Best of LinkedIn: Digital X 2026

00:00:00: This episode is provided by Thomas Allgaier and Franis, based on the most relevant LinkedIn posts about DigitalX.Franis supports ICT enterprises in the form of delivering precise ICT market and pricing intelligence that analysts subscriptions and existing databases cannot provide.

00:00:17: you can find more info

00:00:19: Yeah.

00:00:19: So today we are digging into the top takeaways from DigitalX, twenty-twenty six which just wrapped up right

00:00:26: straight out of The RhinoHoff in a cologne

00:00:27: exactly.

00:00:28: We've been going through the most impactful LinkedIn posts From the attendees you know Just to see what the actual professionals Are talking about.

00:00:34: yeah and if there's one big cohesive message To pull from all that noise I mean it's really the era Of AI.

00:00:40: experimentation is just entirely over

00:00:43: totally Over.

00:00:44: the mandate across the board was worth stat wow Right, so value instead of wow.

00:00:49: Yeah

00:00:49: industrializing AI to create actual tangible economic value for you the enterprise user

00:00:56: right and To get that value I mean we really have to look at the physical infrastructure first like imagine trying to fit Ten thousand cutting-edge Nvidia Blackwell GPUs through this narrow underground tunnel in Munich,

00:01:08: right?

00:01:09: It sounds like something out of a heist movie.

00:01:10: it Really does yeah But it is actually the physical reality Of what it takes to build the Industrial future right now

00:01:17: Because for the last couple of years, everyone has been treating AI like it is this... I don't know.

00:01:23: This weightless virtual magic trick that just happens in a browser window?

00:01:27: Yeah type of prompts you get a picture.

00:01:29: Magic

00:01:30: Exactly.

00:01:31: But when look at enterprise tech landscape today It's incredibly heavy.

00:01:36: I mean, it requires moving Earth laying fiber building these massive data centers.

00:01:41: Which brings us back to that underground tunnel.

00:01:43: Dr.

00:01:44: Ferri Abul-Hasan posted about the launch of a new industrial AI cloud in Munich and logistics have pulling this off.

00:01:50: just six months are frankly staggering.

00:01:53: Yeah they really are.

00:01:54: We're talking about one hundred fifty truckloads equipment.

00:01:56: They laid down four hundred kilometers And yeah, they had to navigate all those Blackwell GPUs through an underground passage just to get them to the secure server racks.

00:02:06: Well and the output of that logistical nightmare is what actually matters to the enterprise space right?

00:02:11: That single facility increases Germany's available AI compute capacity by roughly fifty percent overnight!

00:02:18: THAT IS

00:02:19: INSANE!!

00:02:19: It delivers upto point five Exif Law PS of Compute Power & twenty petabytes of storage.

00:02:25: I mean... Just to put point five exoslopias into perspective for a second.

00:02:29: Yeah, that is five hundred quadrillion calculations every single second.

00:02:33: yeah You don't need that kind of brute force too you know.

00:02:36: write some marketing copy or generate A fun image.

00:02:39: right

00:02:39: no you

00:02:40: need That power.

00:02:41: when?

00:02:41: You are trying To run a real-time digital twin Of an entire manufacturing plant

00:02:45: exactly if you Are simulating fluid dynamics supply chain logistics and robotics telemetry all at the exact same time.

00:02:52: Your compute has to be massive

00:02:54: And it has to Be local.

00:02:55: Yes, local and sovereign.

00:02:57: The conversations from the event clearly positions sovereignty as the foundation of this new infrastructure

00:03:02: Which makes sense.

00:03:03: Yeah Like, Nixon and Markus Erl both highlighted how the ServiceNow AI control tower is now running directly on Deutsche Telekom's Sovereign T-Cloud.

00:03:14: And Philip Sauertyk also posted about a new zero outage.

00:03:18: two point O agreement with Dell Technologies.

00:03:22: so they are building these open sovereign cloud architectures focused entirely on operational excellence.

00:03:28: but let's be real.

00:03:29: for second though Half the time I hear the word sovereignty in tech, it just sounds like you know a political buzzword designed to keep data trapped inside national

00:03:38: borders.

00:03:39: Sure yeah i get that.

00:03:40: so is there actual tangible operational value here?

00:03:43: or?

00:03:43: Is it just regulatory compliance kind of dressing up his innovation?

00:03:47: Well, that skepticism is super common but sovereignty in this context is fundamentally about operational control and really your exit strategy.

00:03:54: Okay

00:03:55: how so?

00:03:55: It has the guarantee that you're critical infrastructure won't go down just because a foreign provider changed their terms of service or got caught on geopolitical crossfire.

00:04:04: true sovereignty as power choice It means an enterprise can explicitly decide,

00:04:09: hey

00:04:10: this highly sensitive proprietary data must stay in our Munich facility.

00:04:15: but this non-sensitive workload that

00:04:16: can scale globally across public clouds.

00:04:19: Exactly!

00:04:20: Is basically like building a high speed rail network.

00:04:22: you can have the fastest most advanced AI trains in the world But if you don't actually own the tracks yeah The sovereign cloud infrastructure I mean Someone else can always flip the switch and stop your trains from running.

00:04:34: You are completely at the mercy of the track owner

00:04:36: Precisely.

00:04:37: And what Deliria's zero outage?

00:04:40: two point oh framework acknowledges is that in a hyper connected physical environment an Outage isn't just you know, a four or four error on our website anymore.

00:04:48: No it was physical

00:04:49: Right.

00:04:50: If the cloud drops out for even a few seconds, our robotic arm on an assembly line might stop mid-motion.

00:04:56: That causes a multi million dollar supply chain pile up.

00:04:59: so that standard of reliability just has

00:05:01: to be absolute.

00:05:02: Okay.

00:05:02: So with heavy lifting done and sovereign rails in place The payload is starting to shift.

00:05:07: Because looking at the LinkedIn post from event There was massive pivot toward agentic AI?

00:05:12: Yes absolutely

00:05:13: We are moving away.

00:05:14: form generative A.I.. that talks to you Toward Agentic Ai...that actually acts on your behalf.

00:05:20: And the mechanism here is a profound shift in software architecture.

00:05:24: Okay, break that down

00:05:25: for us?

00:05:25: Well generative AI predicts.

00:05:30: Agentic AI uses a large language model as a reasoning engine to understand the intent of a goal.

00:05:36: And then it is actually given access to tools, so APIs databases software interfaces and It uses those two.

00:05:43: figure out the steps required to achieve that goal dynamically.

00:05:47: So its kind of difference between a sous chef who only chops carrots when you specifically instruct them chop carrots versus ahead chef Who looks at the ingredients in the pantry?

00:05:58: I'm gonna make a beef stew and then just orchestrates the whole kitchen to do it without you monitoring every step.

00:06:04: That

00:06:04: is great analogy!

00:06:05: Yeah, Ralph Pechman shared some really cool insights on the TAI agent platform which uses NAN as its workflow engine.

00:06:14: yeah The objective is take these clever AI models and embed them into repeatable secure business processes.

00:06:22: NAN essentially acts as the connective tissue, right?

00:06:24: Allowing the AI to reach into legacy enterprise systems.

00:06:27: maybe pull a finance report cross-reference it with compliance rules in this ever database and draft an action

00:06:32: plan Exactly.

00:06:34: And Julianne Palmer and Matthias Buba echoed this exact sentiment when they talked about the agent-force partnership between Salesforce, and Telecom The

00:06:43: agentic enterprise.

00:06:44: Yes!

00:06:44: The vision there is that humans aren't managing software applications anymore.

00:06:49: Humans are managing digital agents who in turn managed the software.

00:06:53: Which is wild because, The barrier to entry for building these agents has practically vanished!

00:06:58: Oh totally?

00:06:58: I mean...the traditional model required a massive team of specialized developers writing really rigid Python scripts But now natural language is basically the compiler.

00:07:08: Yeah if you can speak and code

00:07:10: Exactly Oliver Grabar posted perfect example from this from the Sparkis hosted Vibe Coding area at that event.

00:07:16: Oh

00:07:17: yeah the vibe coding thing?

00:07:19: regular mid-market corporate clients, so accountants HR managers logistics directors with zero coding experience.

00:07:26: They literally sat down and used a platform called lovable to build fully working AI agents in about an hour.

00:07:33: that is amazing right?

00:07:35: you just describe your workflow In plain English or German?

00:07:38: And the tool writes The front end.

00:07:41: it writes the back end connects the API's and deploys It.

00:07:44: the underlying complexity Is entirely abstracted away

00:07:47: Which is great, but that democratization of development unlocks immense speed while also introducing an entirely new class of risks.

00:07:54: Right!

00:07:54: When domain experts can spin up autonomous agents in an hour those agents start taking actions in the real world?

00:08:00: Yeah wait hold on if non-technical people are building AI agents over their lunch break and those agents moving off our screens to control physical supply chains or issue purchase orders Aren't we creating this chaotic Wild West?

00:08:12: Yeah, it's a huge concern.

00:08:14: Like

00:08:14: who is actually accountable when an agent hallucinates and orders ten thousand tons of steel by mistake?

00:08:19: Accountability is definitely the defining crisis of this transition especially as agents move into what Kristoff Leggiter has post termed physical AI.

00:08:29: Right, physical AI?

00:08:30: Yeah we are watching software agents merge with physical IoT devices and digital twins.

00:08:37: We're talking about autonomous industrial systems making real-time decisions on factory floors

00:08:42: And a hallucination out of text prompt is just embarrassing.

00:08:46: A hallucination in a physical AI system that is catastrophic damage

00:08:50: Which is exactly why Leggett Is pushing so hard for our common autonomy framework and strict new ISO standards for manufacturing.

00:08:57: So how do you even standardize that?

00:08:59: Well, you cannot certify an AI model the way you certified traditional software because in AIS decisions are non-deterministic.

00:09:06: It might react differently to the exact same stimulus twice.

00:09:10: so instead of certifying the intelligence You have to certify the boundaries of the sandbox it operates at.

00:09:14: Oh interesting!

00:09:15: You define the physical limits The human oversight triggers and the fail safes

00:09:19: Because the moment an AI agent can move a robotic arm or reroute a power grid, The traditional IT security perimeter instantly dissolves.

00:09:27: Yes!

00:09:28: It's like cyber-security used to be a moat around a castle.

00:09:31: You build high firewalls pulled up a drawbridge and trusted everything inside But with Agenetic AI pulling data from third party APIs & IoT sensors deployed across the globe... ...the Castle walls are gone

00:09:44: Right..The attack surface is essentially infinite.

00:09:46: now To illustrate how physical this threat landscape has become, Ferry-Ebel Hassan also detailed a live drone detection demo at the event.

00:09:55: Oh!

00:09:55: This was crazy... Yeah they

00:09:56: were using advanced radio frequency sensors But the goal wasn't just to spot an unauthorized drone hovering over a facility.

00:10:02: Right?

00:10:03: The AI actually analyzed the control signal, to instantly triangulate and locate the physical position of the pilot on the ground.

00:10:10: That

00:10:11: is wild!

00:10:12: Yeah You are defending against a physical threat by analyzing the digital spectrum in real time And then sending physical security personnel To intercept the human holding the joystick.

00:10:22: That's what protecting critics you know, critical infrastructure like water supplies and logistics hubs actually looks like today.

00:10:29: You need a system that bridges the digital and physical worlds instantaneously because if an AI agent can execute a complex workflow in milliseconds well a malicious AI can execute and attack just as fast.

00:10:42: human response times are no longer sufficient.

00:10:45: No Not at all.

00:10:46: You need a digital immune system, right?

00:10:48: And an immune system doesn't wait for the central brain to approve of response to a virus.

00:10:52: Right it detects the anomaly and neutralizes it autonomously on a cellular level.

00:10:56: Yeah, and Helmut Reisinger and Thomas Mazicik highlighted The Sovereign Cortex partnership with Palo Alto Network which does exactly this.

00:11:03: okay.

00:11:04: by using AI driven security operations They are correlating anomalies across network endpoint in cloud environments To shrink incident resolution times from days down to mere minutes

00:11:14: well days to minutes.

00:11:16: But no single vendor has the visibility to build that immune system alone,

00:11:20: right?

00:11:21: Exactly!

00:11:22: It requires a radically open ecosystem.

00:11:25: Nirani Azizi discussed the launch of T-Secure Fabric positioning it as a global partner based secure networking platform and Tima LeBrens also celebrated Deutsche Telecomers recognition.

00:11:45: So the defense mechanism really has to be as networked and automated, As the systems it's trying to protect.

00:11:51: Absolutely By the way if you're finding this kind of breakdown helpful for keeping ahead Of these massive shifts in enterprise tech You should absolutely subscribe To This Deep Dive.

00:12:00: We do a heavy lifting of sorting through The noise pulling out the actual operational mechanisms so that

00:12:06: Yeah, and staying ahead of the curve is crucial because the final theme that emerged from DigitalX.

00:12:12: Well it actually had nothing to do with servers AI agents or cybersecurity frameworks

00:12:16: right?

00:12:16: The tone really shifted.

00:12:18: It

00:12:18: did!

00:12:18: The technology stack is largely ready...the computer's there but the agents are being deployed..The security ecosystems are forming.

00:12:25: What's the holdup?

00:12:26: The

00:12:26: bottleneck is the human element.

00:12:28: yeah the conversation got surprisingly philosophical.

00:12:31: We saw a major political and societal voices stepping into the tech debate.

00:12:35: Yeah, that's Angela Merkel Robert Habeck Hendrick Foost.

00:12:38: And just to be clear we're not taking any sides here.

00:12:41: We're just reporting impartially on what the attendees highlighted from these sessions.

00:12:46: right of course

00:12:47: But they approach this not as an engineering problem but is a crisis of psychology in leadership.

00:12:53: Former Chancellor Muckel issued a very direct challenge.

00:12:57: She flat out said we are not excited enough

00:13:00: which was a powerful statement.

00:13:01: She was calling out a pervasive culture of risk aversion, urging leaders to embrace the transformation while maintaining independent human judgment.

00:13:09: Yeah and Tim Hodges took aim at this directly too identifying what he called The German Skepticism Bubble.

00:13:14: Yes And his solution forcing executives into kill-the-company workshops.

00:13:20: I love that concept!

00:13:21: I

00:13:21: love the mechanics of A Kill the Company Workshop.

00:13:24: You take your executive team Split them in half and assign one group to act as a ruthless startup with zero legacy tech debt, an unlimited access to AI agents.

00:13:32: Just

00:13:33: tell them go destroy us!

00:13:34: Exactly their only goal is to figure out how to destroy your current business model.

00:13:39: it shatters that cautious linear thinking because It forces you realize the risk of inaction is vastly more dangerous than the risk of experiment.

00:13:49: When the urgency for that experimentation is driven by demographic reality, Lara Sophie Bother shared The Math and it's frankly sobering

00:13:57: It really is.

00:13:57: Germany is projected to lose thirty percent of its working population by twenty forty.

00:14:02: Thirty percent?

00:14:03: Yeah, so if you are running a mid-market manufacturing firm right now this isn't some abstract futuristic problem.

00:14:09: You're already feeling the talent gap.

00:14:11: AI

00:14:12: and robotics are not here to steal jobs.

00:14:14: They are only viable mechanism to compensate for human workforce.

00:14:18: we simply will not have.

00:14:19: Exactly And Bother pointed out that Germany actually holds massive structural advantage here.

00:14:25: Silicon Valley might Consumer data?

00:14:27: Sure, but Germany's ultimate treasure is decades of proprietary physics-based industrial engineering telemetry.

00:14:35: Oh that's a really good point because large language models are becoming starved for high quality non synthetic training data.

00:14:41: right?

00:14:42: so if you feed forty years of precision manufacturing data into sovereign agentic AI You create an economic engine.

00:14:50: the consumer tech companies simply cannot replicate.

00:14:52: yeah they just don't have.

00:14:54: Roman Howe used the classic lily pond riddle to explain where we are on this adoption curve, which I thought was brilliant.

00:15:01: The exponential growth one?

00:15:03: Yeah!

00:15:03: So if a pond has lilies that double every day and it takes thirty days to completely cover the pond... On what day is the pond half full?

00:15:10: Day twenty-nine

00:15:11: Exactly.

00:15:12: For twenty eight days the pond looks mostly empty And then suddenly It just chokes the entire ecosystem.

00:15:17: Wow How argues that AI capabilities are on the exponential curve, but corporate planning remains strictly linear.

00:15:24: We're treating AI like a slightly faster spreadsheet completely underestimating the exponential leap in capability.

00:15:31: Yeah technology is scaling exponentially But human trust scales linearly if it scales at all.

00:15:36: Very true.

00:15:37: And this where historian Yuval Noah Harari delivered really profound warning during his keynote which was widely discussed by attendees like Dominic Roche and Stephanie Lysen.

00:15:46: Harari used the metaphor of The Tower Of Babel.

00:15:49: All right, the breakdown of language?

00:15:51: Exactly!

00:15:52: Language is the foundational layer of human cooperation.

00:15:55: It's how we negotiate contracts write laws form institutions and establish reality.

00:16:01: Yeah And Harari argues that as AI masters language it actually gains the ability to hack our cooperative infrastructure

00:16:10: Which is terrifying when you think about the implications for enterprise compliance in HR.

00:16:14: Right,

00:16:15: if you get a memo from your CEO and the tone...the vocabulary of the cadence sound perfectly like your CEO but it was actually generated by an autonomous agent trying to optimize a quarterly metric based on a misinterpreted prompt.

00:16:30: Where Is The Truth?

00:16:31: That's the million dollar question!

00:16:33: If

00:16:33: language is how we build trust And machines can perfectly mimic language How do companies maintain a single source of truth going forward?

00:16:41: Trust is the ultimate currency of the digital age.

00:16:44: When you can no longer implicitly trust the digital outputs on your screen, You have to anchor your trust in the architecture itself!

00:17:00: The mandate from DigitalX twenty-twenty six is clear The tech works, the pilots are done and the demographics demand immediate industrialization.

00:17:09: Absolutely.

00:17:10: But I want to leave you with a final thought to mull over building on that tension between risk aversion and trust.

00:17:16: we talked A lot about the skepticism bubble in Europe's hesitation compared To the aggressive move fast and break things.

00:17:23: culture We see elsewhere.

00:17:25: right because the prevailing narrative is always That hesitation as a weakness.

00:17:28: exactly.

00:17:29: but what if?

00:17:31: If that deep-seated cultural hesitation is forcing Europe to build highly secure, rigorously sovereign and deeply standardized AI focusing heavily on physical safety.

00:17:41: And verifiable operational control could this cautious high regulated approach actually become its greatest competitive advantage?

00:17:49: That's a fascinating way to look at it

00:17:51: Right.

00:17:52: In a global market that is suddenly realizing the catastrophic risks of untethered AI, A foundation built entirely on trust and security might be exactly what the enterprise world is desperate to buy.

00:18:03: Yeah because when The Wild West becomes too chaotic... ...the Market always retreats to safety!

00:18:08: The massive physical weight of those servers, the strict ISO standards.. ..The rigid sovereignty requirements.

00:18:14: That friction may just be an anchor that makes the future viable.

00:18:17: If you enjoyed this episode new episodes drop every two weeks.

00:18:20: Also check out our other editions on Cloud Insights and Sovereignty, ICT & Tech Insights, Digital Products & Services, AI and Agentec Systems, Green ICT in Sustainable AI, DefenseTech and HealthTech.