Best of LinkedIn: ICT & Tech Insights CW 25/ 26

00:00:00: This episode is provided by Thomas Allgeier and Frennis, based on the most relevant LinkedIn posts about ICT and tech insights from CW-TwentyFive and TwentySix.

00:00:08: Frennes supports ICT enterprises in the form of delivering precise ICT market and pricing intelligence that analysts' subscriptions and existing databases cannot provide – you can find more info Right now, somewhere in the world.

00:00:23: heavily funded nation states are quietly hoarding mountains of your encrypted data.

00:00:27: I mean intellectual property health records banking details just

00:00:31: stockpiling

00:00:32: exactly.

00:00:33: they can't read a single line But they're stockpiling it anyway, basically betting billions that there are just a few years away from building an eschene and we'll crack all wide open.

00:00:42: It's the ultimate long con.

00:00:43: like their stealing is safe today knowing they will invent the key tomorrow

00:00:47: right?

00:00:48: And not impending.

00:00:49: reality completely changes how you have to think about digital infrastructure.

00:00:53: So, welcome to The Deep Dive.

00:00:54: Our mission today is cut through all the noise and distill the absolute top ICT and tech insights sourced directly from industry leaders on LinkedIn across calendar weeks twenty-five and twenty six

00:01:05: And we're really looking at a landscape where fundamental rules are being rewritten in real time.

00:01:09: Yeah they really are.

00:01:10: We'll explore how cybersecurity has morphed into this massive board level resilience mandate.

00:01:16: Also how artificial intelligence while simultaneously attacking and defending our networks.

00:01:22: Plus the very quiet shift to AI-defined vehicles.

00:01:25: Oh yeah, that's a huge one!

00:01:27: And finally we'll get into that terrifying quantum computing race I just mentioned...

00:01:31: ...I think throughline connecting all these insights is a total redefinition of security.

00:01:36: We really have to start there.

00:01:38: Dr.

00:01:39: Drijesh Pee and Cheetan Prakash are both posted some brilliant analysis on this exact shift over the past two weeks.

00:01:46: Right, about how the baseline has moved?

00:01:48: Exactly!

00:01:49: The overarching theme is that cybersecurity is no longer an isolated IT function.

00:01:54: I mean for the last twenty years cyber was treated as a cost center.

00:01:57: You know.

00:01:57: Yeah you just threw money at the IT department to buy firewalls and antivirus software

00:02:02: And their job.

00:02:03: basically it's risk reduction But now its fundamentally tied business resilience operational continuity and well protecting revenue.

00:02:12: So it's a shift from value protection to value creation, which means the metrics that leaders actually bring to the boardroom have to change entirely.

00:02:22: If you're a chief information security officer reporting to a board You can't just hand them a spreadsheet showing how many thousands of phishing emails you block this month.

00:02:30: They don't care about the operational trivia.

00:02:32: No they dont Nor

00:02:33: should they.

00:02:34: Andrei Prozorov broke this down beautifully by sharing his top three board-level metrics and they captured the maturity shift perfectly.

00:02:42: What were the three again?

00:02:43: First, current risk level.

00:02:45: Second, estimated financial impact.

00:02:47: Third, aggregated control effectiveness.

00:02:50: Notice what's missing there.

00:02:51: Right, they're a zero technical tool level metric

00:02:54: Exactly.

00:02:55: the board needs to know the actual financial exposure in dollars and cents And they want proof that the millions They are spending on defensive controls or mathematically reducing that exposure.

00:03:06: Okay Let me push on that.

00:03:07: third one aggregated control effectiveness because that essentially means Are the tools we bought actually working together?

00:03:13: To stop a breach Which is honestly a perfect segue into how governments are stepping in because if you can't prove your controls work Regulators, they're gonna force your hand.

00:03:23: Oh absolutely regulation Is turning digital trust?

00:03:26: Into a strict operational requirement.

00:03:29: yeah Adrian Sauer pointed out something wild happening In the European Union right now.

00:03:33: he noted that The Digital Operational Resilience Act door.

00:03:37: well It isn't even enough anymore.

00:03:39: The EU Cyber Resilience Act, the CRA is now going to require a CE marking for banking apps

00:03:45: Which completely upends how software has delivered.

00:03:48: Okay

00:03:48: let's unpack this.

00:03:49: We are now regulating banking apps like we regulate physical toasters with a CE mark.

00:03:54: I mean, if you make a toaster You have to prove the wires won't short circuit and burn A house down but a toster doesn't get updated over The air every Tuesday.

00:04:02: No it definitely doesn't.

00:04:03: And It definitely Doesn't Get attacked by nation stakes.

00:04:06: does applying physical Manufacturing standards To digital products actually work or is it just compliance theater?

00:04:12: Well, it forces a philosophical shift in software engineering.

00:04:15: Regulators are well aware that software is dynamic but they're arguing That the baseline architectural safety needs to be standardized from day one like

00:04:23: baking it in

00:04:24: exactly.

00:04:25: You cannot chip a product with fundamental known vulnerabilities and just hope to patch It later.

00:04:30: The CE mark for software means you have baked security into the core design not bolted on as an afterthought.

00:04:38: And if you think the concept is aggressive, the timeline is even tighter.

00:04:41: Right

00:04:41: because Clayton Inge highlighted what's happening with Dutch cyber bevelaging sweat which their implementation of the NIS-II directive Yep.

00:04:50: that takes effect in August twenty twenty six.

00:04:52: it'll hit roughly eight thousand one hundred organizations.

00:04:55: and kicker there zero transition period.

00:04:58: Wow!

00:04:59: None

00:05:00: Zero.

00:05:01: The obligations apply on day one.

00:05:02: We're talking about directors facing personal fines up to twenty five thousand euros for non compliance.

00:05:07: A zero transition period is a massive wake-up call for any leadership team dragging their feet.

00:05:13: But you know, even if you are compliant compliance doesn't equal survival.

00:05:18: Jane Franklin posed what I consider the ultimate resilience question of the week

00:05:22: The one about vendor tool.

00:05:23: Yeah

00:05:23: She asked What happens at the defensive tools?

00:05:25: You rely on most.

00:05:27: simply get switched off by someone else.

00:05:29: not breached by hackers Not broken by bad code just switched off

00:05:33: by your third-party provider.

00:05:35: Exactly!

00:05:36: Due to a commercial dispute or maybe geopolitical sanction, if you're primary defensive capability goes dark tomorrow can your business still function?

00:05:46: That defines true cyber resilience.

00:05:48: today.

00:05:49: it is capacity to operate under constant threat and sudden degradation Like if you lose your primary cloud firewall, does your network fail open or doesn't it feel closed?

00:05:59: Can you reroute traffic?

00:06:00: and If that level of resilience is the new baseline The most disruptive force testing.

00:06:05: That baseline right now Is artificial intelligence because AI is acting as both the ultimate weapon And the ultimate shield.

00:06:11: we are seeing It completely reshape the security playbook.

00:06:14: Tamer Egel all shared an incredibly sharp observation about how this changes the daily reality for security teams.

00:06:22: He noted that AI agents are now conducting routine security reviews instantly.

00:06:26: Right, taking over the manual work

00:06:28: completely.

00:06:29: in The old model a human Security engineer acted as a routing mechanism.

00:06:33: They would manually review code spot an error write a mitigation recommendation and toss it back Over the fence to the developers but now Now an AI agent can do that natively inside the developer's coding environment in real time.

00:06:46: So, The Human Engineer's primary value to the company is no longer their speed it's their judgment.

00:06:51: It's there ability to catch what machine gets wrong or frankly knowing which of a thousand vulnerabilities that AI found actually matter To this specific business context.

00:07:00: you are not racing the machine You're managing it

00:07:03: Exactly and as these systems evolve from just generating text taking actions fundamental security perimeter shifts.

00:07:11: Brent Allers brought up the rise of Agenic AI.

00:07:14: Let's define that for the listener, because Agenics AI is a major leap.

00:07:18: we aren't just talking about a chatbot answering questions.

00:07:20: No not at all.

00:07:21: We are talking about autonomous software agents.

00:07:24: they're given a goal like audit our cloud storage and They can independently log in navigate systems make decisions and execute changes without human intervention.

00:07:35: Which

00:07:36: is precisely why Allers argues that identity is now the critical security perimeter for non-human workflows.

00:07:43: If you have autonomous AI agents zipping around your enterprise networks, traditional security methods like static IP whitelists completely fail

00:07:51: because an IP doesn't tell you the context.

00:07:53: Right, a static IP just tells you where the traffic came from.

00:07:55: it doesn't Tell You The Intent?

00:07:57: You now need cryptographic agent identities.

00:07:59: your systems have to continuously verify not Just who the human is but which specific AI agent Is requesting access what its parameters are and whether Its current actions match It's assigned Context

00:08:11: Because I mean if a malicious actor hijacks that AI Agent they inherit all its automated privileges.

00:08:16: And regulators Are watching this rapid integration very closely.

00:08:19: Gerald A to rather stark warning from the European Banking Authority.

00:08:23: Oh, their latest risk assessment?

00:08:24: Yeah!

00:08:25: The EPA stated that while banks are broadly resilient today... ...the rapid development of highly capable AI models is heavily amplifying their operational and cyber risks.

00:08:35: It accelerates the automation of attacks And creates massive opaque third-party dependencies on a few companies That control these frontier AI models

00:08:43: Which means traditional human driven threat assessments we use Are becoming entirely obsolete.

00:08:50: MJ Vadia highlighted IBM's new AI cyber resilience assessment, which is built specifically for these new threats.

00:08:57: So it's AI fighting AI?

00:08:58: Pretty much human red teams simply cannot keep up with adversaries who are using AI to map vulnerabilities at machine speed.

00:09:05: Here's where he gets really interesting though and honestly a bit concerning.

00:09:09: If AI is doing the security reviews and AI also generating attacks, aren't we just building a high-speed machine versus machine war where human teams are stuck on sidelines?

00:09:19: Tactically yes.

00:09:20: This

00:09:20: an arms race of algorithms But strategically The human element becomes more critical than ever.

00:09:26: Humans have to define rules for engagement.

00:09:28: Humans design governance architectures And strategic guardrails that keep these automated systems from spiraling out control.

00:09:35: We have to govern the physical footprint of it all too, because scaling this level artificial intelligence requires immense specialized physical infrastructure.

00:09:44: Eric Skerron discussed Helix Digital Infrastructure a company that just raised ten billion dollars from KKR and NVIDIA

00:09:51: To build purpose-build AI data centers.

00:09:53: Right!

00:09:54: And understand why this is big deal.

00:09:56: you need to understand why AI data center are fundamentally different than normal cloud datacenters.

00:10:01: It comes down power density in heat.

00:10:03: A traditional server rack might draw what, ten to fifteen kilowatts of power?

00:10:07: Something

00:10:07: like that.

00:10:07: But a rack fully loaded with AI GPUs—like NVIDIA H- one hundredths —can draw forty to one hundred kilowattes!

00:10:15: You cannot cool that with traditional air conditioning.

00:10:17: The freshwater requirements for evaporative cooling in these massive facilities are straining local municipalities to the breaking point

00:10:24: Which is why Helix doing something so fascinating.

00:10:27: To bypass local power grid limitations and those massive freshwater constraints, they are utilizing an offshore model.

00:10:34: Right in the ocean.

00:10:35: Yeah

00:10:36: They're building infrastructure that uses cold water for cooling similar to highly efficient models proven in places like Norway.

00:10:44: You draw in freezing water from deep in the ocean or a fjord, run it through heat exchange loop and you entirely eliminate need for fresh drinking

00:10:53: water.".

00:10:53: It solves the physical bottleneck of AI beautifully but building these massive centralized A.I brains naturally leads into next problem... Who actually controls

00:11:03: them?

00:11:04: The sovereignty issue.

00:11:05: Exactly,

00:11:06: if European companies are relying on offshore data centers or American cloud providers for their fundamental AI infrastructure that creates a massive geopolitical vulnerability

00:11:15: which is exactly what is driving the intense conversation around digital sovereignty in Europe right now.

00:11:21: Christine Neckfuss-Nicolik and Bern Wagner both emphasized on LinkedIn that digital sovereignty isn't about isolating Europe or building protectionist walls, it's a necessary baseline for competitiveness and security

00:11:33: because if you don't control the critical AI technologies rely on You simply cannot guarantee your own security And

00:11:40: we are seeing market reward companies provide this control.

00:11:44: Elk Andrew noted that T-Systems maintained its number one rank as an IT service provider in Germany, That proves sovereign cloud environments and industrial AI platforms are the absolute top priorities for CIO's right now.

00:11:57: They want power of AI but they demand data sovereignty

00:12:01: And we're watching this battle for sovereign advanced AI play out In real time on streets with autonomous vehicles.

00:12:07: This is a great pivot

00:12:09: If you wanna see where AI meets physical infrastructure.

00:12:11: Look at the auto industry.

00:12:13: Augustin Friedl provided an incredible analysis of how this is manifesting in what he calls next-gen vehicle

00:12:18: intelligence.".

00:12:19: The shift, he outlines...is staggering honestly!

00:12:22: The automotive industry is pivoting away from fragmented level three autonomous ambitions and moving straight into Level Four ecosystem execution.

00:12:31: Companies like Waymo, Uber & Pony.ai

00:12:34: Exactly pushing incredibly hard to fully autonomous robotaxes and ridehailing networks.

00:12:40: But to get there the underlying architecture of the car itself has to change.

00:12:45: We are transitioning from software-defined vehicles, SDVs To AI defined vehicles or ADV's

00:12:51: and The bottleneck to scaling an AI defined vehicle isn't actually getting access to the AI?

00:12:56: The roadblock is something called legacy ECU fragmentation.

00:12:59: explain that for us because That's a very hardware specific problem.

00:13:02: right it's the core engineering hurdle.

00:13:05: ecu stands for electronic control unit.

00:13:07: In older legacy vehicle designs, a car might have seventy or eighty different microchips.

00:13:12: One controls the windshield wipers and other controls of brakes.

00:13:14: another manages the transmission.

00:13:16: They're all highly fragmented running different code.

00:13:18: Wow!

00:13:19: Eighty different chips.

00:13:20: Yeah You cannot drop an advanced centralized AI brain into a car And ask it to drive if has negotiate with eighty different outdated microcontrollers simultaneously.

00:13:31: It's like trying to conduct a symphony where every musician is playing different song and time signature.

00:13:36: You need a centralized compute architecture!

00:13:39: And according to the insights shared, Chinese OEMs like BYD are vastly outpacing German legacy OEM's in building these centralized architectures.

00:13:48: Their vertical integration is just on another level?

00:13:50: It really is!

00:13:51: it allows them to execute over-the-air or OTA update cadences much faster.

00:13:57: they can upgrade the entire car's brain seamlessly exactly like updating a smartphone.

00:14:02: and that speed of iteration Is exactly why legacy automakers are changing their strategy.

00:14:06: So BMW and Mercedes stopped their L-III offerings to focus on level two plus cost efficiency.

00:14:12: It's almost like the industry realized they don't need to build a flawless robot butler right now.

00:14:16: When you think about it, Level III autonomy puts the manufacturer in terrible legal position.

00:14:22: The car drives itself under specific conditions meaning the manufacturer assumes liability if it crashes.

00:14:29: But this system still requires human driver suddenly take over If things get too complex.

00:14:34: It is the worst of both worlds for The Automaker.

00:14:36: Level Two Plus keeps legal liabilities squarely on the human driver while still providing, you know, ninety percent comfort of automated driving in the highway

00:14:46: From a business scaling perspective.

00:14:48: optimizing level two plus for consumers which what people actually want to pay today While simultaneously partnering with tech companies to build separate level four robotaxi fleets seems be winning economic strategy.

00:14:59: Yeah, a highly advanced scalable cruise control is what the market is actually willing to pay for.

00:15:04: But speaking of edge cases that are suddenly becoming practical realities while AI's defining data centers and vehicles today, quantum computing.

00:15:14: And

00:15:20: the most critical insight from the past two weeks is that quantum Is no longer just a theoretical physics experiment confined to university labs.

00:15:28: No, it's very real.

00:15:29: now

00:15:30: It has officially crossed the line into an all-out engineering and manufacturing race.

00:15:34: The numbers definitely back that up.

00:15:37: Jay Gambetta highlighted IBMs announcement of a massive ten billion dollar investment Into project Starling?

00:15:50: It is.

00:15:51: We aren't talking about tiny, noisy experimental machines anymore.

00:15:55: we're talking about scaling up manufacturing to produce systems capable of reliably executing a billion operations without an error.

00:16:03: A ten-billion dollar commitment Is a massive market signal.

00:16:06: it tells the industry that The foundational quantum physics are solved and now it is purely a massive engineering challenge.

00:16:12: And Nick Rosa pointed out that this acceleration is happening across the entire industry.

00:16:16: Yeah,

00:16:16: it was atom computing, ion Q and IBM right?

00:16:19: Exactly!

00:16:20: They all hit major quantum error correction milestones in the exact same week...

00:16:24: ...and what's fascinating is they're doing using completely different physical substrates like adicomputing uses neutral atoms iNQ use trapped ions IBM superconducting circuits.

00:16:36: To translate that for you, these are basically just different microscopic methods for trapping and controlling a quantum bit or quivit.

00:16:43: so it doesn't lose its fragile quantum state.

00:16:45: But the fact all three distinct methods hit major stability milestones simultaneously proves this is technological convergence not a fluke.

00:16:55: But while the media focuses entirely on quantum computers, Donald Butts noted that Quantum Sensing is actually the overlooked commercial gold mine right now.

00:17:03: It's already growing at thirty-two percent annually.

00:17:06: Yeah!

00:17:06: Quantum sensing is incredible.

00:17:07: Explain how it works because sounds like science fiction.

00:17:14: physical forces like gravity, magnetic fields or time with a precision that classical sensors just cannot physically match.

00:17:21: And importantly it doesn't require massive cryogenic quantum computer to work.

00:17:25: So its practical today?

00:17:27: Very much so.

00:17:28: We are already seeing these sensors deployed in next generation atomic clocks and for GPS denied navigation systems.

00:17:35: If a military sub or an aircraft loses satellite GPS, quantum sensors can calculate exact positioning purely by measuring the subtle shifts in Earth's gravitational and magnetic fields.

00:17:46: That

00:17:46: is incredible technology but it doesn't erase the massive shadow hanging over the quantum space which brings us right back to our opening hook!

00:17:55: Dr.

00:17:55: Aaron C. Kemp issued a very stark warning about Harvest now, decrypt later strategy.

00:18:01: The stockpiling

00:18:02: we talked about?

00:18:02: Right!

00:18:03: Nation states are pulling encrypted data off the internet today.

00:18:06: they store it on massive servers knowing that by twenty-twenty nine or twenty thirty a fault tolerant quantum computer will be able to crack them.

00:18:13: mathematical algorithms that currently protect all global communications

00:18:17: and governments are finally realizing the clock is running out.

00:18:20: Richard Entrip and Christopher Ocpala highlighted two recent White House executive orders mandating a massive federal migration to post-quantum cryptography by twenty thirty.

00:18:29: So, a strict deadline?

00:18:30: Exactly!

00:18:31: The Department of Defense is requiring all high value systems to support quantum resistant encryption that same year And Michael Osborne shared that France's cybersecurity agency AN SSI Is moving even faster.

00:18:44: They are flat out refusing to certify any security products That aren't Quantum Safe starting in twenty twenty seven.

00:18:50: Wait, let me make sure I have this straight.

00:18:51: The transition to post-quantum encryption involves updating cryptographic libraries across thousands of massive legacy enterprise networks.

00:18:59: IT experts estimate that migration takes ten or twelve years to complete properly.

00:19:03: It's a huge undertaking

00:19:04: And the US mandate is twenty thirty.

00:19:07: But current estimates say quantum computers could realistically break standard RSA encryptions somewhere between twenty twenty eight and twenty thirty three.

00:19:15: If data already being harvested today.

00:19:17: aren't we mathematically too late protect A massive chunk of our long-term secrets.

00:19:22: For any data that has already been harvested and is sitting on a foreign server, yes the mathematics are grim That data will almost certainly be decrypted in next decade.

00:19:30: The race right now Is to stop the bleeding.

00:19:34: Organizations have to secure the data currently In motion And protect the integrity Of all future communications.

00:19:40: It requires fundamental shift To what we call cryptographic agility

00:19:44: Meaning your network can't Be hard coded Just one type of encryption anymore.

00:19:48: Exactly Systems must be designed so you can swap out encryption algorithms seamlessly as new quantum threats emerge without having to rebuild the entire software stack.

00:19:59: The organizations that are manually mapping their cryptographic inventories today, finding exactly where old encryption lives in code... ...are only ones who will survive this transition intact!

00:20:09: It's a modernization challenge that touches every single digital transaction on the planet.

00:20:13: I mean from your banking apps to those AI-defined vehicles we talked about, to national defense grids...

00:20:19: it is absolute foundational change.

00:20:21: but you know if you look at the ultimate trajectory of quantum technology Steve Suarez shared a fascinating insight, citing in McKinsey report which noted that mature quantum computer could theoretically cut pharmaceutical drug discovery time from roughly seven years down to just one year.

00:20:41: Down to single-year?

00:20:42: Yes!

00:20:43: By modeling molecular interactions perfectly at the quantum level it can unlock personalized medicine and cure complex diseases at an unprecedented speed.

00:20:52: But, as we just discussed the exact same machine that delivers those medical miracles is a machine which will break all current digital trust protocols.

00:21:01: That's a massive

00:21:01: paradox!

00:21:02: It forces us into really tough corner.

00:21:04: Are We As A Society prepared to manage world with exactly technology capable of curing terminal diseases in matter months?

00:21:11: Is Simultaneously The Greatest Existential Threat To Global Data Security?

00:21:16: If you enjoyed this episode, new episodes drop every two weeks.

00:21:19: Also check out our other editions on cloud defense tech digital products and services artificial intelligence sustainability in green ICT Defense Tech and HealthTech.

00:21:28: Thank You for joining us On This Deep Dive.

00:21:30: Don't forget to subscribe And we'll see ya next time!