Best of LinkedIn: Cloud Insights CW 11/ 12

00:00:00: This episode is provided by Thomas Allgaier and Frennis based on the most relevant LinkedIn posts about cloud in calendar weeks, eleven and twelve.

00:00:08: Frenis has built a sovereign cloud market radar for ICT providers with weekly hot news monthly reports quarterly leadership presentations an AI podcast for field teams.

00:00:19: you can find more info.

00:00:23: pretty wild reality that over seventy percent of Europe's so-called sovereign public cloud is actually controlled by just three American companies.

00:00:33: Right, I mean the whole era of the Cloud being this.

00:00:36: you know this invivible magical place where you'd just swipe a credit card and all your infrastructure headaches disappear... That is officially dead!

00:00:43: Oh

00:00:43: it really is completely dead when you look at the raw signals coming from tech leaders on the front lines right now The entire conversation has just fundamentally shifted.

00:00:51: Yeah And if you are tuning in whether you know, managing an engineering team or trying to map out a corporate IT strategy.

00:00:57: The focus is no longer just about adoption

00:01:00: Exactly.

00:01:00: It's not about simply lifting and shifting workloads into a data center anymore, right?

00:01:04: The whole industry is currently fighting these three massive battles And it's over control its over economics and operational outcomes.

00:01:14: Okay Let's unpack this.

00:01:15: our mission for this deep dive Is to really cut through all the marketing noise.

00:01:20: Right we're gonna look at curated insights shared across LinkedIn during calendar weeks eleven and twelve.

00:01:25: Yeah We're diving into the battle for control in jurisdiction for financial accountability and the battle for operational discipline.

00:01:34: And I want to jump straight into that first one because sovereignty cloud is a term that gets thrown around so much... Yeah,

00:01:40: constantly!

00:01:41: Right usually it's this really dry compliance checkbox.

00:01:45: In today's geopolitical climate, it is basically the core strategic requirement.

00:01:51: Yeah because to understand control in the modern cloud you really have to look at the physical and logical reality of where your data actually sits...

00:01:58: ...where the servers are

00:01:59: Exactly.

00:02:00: And the major hyperscalers are making these concrete incredibly expensive moves to prove that they can offer this control, like Sergio Canbello-Bajillo provided this brilliant breakdown of those recently... Oh

00:02:11: I saw it!

00:02:12: He was analyzing the new AWS European Sovereign Cloud right?

00:02:15: Yeah yeah and he pointed out something crucial.

00:02:18: AWS isn't just opening up a new region in Europe.

00:02:21: They're creating a completely new sixth partition.

00:02:26: Wait, I think we need to define that for the listener.

00:02:28: Because a partition sounds like...I don't know-a minor technical detail but it's actually massive distinction.

00:02:35: It

00:02:35: really is!

00:02:36: The difference between renting a room in someone else house versus building completely separate houses on totally different electrical grid.

00:02:45: In whole history of AWS they have rarely created entirely new partitions.

00:02:50: This move means you're dealing with distinct accounts.

00:02:52: Your standard AWS credentials will not work there.

00:02:56: just completely separate?

00:02:57: Completely separate.

00:02:58: There are strict regional boundaries for identity and access management.

00:03:03: Sergio highlighted that there is absolutely no cross-partition VPC peering

00:03:07: which in plain English means you can't just build a digital bridge between your new European data environment and your old US based systems.

00:03:16: Exactly, the networks are fundamentally logically isolated from each other.

00:03:20: I mean even management consoles are separate And billing is entirely in euros.

00:03:25: They're engineering this hard separation to keep data strictly within European borders.

00:03:30: But i have push back here.

00:03:32: Well, I should say the industry professionals on LinkedIn are pushing back.

00:03:35: Right?

00:03:35: Because is this just like putting a high tech safe inside your house but someone else still legally holds a master key?

00:03:41: that Is the million dollar question.

00:03:43: yeah

00:03:43: there's This highly factual really neutral debate happening right now about this structural reality.

00:03:50: Anthony Monti and Vasco Afonso both pointed out That despite these new partitions over seventy percent of Europe public cloud it Still operated by three US based companies.

00:04:00: And that structural reality brings us directly to the tension around the U.S.

00:04:05: Cloud

00:04:05: Act, right?

00:04:06: The concern being raised across the industry and again this is just what people are discussing Is that regardless of where the physical servers are located or how isolated the network partition is if the parent company is headquartered in the United States those subsidiaries might still be subject US legal jurisdiction and

00:04:24: data requests exactly.

00:04:26: And just to be clear, To the listener you know we are not taking sides on The legal framework of the cloud Act here.

00:04:32: We're simply reporting that this specific legal reality is a massive point Of friction right now?

00:04:38: Oh absolutely It's huge issue.

00:04:40: European enterprises in governments Are actively struggling to figure out what the word sovereign actually means.

00:04:45: if A foreign government can theoretically compel Data

00:04:48: access.

00:04:48: yeah and That Friction is forcing providers to engineer some pretty extreme solutions, To answer this demand for absolute control.

00:04:57: I mean look at how Microsoft's positioning itself.

00:04:59: Yeah Jonathan Vella shared some really fascinating insights on this.

00:05:03: On Microsoft push into what they're calling disconnected operations.

00:05:07: Right we are talking about Azure local and Foundry Local.

00:05:10: This where get in the Submarine analogy right?

00:05:13: Disconnected operations mean running advanced large-scale AI models with zero internet connectivity.

00:05:20: Zero, none!

00:05:21: It's

00:05:22: literally like a submarine navigation system.

00:05:24: it has to possess all the intelligence internally because it physically cannot surface.

00:05:30: To ask a satellite for directions.

00:05:32: that is a perfect way to visualize it.

00:05:34: I mean if you're operating in a highly regulated industry, say defense or critical national infrastructure.

00:05:39: You just cannot afford a persistent connection to the public internet right?

00:05:43: The risk is too high

00:05:44: exactly so.

00:05:45: Microsoft has allowing these organizations to run multimodal AI inside their own sovereign boundary fully offline and Antonio F actually highlighted A similar philosophical approach from Oracle with what they call realms.

00:05:58: So how does a realm differ From What AWS Is Doing With Their Partitions?

00:06:02: Well, instead of offering one giant global cloud with a few isolated partitions, Interneo describes how Oracle is building dozens.

00:06:14: Yeah, each realm is tailored specifically to the unique legal geographic and security requirements of different organizations or governments.

00:06:23: It's basically an acknowledgement that one size absolutely does not fit all when international laws are involved.

00:06:29: Well

00:06:29: because this ongoing friction regarding jurisdiction we're seeing a major coordinated push for homegrown domestic infrastructure in Europe.

00:06:38: Oh yeah big time.

00:06:39: Alexander Zahetmeyer noted a recent open letter that was signed by over thirty European cloud CEOs.

00:06:45: They are actively demanding that domestic providers be prioritized by European institutions to ensure true digital sovereignty,

00:06:53: right?

00:06:53: They're essentially saying we cannot rely on American tech companies to protect European data sovereignty

00:06:58: and those domestic alternatives are actually proving their capability in real time like Elkanderl shared a great real world example of this.

00:07:06: oh the Deutsche Telekom one.

00:07:07: yeah

00:07:08: Deutsche Telekom's T-Cloud public is currently running the Moodle education platform for roughly six hundred and twenty thousand users across Germany.

00:07:17: Wow!

00:07:18: I know right?

00:07:19: That it not a small experimental project that has massive, highly sensitive workload involving personal data of hundreds thousands students operating entirely on European built and operated infrastructure.

00:07:33: That's huge, but you know before we move off the topic of sovereignty there is this fascinating twist to all of these and it has been driven entirely by AI.

00:07:41: Oh right.

00:07:42: Stefan Gimmison posted about their recent boom in what he was being called Vibe Coding.

00:07:46: I have seen that term floating around its essentially using AI coding agents where a user just types a prompt describing an app they want Deploys it

00:07:56: exactly.

00:07:57: It's totally revolutionizing how fast you can build software, but Gimmison points out this massive almost invisible blind spot.

00:08:04: Well thousands of these apps are being built every single day by European developers But by default These AI tools are funneling massive amounts of European user data into us-based back ends.

00:08:16: Oh wow

00:08:17: Yeah, the developers typing the prompts often don't even realize that AI agent is defaulting to a server in Virginia.

00:08:26: So AI is moving data faster and more fluidly than human engineers ever did which I guess is rapidly accelerating the urgent need for default European infrastructure.

00:08:36: Precisely Now, if Sovereignty is the battle for controlling legal and physical reality of your cloud.

00:08:42: The second major theme we are seeing in this data is the Battle For Controlling the Financial Reality.

00:08:48: We're talking about fine ops but just like sovereignty...the definition of Fine Ops has completely transformed!

00:08:53: It's no longer a passive exercise looking at a dashboard with past expenses asking our team why that number was so high.

00:09:00: Yeah.

00:09:01: If you are listening to this right now and you manage an engineering team, You have probably stared at a massive unexpected AWS bill And felt totally powerless To do anything about it.

00:09:10: We've

00:09:10: all been there

00:09:10: Right?

00:09:11: Diana Ramos actually posted About this exact psychological dynamic.

00:09:15: She noted A striking observation Which is that simply showing An engineering team A dashboard of their cloud costs Does absolutely nothing to change Their behavior

00:09:25: which I mean seems counterintuitive At first glance.

00:09:28: if the teams sees They're burning through cash the logical assumption is they would optimize their code.

00:09:33: You'd think so, but Ramos points out that if engineering team doesn't have true ownership of underlying architecture like... If structural decisions were made six months ago by an enterprise-architecture or forced to share clunky infrastructure with three other departments….

00:09:50: They physically cannot influence cost!

00:09:53: Yeah, the dashboard just becomes this terrifying number that they are judged on but can't control.

00:09:58: And Rezatee echoed this exact sentiment in his own post.

00:10:02: He argued that Phenops is entirely useless unless you actively empower the engineering teams to make architectural changes rather

00:10:09: than Just leaving The financial burden On the platform or IT Teams To police.

00:10:13: What's fascinating here is that the old model of FindOps, it essentially like showing someone massive itemized bill for a multi-course dinner after they have already eaten.

00:10:23: You can't uneat steak but you still pay to restaurant.

00:10:26: so how do these companies stop ordering process before the bill gets run up?

00:10:31: That is where industry is aggressively pivoting toward what called shift left FindOps.

00:10:36: Okay So what does the actual mechanism there?

00:10:38: How Does Shift Left Work in practice.

00:10:41: So shift left means moving the cost analysis to the very beginning of this software development life cycle, Benjamin Laneras pointed out that Gartner recently recognized Harness as a sole leader in this exact space.

00:10:52: okay

00:10:53: instead of reporting metric after deployment costs becomes an active input during design and pull request phase.

00:10:59: wait

00:10:59: so like before code is even merged.

00:11:01: exactly imagine engineer writing code to spin up new database In a shift-left model, the automated integration pipeline will actually calculate the projected monthly cost of that database.

00:11:13: And if it exceeds a predefined budget... ...the system will automatically reject code deployment before ever goes live.

00:11:19: That is brilliant!

00:11:20: You are putting price tag on menu for engineers to order.

00:11:24: infrastructure and scope of what Finops covers is exploding beyond just server costs too.

00:11:30: Stephen Old noted in his insights that Finnoff's mandates are rapidly expanding.

00:11:35: Teams are now being asked to track AI spend, sauce application costs even complex software licensing across the enterprise.

00:11:42: It is evolving into a unified discipline for all technology economics which brings us to most urgent financial crisis facing IT leaders right now.

00:11:51: Let me guess... The generative AI cost explosion

00:11:54: Nailed it.

00:11:55: Mostafa Zaffer from IBM highlighted some pretty sobering data on this.

00:11:59: He noted that generative AI cost projections of enterprise leaders are tripling in just a matter.

00:12:13: It is terrifying because AI introduces a completely different type of consumption model.

00:12:20: A traditional web server costs roughly the same, whether ten people visit this site or one hundred.

00:12:25: but with generative AI Every single prompt, every single token generated costs compute power.

00:12:32: It's mirrored

00:12:33: Exactly!

00:12:34: Anil Prakash Singh framed this economic reality perfectly.

00:12:37: He warned that in the AI era The winner is not just going to be company with smartest AI model The true winner will be the company with the most economically efficient AI.

00:12:47: Because if your AI inference costs scale linearly, you haven't built a sustainable business model.

00:12:54: You have just build a machine that burns cash faster than more successful.

00:12:57: yet Which means optimization is no longer a back office accounting task.

00:13:02: It's core day one engineering competency.

00:13:05: But here's the catch...you can actually execute this shift-left Finov strategy And you certainly cannot enforce data sovereignty if your underlying infrastructure is a tangled mess of legacy servers and fragmented clouds.

00:13:19: Which naturally transitions us into the third-and final battleground from our LinkedIn insights, Hybrid Cloud & The Battle for Operational Discipline?

00:13:27: Yes!

00:13:28: For years this industry was consumed by cloud vs on premises debate.

00:13:32: but that debate officially over.

00:13:35: Hybrid is the baseline reality in almost every major enterprise today.

00:13:39: The reality is that you're gonna have some highly secure workloads sitting on bare metal in a basement, Some customer-facing apps running in public clouds and IoT devices processing data at the edge.

00:13:50: Yeah

00:13:50: it's everywhere.

00:13:51: And the real gap right now isn't choosing the perfect architecture...the gap Is the execution of managing that chaos.

00:13:57: Jogendra Iatha captured the philosophy Of this transition beautifully in his post, he wrote On-Prem gave us discipline.

00:14:04: Cloud gave us velocity.

00:14:06: Let's expand on that because it gets to the heart of modern engineering struggle.

00:14:10: So think

00:14:10: about physical constraints for On-premises infrastructure.

00:14:14: When an engineering team had to wait six months for a physical server... ...to be ordered, shipped and racked in the data center.

00:14:20: They were incredibly disciplined about how they wrote their code.

00:14:23: Because compute was scarce.

00:14:25: You couldn't just spin up more Exactly!

00:14:27: They optimized everything.

00:14:29: The cloud completely removed that friction.

00:14:31: It gave us velocity to spin-up a thousand servers In one minute with single script Right But in gaining that velocity We completely lost the discipline.

00:14:39: Today's winning engineering organizations are the ones who can master both, bringing.

00:14:50: And because this hybrid reality is so complex, we're seeing the market respond by intensely productizing solutions to help engineers manage the mess.

00:14:59: They are trying to hide the complexity behind unified tools.

00:15:02: like Anil Kumar highlighted how Grafana is emerging as a critical observability layer.

00:15:08: and for

00:15:14: It's so true.

00:15:14: When your applications are fragmented across AWS, Azure Google Cloud and your own private data centers you desperately need a single pane of glass to aggregate all those metrics in logs.

00:15:25: You need that operational visibility just to survive the complexity.

00:15:29: I mean you cannot secure optimize what you can't see

00:15:31: Exactly.

00:15:32: And we're also seeing new tools emerge That bridge the gap between strict security requirements and The need to modernize legacy systems.

00:15:41: Alan Smyotsky posted a fascinating look at the tool called CloudShift.

00:15:45: Oh, right!

00:15:46: It allows for what is called air-gapped code refactoring

00:15:49: which has vital capability of those highly regulated industries we talked about earlier like banking or defense.

00:15:55: Exactly because how does an engineer actually use AI to rewrite thousands lines legacy codes without sending that proprietary code out into public internet?

00:16:04: Right you can't just paste bankcode in chat GPT.

00:16:07: No So, Smyotsky explains that if a highly regulated bank is migrating to Google Cloud they can use tools like CloudShift To localize the AI model weights within their own secure perimeter.

00:16:19: They could used advanced LLMs to refactor code on-premises without proprietary source codes ever hitting an external API.

00:16:27: It's amazing!

00:16:28: it

00:16:29: proves you do not have.

00:16:30: sacrifice security get velocity of modernization.

00:16:34: You know, if we connect all of these pieces the physical sovereignty The automated financial controls and the hybrid complexity A new entirely different problem emerges And it is actually not a technical problem.

00:16:46: Cushop A shared a truly profound thought regarding the integration of AI into these environments.

00:16:52: He pointed out that as AI workflows take over the management of complex hybrid systems, we are actually experiencing a massive leadership gap.

00:16:59: Wait so let's play out scenario here Let say an intelligent AI agent is given authority to automatically scale infrastructure.

00:17:06: Over long weekend The AI predicts a massive traffic spike and decides move a massive workload from private data center in public cloud And in doing so, it completely blows the monthly Phenoms budget.

00:17:19: Who gets fired?

00:17:19: That

00:17:20: is the exact tension Kashyap was highlighting.

00:17:22: who actually owns the decision rights for what that autonomous agent does?

00:17:27: Is It The Architecture Team Who Built The Initial Infrastructure Pathways Right... Is At The Business Unit Who Paid For The AI Tool?

00:17:35: or is it the security team who failed to set boundaries?

00:17:38: That's wild!

00:17:40: When automation starts making financial and architectural decisions that used to sit firmly within human authority, organizational boundaries blur completely.

00:17:49: The technology works flawlessly, but the human accountability is broken.

00:17:53: It forces a total redesign of the corporate operating model.

00:17:56: I mean you can have the most advanced sovereign cloud architecture and the tightest shift left fine ops controls in the world.

00:18:02: But if your leadership doesn't know who was legally and financially allowed to authorize an AI To pull the lever You are going to experience massive internal chaos.

00:18:12: Yeah that is the ultimate bottleneck.

00:18:14: we've built these incredibly powerful automated distributed global systems, but our human organizational charts still look like they were drawn in nineteen ninety five.

00:18:22: Exactly!

00:18:23: So to wrap up our deep dive today I want to leave you the listener with one final provocative thought to mull over.

00:18:31: we started this conversation by talking about the physical reality of The Cloud where the servers actually sit and what legal jurisdiction They fall under right?

00:18:40: Well, Matthews Freed has brought a sobering point About What True Resilience Really Means In This Physical Reality.

00:18:46: Globally, organizations are spending roughly six hundred billion dollars a year on cloud infrastructure.

00:18:53: Engineering teams spend countless hours building these incredibly complex, logical disaster recovery plans.

00:18:59: Ensuring they can failover from one availability zone to another if a server rack goes down?

00:19:04: Right!

00:19:04: The underlying assumption being that the cloud will always just be there, humming along in the background….

00:19:09: But Frida's points out most of those logical disaster-recovery plans assume physical undersea.

00:19:15: fiber optic cables connecting our continents cannot be cut In a world of increasing geopolitical tension, true resilience is not just about picking different software region in the drop-down menu on an AWS dashboard.

00:19:28: True Resilience Is About Physics It's About Geography.

00:19:32: That's A Great Point!

00:19:32: it

00:19:33: goes back to difference between virtual partition and physical building.

00:19:37: you can have nicest luxury apartment with most advanced AI thermostat and strictest budget controls But if the physical foundation of the building is sitting on a fault line, or if the main water pipe under the street gets severed.

00:19:51: all your software optimizations mean nothing.

00:19:53: The cloud is a physical entity and Controlling it means understanding the physical geographical world that relies on

00:20:00: absolutely.

00:20:01: If you enjoyed this episode new episodes drop every two weeks.

00:20:05: Also check out our other editions on ICT and Tech, Digital Products & Services, Artificial Intelligence, Sustainability in Green ICT, DefenseTech and

00:20:13: Health.

00:20:14: Awesome!

00:20:15: Thank you so much for joining us on this deep dive.

00:20:17: we hope you found these insights valuable.

00:20:19: don't forget to subscribe wherever you listen.