Best of LinkedIn: ICT & Tech Insights CW 35/ 36

00:00:00: This episode is provided by Thomas Allgaier and Frenas, based in the most relevant LinkedIn posts about ICT and tech insights from CW-ThirtyFive and ThirtySix.

00:00:09: Renez supports ICT enterprises in the form of delivering precise ICT market and pricing intelligence that analysts' subscriptions and existing databases cannot provide.

00:00:19: You can find more info at the description.

00:00:20: And we have a really fascinating deep dive for you today.

00:00:24: Yeah We do So.

00:00:26: i want to imagine building this state billion dollar digital bank fault.

00:00:31: Right, you reinforce the wall as you install these multi-factor biometric scanners and You hire this top tier security team to monitor the perimeter

00:00:40: right doing everything by the book

00:00:41: exactly And then you accidentally just tape The master blueprint and the vault combination directly to the front door for anyone walking By to read

00:00:48: yeah which sounds completely ridiculous.

00:00:50: it sounds absurd but As we uncovered today looking at the source material That is fundamentally the reality of how roughly thirty nine million streaming accounts were just exposed.

00:01:01: So welcome to The Deep Dive.

00:01:02: today we are looking at the actual, you know on-the-ground engineering realities happening right now across physical infrastructure cyber security and quantum computing.

00:01:11: Yeah, we're really trying to cut past the marketing hype here and just dig into the structural math and physics of where tech is actually headed.

00:01:18: Exactly!

00:01:19: Because what really emerges from the curated sources were looking at... ...is a massive shift in physical reality.

00:01:25: I mean We spend so much time in this industry obsessing over software right?

00:01:29: Like large language models algorithms This is often yeah always.

00:01:34: but right now The literal physical hardware of our world and the deeply urgent security frameworks required to keep it running are undergoing a transformation that is honestly almost unprecedented in its scale, at speed.

00:01:47: Okay

00:01:47: so let's unpack that scale because you really cannot have this software.

00:01:51: an AI revolution without that kind of puts this into perspective.

00:02:02: Yeah, the numbers are staggering.

00:02:03: Staggering is the right word.

00:02:05: they estimate approximately thirty one point six trillion dollars in cumulative global data center capital expenditure through twenty fifty.

00:02:14: Wow yeah almost thirty two trillion dollars.

00:02:17: and The US alone could capture about forty eight percent Of that central case investment.

00:02:22: now if you're listening to listen thinking well there's no way our current power grids and supply chains can handle that, you're entirely correct.

00:02:29: No your spot on the grid constraints are basically the defining bottleneck of this entire supercycle.

00:02:36: I mean the architectural reality of a data center is being pushed to its absolute breaking point right now.

00:02:40: How so exactly?

00:02:41: Well, Diana Disa highlighted that the data center rack market is growing at a ten percent compound annual growth rate right through the early twenty thirties.

00:02:49: But that growth isn't just about building more of the exact same racks.

00:02:54: it has driven entirely by the need for high density liquid cooling ready architectures.

00:03:00: because

00:03:00: these AI racks are now approaching what like one megawatt?

00:03:04: Of power consumption per single rack

00:03:07: exactly and you have to think has a terrible heat capacity.

00:03:13: Right!

00:03:13: When you're pushing full megawatt into single dense cluster of GPUs, blowing cold air down an aisle... It's

00:03:19: like trying to put out bonfire with the desk fan.

00:03:22: That is exactly what it's like.

00:03:23: The thermal density for these new chips requires fundamentally new plumbing.

00:03:28: You are literally piping dielectric fluid or direct chip liquid cooling systems right up to the hot silicon.

00:03:34: Which means that physical layout of a building has to change, just support weight and mechanics for all those fluids?

00:03:39: Yes which creates massive logistical nightmare.

00:03:43: if you are trying to build greenfield sites from scratch I mean getting permits securing land then waiting in this like five-to seven year interconnection queue Just get a gigawatt power from utility company.

00:03:55: it's stalling a lot of these hyper scale projects

00:03:57: Seven years.

00:04:01: But there's a workaround.

00:04:03: Dave Perez brought up a strategic shift where legacy data centers are being modernized simply by reusing their existing power capacity.

00:04:11: Oh, I found that fascinating in the sources.

00:04:13: instead of waiting half a decade to build a massive new campus companies just taking older much less efficient data centers maybe facilities that were only drawing...I don't know.

00:04:23: ten or twenty kilowatts a rack

00:04:25: Yeah!

00:04:25: The old enterprise stuff

00:04:26: Exactly and they're just gutting them.

00:04:28: They retrofit with high density liquid loops And they bypass the permitting and grid constraints completely because there just optimizing to power draw.

00:04:37: They already legally possess from the utility, it's

00:04:40: a pure speed-to-market play.

00:04:42: It's really smart.

00:04:43: But you know The infrastructure boom isn't Just about these massive centralized hubs drawing down megawatts of Power?

00:04:49: The compute is also pushing out To the extreme edge in ways that frankly redefine what A network actually Is.

00:04:56: Chelsea Larson Andrews brought up a vision for SixG that moves us far beyond just, you know faster downloads bees on your phone.

00:05:03: It's called integrated sensing and communication or ISAC.

00:05:07: Okay Let's pause and explain the mechanism behind this because it honestly blew my mind when I read it.

00:05:12: The concept here is that the network itself becomes a spatial sensor

00:05:16: Exactly.

00:05:17: Like instead of just transmitting packets of data from a router to a device, the six G radio waves are physically interacting with the environment?

00:05:24: Yeah think about it kind of like a bat using echolocation but with radio frequency.

00:05:29: Oh that's

00:05:30: a good way to picture.

00:05:31: The network is actively analyzing the microdoppler signatures, signal bounds and perturbations of waves as they move through a room.

00:05:40: So imagine you have smart warehouses or like a highly robotic hospital wing.

00:05:45: You wouldn't necessarily need separate heavy camera system for every single blind spot in building Or battery powered RFID tag on each pallet.

00:05:54: The communication network's radio waves themselves can track the exact physical location of an autonomous forklift or monitor movement inventory and feed that spatial data in real time right back to local AI control systems.

00:06:08: So you essentially turn a physical environment into planetary nervous system?

00:06:12: You're no longer just transmitting data over air, the infrastructure is like actively feeling its environment.

00:06:18: it's completely ambient intelligence

00:06:20: It IS.

00:06:21: but if follow logical implication When your warehouse Wi-Fi isn't just routing traffic but is actively tracking the physical movement of heavy autonomous forklifts, Your network is physically interacting with the real world.

00:06:35: Yeah That means if a hacker breaches your network they aren't just stealing your data anymore They're blinding your actual physical operations when you attack surfaces literally The environment around you.

00:06:45: your entire approach to security architecture has to fundamentally change

00:06:50: and that brings us directly into the stark reality of modern cyber defense.

00:06:54: Heathran from made a really critical distinction in this sources between resistance and resilience,

00:06:59: A very important distinction?

00:07:00: Yeah

00:07:00: And we really need to define these terms because people use them interchangeably when they absolutely shouldn't.

00:07:05: So resistance is your endpoint.

00:07:07: detection and response It's you're multi-factor authentication Your firewalls.

00:07:12: Resistance tries to stop The attack at the front door.

00:07:15: Right, but resilience starts with a deeply uncomfortable engineering assumption which is prevention will eventually fail.

00:07:21: The adversary's gonna get inside?

00:07:23: They are!

00:07:24: the question you have to ask yourself as a leader Is not how do we keep them out one hundred percent of time?

00:07:29: But rather can your business still achieve its operational objectives when the network goes dark?

00:07:35: right Can you fail gracefully into a degraded operational state use out-of and management and, you know keep the physical forklifts moving.

00:07:44: Because if you rely purely on resistance The results are catastrophic.

00:07:48: Yeah Let's go back to that bank vault analogy we opened this show with because Esaf Ashkenazi provided a textbook breakdown of recent TV ING breach.

00:07:57: Oh yeah This is massive streaming service And thirty nine point five million user accounts were exposed

00:08:03: And mechanics about how this unfolded Are just painful For any security professional To analyze.

00:08:09: So painful

00:08:10: it started with a single stolen developer key.

00:08:14: Now, in a resilient system with the strict least privilege architecture, a compromised developer key should really only give the attacker access to one maybe highly compartmentalized sandbox...

00:08:25: Right limited blast radius!

00:08:27: Exactly.

00:08:28: but because of incredibly poor access controls that one single stolen key unlocked three hundred and sixty-one different development projects across the entire organization.

00:08:37: But the real kicker is what they actually found inside those projects.

00:08:41: The attacker uncovered forty-one production keys hardcoded directly into the source code.

00:08:45: Unbelievable!

00:08:46: I know, developers do this out of convenience right?

00:08:48: Just to speed up their CICD pipelines.

00:08:51: but it's exactly equivalent of taping the vault combination onto front door.

00:08:56: It completely negates whatever expensive multi factor authentication or perimeter resistance layer you bought

00:09:01: Which ties directly in with regulatory hammer that will fall on industry.

00:09:06: Helen Yu pointed out the implications of EU Cyber Resilience

00:09:09: Act.

00:09:10: Security by design is no longer just a technical engineering problem anymore.

00:09:14: It is a literal leadership obligation.

00:09:17: Under this act you have to prove supply chain provenance

00:09:21: meaning what exactly?

00:09:22: You need a verified software bill of materials.

00:09:25: if you are shipping products with hard-coded keys or unpatched open source dependencies The liability shifts straight to the executive level with massive regulatory consequences.

00:09:36: Wow

00:09:36: By the way, if you are finding this deep dive into the underlying mechanics of tech and security.

00:09:41: valuable.

00:09:42: make sure to hit subscribe on your platform of choice so that we don't miss our future additions.

00:09:46: We're constantly tracking how all these architectures are evolving

00:09:49: And they are evolving incredibly fast in the adversary side though perhaps not exactly what mainstream media is portraying

00:09:55: it.

00:09:57: Well, there's this pervasive narrative right now that generative AI is handing every script kitty a magic wand to just breach enterprise networks effortlessly.

00:10:05: But Marcus Hutchins made a vital observation about the current reality of AI-assisted attackers.

00:10:10: Oh yeah He noted that unsophisticated attackers using AIs still severely struggle with complex exploit development.

00:10:17: Yes Hutchins detailed spending weeks trying to use frontier large-language models to exploit a critical remote code execution vulnerability, specifically CVE.

00:10:27: twenty-twenty four three eight zero six three.

00:10:30: And the models could write basic scripts sure but they failed completely at writing the actual exploit.

00:10:36: Because going from just triggering a bug to actually executing code remotely requires deep, deep understanding of environmental context.

00:10:43: Right?

00:10:43: It's not just copy-pasting code.

00:10:44: No and LLM doesn't inherently understand how to map memory layouts or bypass address space layout randomization Or chain together these highly specific kernel primitives.

00:10:55: The AI models trade time for money like they'll you write phishing emails way faster But they do not currently solve the deep architectural problem solving required For zero day remote execution.

00:11:05: Defenders are absolutely not waiting around to see if the attackers models get better at bypassing ASLR, right?

00:11:10: Right.

00:11:11: Jessica Dabbs and Beeple Seenhub both highlighted a massive shift discussed in recent major industry events like CrowdStrike Falcon and Rubrik Forward Tokyo.

00:11:20: Yeah!

00:11:20: The new standard shifting into production is agentic AI in cybersecurity

00:11:25: And we really need separate agents from chatbots here.

00:11:28: Totally different things.

00:11:30: A Chatbot analyzes logs tells human analysts what happened.

00:11:34: An AI agent is given the autonomy to actually take action.

00:11:37: Exactly,

00:11:38: we are moving towards systems where AI agents handle preemptive recovery and active defense.

00:11:45: if an agent detects anomalous encryption behavior that looks indicative of ransomware It doesn't just send a slackalite to an analyst.

00:11:53: Right, it actively quarantines the subnet and automatically rolls back affected servers into clean, immutable snapshots before human analysts even finishes their morning coffee!

00:12:05: And the resourcing for that race at national level is currently subject of pretty intense debate... Dave Schroeder shared a Wall Street Journal report noting that the US Cybersecurity and Infrastructure Security Agency, CISA recently had its budget cut from two point nine billion to two point six billion dollars.

00:12:21: Yeah!

00:12:22: And to be clear we are just looking impartially at the reported facts here.

00:12:27: according To The Report this funding reduction is leading to the closure of several regional security assessment programs for critical infrastructure.

00:12:35: Right purely reporting the numbers but it highlights a massive tension.

00:12:40: On one hand, you have this thirty-one trillion dollar infrastructure boom pushing compute out to the edge where the network acts as a physical sensor.

00:12:49: Yes and on the other hand The federal resources allocated to proactively assess the vulnerabilities of those expanding physical networks are contracting.

00:12:58: It places the burden of defense squarely on the enterprise itself

00:13:01: Which makes perfect sense when you look at where the enterprise money is quietly flowing right now, because while AI and edge compute are definitely defining today's spending.

00:13:09: The smartest capital is already pouring massive amounts of money into the underlying physics of the next paradigm.

00:13:15: Oh

00:13:15: this is a quantum computing inflection point?

00:13:17: Yes!

00:13:18: For long time, Quantum was basically treated as science fiction R&D project rate.

00:13:22: it was always decade away

00:13:24: Always ten years out

00:13:25: But Isabel Busquette shared data from BCG that completely shatters the narrative.

00:13:31: Enterprise quantum spending hit three hundred million dollars in twenty-twenty

00:13:35: five.

00:13:36: Wait, enterprise spending?

00:13:37: Yes!

00:13:38: And critical detail here is that enterprise spending has now officially surpassed both academia and government funding combined.

00:13:46: That's a real signal.

00:13:47: through all of this noise Sixty-two percent of the top twenty five global companies by market cap are now spending at least one million dollars annually on quantum initiatives.

00:13:57: Yeah, and a huge driver for this is pure FOMO.

00:14:01: I mean these executives were caught completely flat footed by the generative AI boom And they're basically refusing to let it happen again with quantum.

00:14:08: But they aren't just funding abstract theory anymore.

00:14:11: They are running real workloads today!

00:14:13: Andre Luis Rodriguez shared that Tier One banks are investing millions to actively use quantum annealers right now.

00:14:19: Now, To understand why a bank cares about this you really have look at how classical computing fails at massive optimization problems like say balancing of fixed income portfolio with discrete lot sizes.

00:14:29: It's a hugely complex math problem

00:14:32: is.

00:14:32: Let me throw an analogy At This to explain the mechanism because Optimization Math can get super dense.

00:14:38: Imagine you are a hiker and your trying to find the absolute lowest valley in massive rugged mountain range.

00:14:46: A classical computer is like a hiker who has to physically walk up an over every single hill, every ridge ,every mountain peak.

00:14:55: just check depth of valleys on other side.

00:14:58: They get stuck behind what are essentially mathematical ridges in the data, right?

00:15:02: And it just takes an impossible amount of time to find The true bottom.

00:15:05: Right whereas quantum mechanics allows the system To literally tunnel through those complex energy landscapes tumbling.

00:15:12: Yeah instead of climbing over the Mathematical Ridge the Quantum algorithm leverages a property called quantum tunneling to drop straight Through the mountain into the optimal risk-adjusted portfolio Valley.

00:15:22: you find the absolute most efficient answer In a fraction of the Time

00:15:26: And the hardware making that tunneling possible isn't just sitting in a university lab anymore, it's being commercialized at this incredible pace.

00:15:34: Steve McDowell reported on Quantinuum pulling an eighty-one million dollars in year to date bookings.

00:15:40: That is real enterprise revenue It

00:15:43: is!

00:15:43: They even sold a Helios quantum system outright to Oracle To be physically installed and integrated inside of traditional US data center.

00:15:51: The engineering breakthroughs across different qubit modalities are Compounding.

00:15:57: J. Gambetta reported that IBM RIS Nighthawk r-two processor is hitting a hundred thousand circuits per second.

00:16:03: Wow, yeah We often obsess over the raw number of quibits A machine has but throughput like The sheer volume Of quantum circuits executed per second Is just as vital for running complex algorithms at scale.

00:16:15: That hundred thousand mark is a twenty five x increase in throughput Over their previous generations.

00:16:20: and then you have the physical footprint problem being solved too.

00:16:23: Craig Lenahan noted that Japan just switched on Shunkai, which is a room-temperature neutral atom quantum computer approaching fifty cubits.

00:16:30: The mechanism there so crucial to understand.

00:16:33: traditional superconducting quantum computers like the ones IBM builds require extreme cryogenic cooling Like

00:16:41: deep space cold Exactly!

00:16:42: They have to be kept in dilution refrigerators at fractions of degree above absolute zero.

00:16:47: Just keep thermal noise from destroying the quantum state.

00:16:50: But Shinkai uses neutral atoms.

00:16:54: They use highly focused lasers, essentially optical tweezers to trap individual atoms in a vacuum and hold them perfectly still.

00:17:02: That

00:17:02: sounds like sci-fi

00:17:03: I know but because the atoms are neutral and isolated by lasers they don't interact with ambient thermal environment In the same way meaning you don't need a massive cryogenic chandelier To run it.

00:17:13: It proves that physical ecosystem is arriving at multiple totally viable forms.

00:17:19: You know, this hardware boom brings a massive terrifying shadow with it.

00:17:22: It does!

00:17:23: If fault-tolerant quantum hardware is arriving faster than anticipated... ...it fundamentally breaks the global encryption standards that currently protect every single piece of digital infrastructure.

00:17:33: we just put the first half of this deep dive.

00:17:34: talking about

00:17:35: Which brings us to postquantum cryptography or PQC?

00:17:43: Neha S and Rajeshwari Mishra both pointed out a strategy currently being utilized by nation-state actors known as Harvest Now Decrypt Later.

00:17:53: Right, for you listening what this means is that attackers are actively siphoning your highly encrypted sensitive data off the wire today!

00:18:02: right now.

00:18:03: They're recording your encrypted VPN traffic, your secure communications and they are just dumping it onto cheap storage arrays –they know that can't read today-.

00:18:11: they're sitting on it waiting for the quantum computer of tomorrow to be powerful enough to slice through

00:18:16: encryption.".

00:18:17: And we have concrete proof that threat is moving from theoretical math.

00:18:22: Ryan Herring and Martin Rudd Teller shared that IonQ just published a fully compiled end-to-end blueprint for breaking two hundred fifty six bit elliptic curve signatures.

00:18:31: They actually published the Blueprint!

00:18:32: they did, they didn't just write a white paper on Shor's algorithm...they mapped the exact logical operations down to physical quantum error correction primitives

00:18:40: And the math on that blueprint was terrifyingly precise.

00:18:44: They calculated it would take both some twenty thousand physical trapped Ion quibbets and twenty five point seven days of continuous runtime to completely break sec up two ferry six K one, yeah?

00:18:55: That is the exact cryptographic curve securing massive amounts of digital assets financial transactions and secure communications.

00:19:02: today

00:19:03: It gives adversaries a literal roadmap.

00:19:05: once a machine with twenty thousand highly coherent trapped ion quibbitz comes online That encryption is effectively zeroed out.

00:19:13: Okay, wait let me push back on the panic here just a little bit.

00:19:16: Craig A. Linton posted a really interesting counterpoint in The Sources.

00:19:19: he noted that the RSA-DF was just solved using classical computing not quantum and you made the argument that Q-Day you know the day quantum computers break the internet isn't actually imminent because the standard RSA twenty forty eight encryption Is still mathematically safe from classical attacks?

00:19:36: And the Quantum hardware to run that massive ion cube blueprint doesn't actually exist yet.

00:19:41: Right, we are at roughly one thousand physical quibits today not twenty-thousand?

00:19:44: Exactly so.

00:19:45: why the immediate alarm bells?

00:19:46: it's an incredibly fair question and its where a lot of IT leaders get stuck.

00:19:50: but Ruben Maher provided the vital insight that changes calculus here.

00:19:55: It is about exactly when cryptographically relevant quantum computer finally arrives on timeline.

00:20:01: Then what does this

00:20:03: mean?

00:20:03: It is about the transition window.

00:20:05: If you look at the history of enterprise IT, migrating cryptography across an entire global organization updating public key infrastructure The hardware tokens and root certificates across every single legacy system takes roughly ten to twenty years to complete fully.

00:20:22: Ah

00:20:23: I see.

00:20:23: so the clock isn't waiting for it to arrive?

00:20:26: The clock has already ticking on migration itself.

00:20:29: Exactly.

00:20:30: If your highly sensitive corporate data, or like national security secrets have a shelf life of high value that lasts more than five years and the migration takes fifteen years to execute.

00:20:41: you are already deep in The Danger Zone today.

00:20:43: You're completely exposed And That means we cannot just wait for individual IT departments To slowly organically adopt post-quantum standards.

00:20:50: Derek Dye made a compelling case for how we solve this structurally.

00:20:54: We need top-down regulatory mandates, we need the major cloud service providers to enable PQC by default in their hypervisors so that protection is just completely transparent to end user.

00:21:07: And most importantly...we have build crypto agility into our systems from day one.

00:21:11: You can't hard code cryptographic libraries into your infrastructure anymore.

00:21:15: No not at all!

00:21:16: you design architecture and swap out encryption algorithms.

00:21:20: why you issue a routine software patch without tearing down the entire network.

00:21:24: Which really brings every single theme we've discussed today full circle, I mean... We are at the very beginning of a thirty one trillion dollar infrastructure supercycle.

00:21:34: Yeah Over the next few decades were going to build massive power dense liquid cooled data centers in human history To support AI and edge computing.

00:21:44: But you have to realize that those very same facilities will eventually house the fault-tolerant quantum computers, which render today's standard encryption entirely obsolete.

00:21:54: So a profound question as you look at your own organization roadmap is this... Are we designing these new massive tech ecosystems with deep cryptographic agility to seamlessly swap out their foundational security in five years?

00:22:10: or in our rush to build the future of AI, are we just constructing bigger faster

00:22:27: targets?