:23: Trap The brutal math Of modern warfare.

00:04:26: Exactly Let's break down the reality on the ground.

00:04:29: Say an adversary launches a one-way attack drone, cost maybe twenty thousand dollars?

00:04:35: To defend that gas compressor station militaries are forced to fire

00:04:41: which is a very long name for an incredibly expensive rocket.

00:04:45: It really is!

00:04:46: it's a radar-guided kinetic kill vehicle, exquisite aerospace tolerances and costs roughly four million dollars per shot.

00:04:54: So you're forced to spend four million to swat a twenty thousand dollar fly

00:04:58: every single time.

00:05:00: That just unsustainable.

00:05:01: Mark K emphasizes that this is what magazine depth warfare looks like.

00:05:05: The enemy isn't even trying to blow up the radar anymore, their target is your budget!

00:05:09: You want to bankrupt you?

00:05:10: Literally!

00:05:11: Yeah.

00:05:11: Bankrupted defense systems by forcing you to fire your most expensive interceptors at they're cheapest garbage

00:05:16: Man... I love how Mark CL framed this dynamic.

00:05:20: He argued That if you are an investor or a defense planner ...you have stop obsessing over the shiny perfectly polished drone system

00:05:28: Because they miss all point of mass

00:05:30: Right.

00:05:31: He used this brilliant metaphor, he said we need the Kalashnikov of drones.

00:05:35: That AK-XVII comparison is incredibly

00:05:38: apt!

00:05:39: It really is because the Ak-XIV wasn't legendary because it was precise...it was legendary because engineering tolerances were deliberately loose.

00:05:47: You could drop in mud and still fires.

00:05:49: Exactly Marciel is saying that drone systems are...and use this word shabby Deliberately unrefined so you can mass produce them by thousands in chaotic environments

00:06:00: without needing access to highly classified fragile supply chains.

00:06:05: And we're seeing this disrupt the traditional prime contractors who built those billion dollar legacy systems, Andrew B Eckhart touched on that.

00:06:11: Oh

00:06:11: right!

00:06:12: The rise of the neo-primes.

00:06:13: Yeah he noted that recent conflicts shattered the illusion you can rely upon small inventories of ultra expensive assets companies like andural and shield AI.

00:06:24: their entire value proposition is scalable mass production Because

00:06:28: the opening night of a conflict might be stealth bombers, but a multi-year campaign is won by whoever prints most drones.

00:06:35: Exactly!

00:06:36: And the scale was happening in real time.

00:06:38: Florian Seibel shared an update about quantum systems scaling up production Not theoretical concepts.

00:06:49: Combat proven systems specifically engineered to take down shahed type drones, going from innovation straight-to massive industrial output.

00:06:57: Okay but let's play this out for a second.

00:06:59: If we're deploying fifteen thousand or eventually millions of these shabby drones We hit a massive logistical wall don't we?

00:07:05: The communications wall

00:07:06: Right.

00:07:07: A swarm is only a swarm if it talks the hive mind.

00:07:10: How do they function when an enemy figures how to sever their comms?

00:07:13: And that is exactly why the battleground is shifting from the physical skies down into the compute layer.

00:07:20: Edge computing is emerging as a critical survival mechanism.

00:07:23: Glenn McCartan articulated this tension perfectly in his post,

00:07:27: The cloud speed versus edge resilience thing?

00:07:30: Yes

00:07:31: think about your smartphone On a five G connection, it's blazing fast because the heavy lifting happens in some massive Amazon data center.

00:07:39: but walk into a concrete basement.

00:07:41: It

00:07:41: gets very dumb...very quickly

00:07:43: Exactly!

00:07:43: McCartan notes that in a denied environment when an enemy turns on those jammers cloud-dependent drone swarms just fail.

00:07:51: they get cut off from their brain.

00:07:52: And Michael Hadji Sidoja added some crucial context there.

00:07:56: satellite communication terminals used to be safe haven Not anymore.

00:08:00: Now they are high-value targets,

00:08:01: so they're actively hunting the satellite dishes.

00:08:03: Yeah using multi domain tactics cyber infusions kinetic strikes electromagnetic weapons survivability isn't about ninety nine percent uptime in perfect conditions any more.

00:08:13: it's About functioning through deliberate attacks.

00:08:15: right but I have to push back a little here on The engineering reality.

00:08:19: if the cloud link is too vulnerable and we shrink the data center down To put it on the physical drone itself Aren't we forcing a cheap piece of plastic to make its own legal decisions in real time?

00:08:32: Yes.

00:08:33: How does it process that much data without a supercomputer?

00:08:36: It's massive hurdle, but the software is leapfrogging hardware.

00:08:40: Adnan Esmail shared this fascinating breakthrough from Stanford.

00:08:43: That proves how fast edge autonomy moving.

00:08:46: Oh robot arm thing.

00:08:47: This blew my mind.

00:08:48: It's wild!

00:08:49: Stanford researchers took foundation AI model called Pi Zero.

00:08:53: It was originally trained entirely on robotic arms doing physical manipulation, sorting blocks understanding weight grip

00:08:59: just desk tasks.

00:09:00: Right

00:09:01: and they fine-tuned it to fly a drone using only language commands.

00:09:04: wait so they taught an AI to move a mechanical arm?

00:09:07: And that exact same brain somehow knew how to pilot a quadcopter in three d space.

00:09:12: yes its concept called cross embodiment transfer.

00:09:16: think about humans.

00:09:17: learn.

00:09:18: if you learned to balance out of bicycle then try ice skate.

00:09:21: You don't have to relearn gravity from scratch.

00:09:24: Ah, the core physics transfer over?

00:09:26: Exactly!

00:09:27: The AI learned spatial uncertainty and physics form a robot arm... ...and transferred that to the drone without ever being trained on thousands of hours flight data.

00:09:35: That is just wild.

00:09:36: It intuitively understood wind resistance because it understood physical resistance.

00:09:40: Yep,

00:09:41: it navigated through gates And picked up objects mid-flight Just for text

00:09:45: prompt.

00:09:46: Okay but this where gets a little unnerving For me.

00:09:49: If edge tech Is advancing so fast that a robot arm brain becomes a drone pilot?

00:09:54: Are our governance frameworks keeping

00:09:56: pace?".

00:09:57: Not even close.

00:09:58: Right, giving a drone camera is one thing—giving it payload and trusting its onboard AI to figure things out in the dark...is entirely different.

00:10:06: You've

00:10:06: hit on massive readiness and governance gap fracturing industry right now.

00:10:11: The military demand for autonomy is accelerating way faster than AI models can guarantee safety.

00:10:17: John Chay-H observed this clash.

00:10:20: The Pentagon needs these systems because of the jamming threats, but the AI industry is looking at math and raising red flags.

00:10:27: Because large language models are probabilistic?

00:10:30: Exactly!

00:10:31: They don't process rigid logic—they predict next token based on statistical weights…and they hallucinate.

00:10:39: John's point is that a system that guesses answers is unfit for a lethal platform where you need absolute deterministic reliability...if X happens Why must always happen?

00:10:50: And this mathematical reality is causing real friction.

00:10:53: We're seeing different major AI labs navigate this in real time.

00:10:57: Anthropic, for example was reportedly blacklisted from certain federal defense contracts because they refused to remove internal restrictions on lethal systems.

00:11:05: Wow

00:11:06: Meanwhile OpenAI reportedly struck a deal with the DOD though They maintain they have similar restrictions on lethal use.

00:11:13: It just highlights the profound lines being drawn by civilian tech companies entering the defense space

00:11:18: and The technical danger goes way deeper than a simple hallucination.

00:11:22: Mike are posted about this concept frankly, it's chilling called meta war.

00:11:26: Oh that kill chain contamination breakdown how?

00:11:28: That actually happens.

00:11:29: so he points out that contamination enters at three levels model weights vector memory in behavioral conditioning.

00:11:36: Basically an AI doesn't store info like a file on hard drive that you can just drag to the trash.

00:11:42: Right, it's all distributed across billions of parameters!

00:11:44: Exactly.

00:11:45: Mike notes there are over four hundred research papers on machine unlearning trying to teach and LLM to forget something... The Consensus.

00:11:54: You could put filter in output make shut up but cannot mathematically teaching stop knowing Wow.

00:12:00: So if an AI is in autonomous lethal targeting chain, the trauma, bias or errors of that specific killchain become permanently baked into its vector space?

00:12:10: The system waged a meta-war on it's own code.

00:12:13: The implications are staggering especially with legal realities modern warfare.

00:12:18: Anthony Mayo published a sharp analysis about this.

00:12:20: his core point at machine speed human and loop.

00:12:23: total myth

00:12:24: because humans just too slow

00:12:25: Precisely.

00:12:26: If an AI generates a targeting recommendation in fifty milliseconds based on radar, thermal and radio chatter... ...a human operator cannot independently verify that before pulling the trigger

00:12:38: The cognitive load is just way too high By the

00:12:40: human isn't IN THE LOOP!

00:12:42: There are rubber stamps trusting machines' math

00:12:44: which creates massive black hole for legal accountability.

00:12:48: if war crime happens who goes to jail?

00:12:51: The operator or the software engineer who tweets the model waits three years ago.

00:12:55: Right,

00:12:56: Dr.

00:12:57: Anke Allen Huffer a lawyer specializing in defense tech posted a stark warning about this.

00:13:02: she argues we cannot allow algorithmic opacity to become new fog of war

00:13:06: shielding decision makers

00:13:07: exactly.

00:13:08: AI systems must have explainable reasoning that auditors can actually dissect.

00:13:12: We have somehow engineered international humanitarian law directly into the map

00:13:17: simple-to-say incredibly

00:13:18: difficult build.

00:13:19: So just to recap the shopping list, we need tens of thousands of cheap drones that think for themselves on the edge when jammed using AI.

00:13:27: That never hallucinates doesn't traumatize its own code and perfectly obeys international law at machine speed.

00:13:33: easy right no

00:13:34: problem.

00:13:35: but seriously solving that puzzle requires a completely revamped defense ecosystem which really brings us to our final theme Ecosystem dynamics and European sovereignty.

00:13:46: You

00:13:47: can clearly see the tectonic plates shifting here.

00:13:50: NOAA has highlighted that the Pentagon just awarded Enderall an enterprise AI contract worth up to twenty billion dollars.

00:13:56: A T-BILLION?

00:13:57: Yep, a ten year deal for AI driven systems... not a pilot program!

00:14:02: ...a definitive signal that U.S is pivoting hard toward software defined warfare trusting neoprimes over legacy aerospace giants.

00:14:09: Meanwhile Europe's feeling immense pressure.

00:14:11: Esben Gadspol posted a pretty blunt warning that Europe is lagging in capital innovation.

00:14:16: Oh, the VC funding

00:14:17: stack?

00:14:18: Yes!

00:14:18: He pointed out.

00:14:19: US-led venture capital is funding significantly more Ukrainian defense tech than EU led capital.

00:14:25: So while the innovation and testing happens in Europe's backyard... ...Europe risks becoming entirely dependent on American software.

00:14:31: But it's not entirely doom & gloom.

00:14:33: Harold Kappa offered a strong counterpoint.

00:14:35: He argues.

00:14:36: Europe is quietly building its own sovereign AI combat network right now.

00:14:40: Like Helsing Centaur, right?

00:14:42: Exactly!

00:14:43: Centaur and the autonomous CA-I Europa node.

00:14:46: The vision is uncrewed swarms flying alongside the Eurofighter Typhoon.

00:14:51: And critical detail is sovereignty of the tech

00:14:53: stack Meaning no foreign kill switches.

00:14:57: Exactly If a diplomatic rift happens No outside nation can revoke the software licenses and ground jets.

00:15:03: It's fully.

00:15:04: European

00:15:05: Institutions are trying to change how they fund this.

00:15:07: Thomas Vigier posted about the EU Commission's Agile program, a hundred and fifteen million Euro initiative designed by bypassed traditional top ten dominant players

00:15:17: Because legacy procurement takes decades By time hardware is delivered.

00:15:21: The threat has changed.

00:15:23: Right Software needs daily updates So the Agile Program gives funding directly to startups And their capping grant approvals at four months Forcing bureaucracy To move with speed of software.

00:15:33: It's a clear recognition that it is not about bending metal anymore.

00:15:51: He

00:15:55: highlighted something that completely changes the trajectory.

00:15:58: Ukraine just released millions of frames of real, annotated battlefield drone footage to allied nations for AI training.

00:16:06: Wow!

00:16:06: Raw validated combat data?

00:16:08: Not simulations – REAL

00:16:10: combat!!

00:16:11: And Swar's Watt points out that competition isn't about the slickest hardware anymore….

00:16:15: The provocative thought I want you to mull over is this... The next global superpower won't be the one with the biggest factories.

00:16:21: It will be the ones that monopolizes the most intense, validated real-world combat data.

00:16:26: Data is new strategic high ground!

00:16:29: Man... If modern defense is a swarm of autonomous bees….

00:16:32: ...the hive mind only as smart it feeds on.

00:16:35: and that feast's happening right now.

00:16:37: Well if you enjoyed this episode New episodes drop every two weeks.

00:16:41: Also check out our other editions on ICT & Tech Insights HealthTech Cloud digital products and services, artificial intelligence and sustainability in green ICT.

00:16:51: Thank

00:16:51: you so much for joining us on this deep dive!

00:16:53: Stay curious remember to subscribe and we'll see ya next time.