Best of LinkedIn: ICT & Tech Insights CW 37/ 38

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-ThirtySeven and ThirtyEight.

00:00:09: 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.

00:00:22: Imagine waking up tomorrow, you know and your city nine.

00:00:26: one-one dispatch is just completely offline.

00:00:28: Oh man That's a nightmare scenario

00:00:30: right?

00:00:30: And not just for an hour but alongside the police routing system fire dispatch and even municipal water billing.

00:00:36: It was everything down at once.

00:00:37: exactly now imagine they stay off line for forty six consecutive days

00:00:41: Which is terrifying because that isn't some theoretical doomsday tabletop exercise.

00:00:45: No,

00:00:45: it just happened.

00:00:46: so welcome to the deep dive.

00:00:48: today We are unpacking a curated stack of insights for you focusing on the harsh realities of digital transformation The massive shift from cyber defense To cyber resilience and the rapid industrial scale maturation Of quantum computing.

00:01:02: yeah in the overarching pattern across all these domains Is this?

00:01:07: sudden almost violent collision with reality.

00:01:09: A

00:01:10: serious reality check?

00:01:11: Exactly, whether we are looking at enterprise IT critical infrastructure or quantum mechanics you know the industry's tolerance for theoretical implementations what you might call science

00:01:22: projects

00:01:23: has basically vanished.

00:01:24: yeah it's gone.

00:01:26: The focus is entirely on measurable business outcomes physical manufacturing realities and honestly just keeping the lights on.

00:01:34: Well, we should probably start with that reality check in the enterprise IT space.

00:01:39: Because the ultimate goal of any tech upgrade is to make the business run better for you right?

00:01:44: But there is this persistent delusion that simply installing a piece of software equals transformation.

00:01:50: Oh,

00:01:50: it's very comfortable to lose.

00:01:51: yeah I mean you look at the deployment checklist You see The progress bar hit one hundred percent and the vendor just hands over the keys.

00:01:57: job done right?

00:01:58: Right but Eric kibbling brought This up in a great breakdown Of why digital transformations actually fail.

00:02:04: A Software Vendor can dilute me flawlessly on their contract

00:02:07: like the code does exactly what the technical spec demands

00:02:10: Exactly but the transformation still completely collapses.

00:02:13: Because employees find it impossible to use in their daily workflow.

00:02:17: Yeah, or the new process actually bottlenecks the supply chain instead of fixing it.

00:02:23: Kimberling's point is that completing a tech implementation is merely milestone

00:02:27: But The C-suite treats technology itself as deliverable

00:02:31: Precisely!

00:02:32: Rather than business capability its supposed unlock.

00:02:35: So take exact same disconnect and apply hardware specifically at edge.

00:02:40: We are deploying massive tech infrastructure without considering the physical or economic realities of the environments we're putting them in.

00:02:48: Yeah, Luke Willwarding highlighted a very concerning trend there.

00:02:51: Edge computing costs are spiraling out at control

00:02:54: because organizations are effectively trying to build many data centers at every single branch location or retail store.

00:03:00: right they're taking the cloud architecture model shrinking it down and just shoving it into a broom closet in thousand different locations.

00:03:08: Which is

00:03:09: wildly over-engineered...

00:03:10: It doesn't scale economically or physically for that matter?

00:03:13: Yeah.

00:03:14: Will wording pointed out, companies like Zebra & Elo are having to actively push their customers to right size the edge infrastructure based on actual workload.

00:03:23: Right!

00:03:24: You don't need massive Kubernetes cluster to run basic inventory scanner.

00:03:27: you

00:03:27: really dont.

00:03:28: And the hardware industry is actually undergoing a massive architectural shift to accommodate this rightsizing, um specifically when it comes to power consumption.

00:03:37: Oh yeah!

00:03:38: If you look at the breakdown Mali O'Shea did on Sambinova's recent one billion dollar series F valuation... The physical physics of the hardware are the real story.

00:03:48: That really stood out.

00:03:49: They're running their SN-Forty AI rack just ten kilowatts and its completely air cooled

00:03:55: which is incredible when you compare that to the alternative.

00:03:58: If you want to run a comparable NVIDIA GB-II rack, You are drawing up to one hundred thirty kilowatts

00:04:05: Which just for context for your listening A one hundred and thirty kilowatt draw Is roughly the power required To run an entire suburban neighborhood.

00:04:12: Wow And it requires liquid cooling!

00:04:14: You can't put on standard IT room with raised floor in fan.

00:04:17: No...you essentially have to pour new concrete Bring in heavy plumbing and build a purpose-built facility just to run the servers.

00:04:25: Right, And the Sambanova architecture achieves that efficiency by moving away from general-purpose brute force GPU processing... ...and utilizing data flow architecture optimized specifically for inference!

00:04:39: That philosophy of purpose built efficiency is moving all the way down into your pocket.

00:04:45: Steve McDowell shared notes from the Qualcomm Snapdragon Summit, regarding their new two-nanometer Snapdragon eight Elite Gen six processor.

00:04:53: What are they doing with that?

00:04:54: They're bringing persistent agentic AI directly to the on device mobile computing layer.

00:04:59: Let's

00:04:59: clarify a genetic AI for second just because it gets grown around a lot.

00:05:02: Yeah good idea.

00:05:03: We aren't talking about a generative tech spot That answers questions when you prompt it.

00:05:08: Agenetic AI refers to autonomous software agents that run continuously in the background have a persistent state and can execute multi-step actions on your behalf across different apps.

00:05:18: And historically running those autonomous agents required constant round trips to massive cloud data centers,

00:05:24: which creates latency and burns massive amounts of energy.

00:05:27: exactly

00:05:28: but Qualcomm is keeping the scalar vector in tensor processing entirely local on the chip.

00:05:34: they bumped up this shared memory by fifty percent to handle

00:05:37: it.

00:05:37: oh wow

00:05:38: yeah.

00:05:38: so the device bypasses the edge server.

00:05:42: So, it brings up serious question for IT teams planning their three-year roadmaps.

00:05:48: Are we currently overengineering the edge?

00:05:51: It really seems like it!

00:05:52: I mean building full liquid cool data centers at remote locations feels like packing a giant steamer trunk for a weekend trip when all you actually needed was a toothbrush.

00:06:02: That's the perfect way to describe it.

00:06:03: How do we stop treating every tech project with massive infrastructure build?

00:06:07: By tying back to Kimberling Orr point.

00:06:09: The technology is not deliverable.

00:06:12: If a ten kilowatt air-cooled rack can handle the local inference, you deploy that.

00:06:16: Right!

00:06:17: And if your smart phone chip can handle an agent work load... You don't even build the edge server….

00:06:22: You fit this tool to the physical reality of our business – not on the other way around.

00:06:26: Yeah exactly….

00:06:27: By the way …if you are finding this deep dive valuable make sure you hit subscribe so you don't miss out future additions.

00:06:32: Now THAT CONCEPT OF PHYSICAL BUSINESS REALITY Bridges us directly into the next theme.

00:06:37: Okay, because having beautifully optimized edge AI and highly efficient workflows means absolutely nothing if The entire system gets taken offline by a threat actor And just stays down.

00:06:49: which brings us to this shift from cyber defense to cyber resilience.

00:06:53: right for years?

00:06:55: The industry was obsessed with building up taller wall you know preventing the breach.

00:07:00: now The paradigm has shifted to assuming the wall will be breached and proving that business can survive

00:07:05: it.

00:07:05: And surviving is brutal!

00:07:07: Let's go back to that forty-six day outage in Susan City that Tommy Flynn documented.

00:07:11: Right, a relatively small California municipality not massive federal agency.

00:07:16: Yeah an malware attack crippled their nine eleven routing Their police and fire dispatch and water billing.

00:07:22: Forty six days to reopen city hall.

00:07:25: Think about the mechanics of that.

00:07:26: On day twenty, you still have citizens who need emergency services...

00:07:29: That's a payroll to process!

00:07:31: You still have a city run?

00:07:33: It highlights the massive often fatal gap between IT disaster recovery and actual continuity operations.

00:07:40: Exactly Disaster Recovery is an IT metric.

00:07:44: it's restoring server from backup.

00:07:46: Continuity of Operations is business metric

00:07:49: Like how police department physically routes officers when screens go black

00:07:53: Right And the data clearly shows that organizations are wildly unprepared for the latter.

00:07:58: Sanjay Poonan and Chris H highlighted a recent Cohesity report, which found seventy six percent of organizations missed their recovery time objectives.

00:08:07: And the most alarming part of that report is that ninety-three percent of cyber recovery plans are based on completely untested assumptions.

00:08:16: They just assume everything will work out perfectly?

00:08:18: Yeah, companies are assuming an incident would be neatly contained.

00:08:21: The backups were pristine and recovery was a smooth linear process.

00:08:25: They assumed they'll have the luxury time which absolutely won't.

00:08:29: This why?

00:08:30: only twenty two percent organizations tested minimum viable company model

00:08:35: But calling it a minimum viable company sounds great in a boardroom.

00:08:40: In practice, how does an organization actually determine that?

00:08:43: It's extremely difficult

00:08:44: because if you're a hospital or logistics firm every single application feels mission critical.

00:08:50: department heads will fight tooth and nail If tell them their system isn't essential.

00:08:55: How do you strip of complex business down to a minimum viable footprint?

00:09:00: It requires a brutal level of cross-departmental triage long before the attack happens.

00:09:05: You have to separate what generates revenue or keeps patient alive from what merely makes operations convenient.

00:09:12: A minimum viable company is absolute smallest footprint applications, data and identity structures required just keep serving customers while rest network is essentially burned to ground.

00:09:23: Attackers are taking advantage that most companies haven't done this.

00:09:28: Mark Trump discussed what he calls the dangerous ITOT seam.

00:09:32: Yeah, this is a big vulnerability.

00:09:33: information technology Is your email?

00:09:36: Your ERP systems?

00:09:37: you're billing and operational technology or OT represents The physical world valves on a pipeline robotics on an assembly line HVAC Systems

00:09:48: And historically security teams treated those as totally separate silos.

00:09:52: But threat actors don't respect internal org charts

00:09:55: they really don't.

00:09:56: Trump noted that threat groups are increasingly taking down the physical plant without ever directly touching the OT control systems.

00:10:03: Oh, wow!

00:10:03: They just compromise the IT virtualization environments that the plant relies on.

00:10:08: they literally severed the brain from the body.

00:10:10: But the data he shared offers a clear solution.

00:10:13: When organizations implement unified visibility across both their IT and OT environments effectively closing that seam The average time to contain a ransomware Incident drops from forty two days down to just five days.

00:10:26: That's a massive difference and the regulatory pressure is actually forcing this shift, too.

00:10:30: Michael Rasmussen pointed out that the EU cyber resilience actress reporting requirements are officially active now.

00:10:35: Well

00:10:36: they aren't.

00:10:36: yeah if you manufacture a product with digital elements You are legally mandated to report an exploited vulnerability within twenty-four hours.

00:10:44: And you simply cannot afford a twenty four hour reporting window if your security teams are spending forty two days trying understand what systems were even touched,

00:10:53: especially given how reckless threat actors have become?

00:10:56: Yeah

00:10:56: it's getting crazy out there.

00:10:57: Cynthia Kaiser noted a bizarre instrument recently where the shiny hunter's ransomware gang tried to extort a rival syndicate, KL-Zero P and retaliated against the FBI in that same week.

00:11:08: Just

00:11:08: absolute chaos.

00:11:09: Chaos!

00:11:10: You can predict the behavior of reckless actors which is why resilience –the ability take a punch & keep operating–is really only viable strategy.

00:11:19: And

00:11:19: while organizations are struggling to defend today's chaotic networks, the next computational era is quietly moving from theoretical physics into industrial manufacturing.

00:11:29: Yes let's talk about quantum computing.

00:11:31: there has been a very justified backlash against mainstream enterprise use for quantum has supposedly been five years away.

00:11:44: All

00:11:44: these five-years away?

00:11:45: Right, companies keep missing their commercialization deadlines and executives are really suffering from quantum fatigue.

00:11:52: Well John Prisco brought up a perfect example of this reality check.

00:11:57: NSKete Corporation just halted their universal quantum hardware development entirely.

00:12:03: Completely

00:12:03: halted it?

00:12:04: Yeah,

00:12:04: they're pivoting to focus solely on annealing because building the hardware for Universal Quantum was taking far too long to produce a return-on investment.

00:12:12: For

00:12:12: listeners who might not be deep into quantum architecture It is probably worth clarifying the difference there.

00:12:18: Universal quantum computing is the holy grail, a machine that can run any complex algorithm.

00:12:24: Exactly But annealing as much narrower it Is essentially a specialized optimization engine.

00:12:29: imagine shaking A complex box of sand until The heaviest rocks naturally settle at the bottom.

00:12:35: It finds the lowest energy state for a highly specific problem like logistics routing but it Can't run general code.

00:12:42: NEC decided to chase the immediate practical use case rather than the theoretical Holy Grail.

00:12:48: But while that skepticism born from timeline theater is valid, it might be blinding us to a massive industrial leap happening right now.

00:12:56: How so?

00:12:57: We are moving from laboratory experiments to merchant foundries.

00:13:01: Keith King and Jay Gambetta reported that IBM finalized A one billion dollar CHIPS Act award To build America's first three hundred millimeter pure play quantum chip Foundry known as Anderon.

00:13:14: Okay, you don't commit a billion dollars in manufacturing infrastructure for a science project.

00:13:18: Precisely

00:13:19: and this isn't just an American push either.

00:13:21: Jack Chuang Heo highlighted that Quantum Foundry Copenhagen is building a three hundred ninety million euro commercial quantum chip fabrication facility.

00:13:29: Wow This no longer race about theoretical physics.

00:13:31: It's fundamentally erased from manufacturing sovereignty.

00:13:34: Who can build the infrastructure to repeatedly and reliably print these chips at scale?

00:13:38: And those foundries are being built to meet concrete demand signals.

00:13:43: Steve Suarez pointed to the Department of Energy as two hundred and fifteen million dollar quantum Genesis Q competition.

00:13:49: Yeah, The money is significant but the technical specifications they're demanding are real story.

00:13:54: Right because They Are specifically Demanding vendors To demonstrate systems with at least a hundred logical quibits.

00:14:01: And the distinction between physical and logical Quibits Is essentially the entire bottleneck Of the Quantum industry.

00:14:07: Exactly Physical quibbits are the raw, incredibly fragile hardware units.

00:14:13: They're notoriously unstable and error-prone.

00:14:16: Whereas a logical quibbit is group of those physical quibits working together using complex error correction codes to act as one perfectly stable reliable unit computation.

00:14:26: So when the DOE demands a hundred logical quibits, they are essentially saying we're done funding norzy experimental prototypes.

00:14:33: We are placing an order for highly stable commercial grade capability

00:14:37: and enterprise customers or echoing that exact demand.

00:14:40: Heather C West IDC reports show that enterprises are finally mapping quantum to real world.

00:14:45: use cases were

00:14:46: actually putting it into work.

00:14:48: Yeah Alex G lead tracked eighty distinct real-world finance projects alone.

00:14:52: but while the finding sector is looking at quantum for optimization The cybersecurity sector is looking at a ticking clock.

00:14:58: The immediate existential threat of post-quantum cryptography, or PQC, is accelerating.

00:15:05: It

00:15:05: really is.

00:15:06: Dr.

00:15:06: Volk and Errol highlighted new research on what it's called the frozen pinnacle architecture

00:15:10: And this research fundamentally alters the threat timeline we were just talking about.

00:15:14: For years, the assumption was that breaking RSA- twenty forty eight encryption which secures basically our entire digital economy would require a massive quantum computer with millions of perfectly error corrected logical quibits.

00:15:28: A machine that might still be a decade away

00:15:30: right but The frozen pinnacle architecture research shows That rsa-twenty fourty eight could potentially factored in just one month using only a hundred and twenty thousand physical qubits.

00:15:39: Oh wow so it doesn't require a futuristic highly flexible error corrected machine.

00:15:44: It works on fixed, near-term noisy hardware.

00:15:47: Exactly!

00:15:48: it proves that the hardware constraints are not going to delay the code breaking potential as long we originally hoped.

00:15:54: The cryptographic apocalypse might not wait for perfect logical quibits which means...

00:16:02: But how do you practically execute a migration of this scale?

00:16:06: You can't just flip a switch and upgrade a global enterprise's entire cryptographic foundation overnight.

00:16:11: No, you can't!

00:16:12: Adi Lamar proposed framework that cuts through the paralysis called The Two Lens Rule.

00:16:17: It is very pragmatic triage system When deciding what to migrate to quantum safe encryption.

00:16:22: first... ...you evaluate data thru two lenses Shelf Life and Blast Radius.

00:16:27: Right, shelf life is simply asking how many years does this specific piece of data actually need to remain secret?

00:16:32: If it's a daily internal cafeteria menu the shelf-life is zero.

00:16:36: but if its blueprint for next generation weapon system or patients genetic sequence The Shelf-Life is

00:16:42: decades Exactly!

00:16:43: And second lens blast radius measures systemic dependency.

00:16:47: If a specific cryptographic key has broken How many downstream systems are compromised?

00:16:52: So you plot your entire infrastructure against those two metrics.

00:16:55: The assets with the longest required secrecy and the widest systemic dependency

00:17:06: are.

00:17:08: George Schwartz noted that Microsoft has officially mandated PQC compliance across its entire suite of systems by twenty-twenty nine.

00:17:15: Wow, That is an extremely aggressive timeline for enterprise IT

00:17:19: It Is.

00:17:19: But you know before we wrap up today I want to pull the lens way back for a second.

00:17:23: Okay We've talked about The massive power constraints Of modern AI data centers Drawing one hundred and thirty kilowatts And this shift toward quantum processing.

00:17:32: Well, David Steenhook shared a conceptual comparison that ties all of this together.

00:17:36: looking at the physics in the universe itself.

00:17:38: I'm intrigued.

00:17:39: let's hear it.

00:17:40: he compared quantum neural networks to the cosmic web specifically how black holes process information.

00:17:47: Theoretical physicists have modeled that black hole store massive amounts of data on their event horizons.

00:17:52: Okay but here is the key.

00:17:55: They store that information holographically, meaning it scales with the two-dimensional surface area of the horizon rather than three dimensional volume.

00:18:04: So they are storing exponentially massive amounts data but within efficiency almost completely eliminates energy waste?

00:18:12: Exactly!

00:18:13: Biological brains actually operate on a similar principle highly interconnected surface level efficiency.

00:18:19: But in enterprise IT We do the exact opposite.

00:18:22: We

00:18:23: build massive three-dimensional brute force server farms,

00:18:26: right?

00:18:27: we fight The natural physics of the universe by trying to shove Massive power through silicon.

00:18:32: quantum computing and holographic storage.

00:18:35: theories hint at a completely different paradigm.

00:18:37: It really raises the question is our current struggle with power hungry AI just a symptom Of us using the wrong architectural approach entirely.

00:18:45: It really makes you wonder,

00:18:46: will post quantum infrastructure look less like a rigid liquid cool data center and more like the organic hyper efficient event horizons found in nature?

00:18:54: it forces us to ask if we're finally learning Computing the way of

00:19:00: the universe does.

00:19:01: That is a profound thought to take with you, The next time you are looking at the power draw on a server wrap!

00:19:06: Absolutely If you enjoyed this episode new episodes drop every two weeks.

00:19:11: Also check out our other editions on cloud insights and sovereignty digital products in services AI and agentic systems green ICT and sustainable AI defense tech And health tech.