Best of LinkedIn: Cloud Insights & Sovereignty CW 29/ 30

00:00:00: Provided by Thomas Allguyer and Frennus, based on the most relevant LinkedIn posts about cloud insights in sovereignty.

00:00:06: In calendar weeks twenty nine and thirty.

00:00:09: Frenness 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 in description.

00:00:21: Yeah, and just to set some expectations before we jump in.

00:00:24: Our mission for this deep dive is to unpack the top digital products all across LinkedIn right now.

00:00:32: Exactly, and today we're splitting our focus along two really critical axes for tech.

00:00:37: professionals basically were looking at where your workloads actually run.

00:00:40: And then who is actually controlling the spend?

00:00:42: Right

00:00:43: which are massive questions right Now.

00:00:44: so let's start with that first theme Which is sovereign cloud.

00:00:48: you know sovereignty Is really shifting from just being this theoretical high-level talking point to actual physical deployment.

00:00:55: Yeah it's not Just policy debates in Brussels anymore right?

00:00:58: It's tangible infrastructure.

00:01:00: It really is.

00:01:00: I mean, Eidt-Vomberg shared a great example of this shift.

00:01:03: AWS is currently building its European sovereign cloud and the data centers are literally the size of IKEA stores.

00:01:11: Just massive concrete warehouses?

00:01:13: Huge!

00:01:14: But the concrete isn't even the main point it's the operational isolation.

00:01:18: they are staffing these facilities exclusively with European residents and citizens to meet those strict jurisdictional requirements.

00:01:26: Oh wow that is huge investment in localization.

00:01:29: You're basically quarantining the operational control.

00:01:32: Exactly!

00:01:33: And, you know we are seeing that same appetite for physical resilience on the client side too.

00:01:38: Thomas Saueraseig pointed out a major move by Airbus recently.

00:01:40: Yeah

00:01:41: with SAP right?

00:01:42: They are adopting Rise With SAP which is moving their massive ERP systems to cloud but they do it through an end-to-end sovereign deployment model

00:01:51: Which makes total sense for aerospace.

00:01:53: I mean they want the agility of a cloud based ERP But absolutely cannot compromise data boundaries.

00:01:59: Right, it's a core business strategy now.

00:02:02: It's actually becoming a competitive differentiator like Dr.

00:02:06: Ferry Appelhausen brought up Deutsche Telekom T-Cloud campaign.

00:02:10: Oh how they're framing around trust

00:02:12: Exactly.

00:02:13: They're pitching digital sovereignty entirely as a trust and business decision for their seven thousand customers.

00:02:20: It's no longer just.

00:02:21: you know buy this so you don't get fine it's Buy This because we guarantee total control over your data footprint.

00:02:27: Yeah, and some providers are leaning into that even harder.

00:02:30: Damien Luke has noted that scaleway.

00:02:32: Just publish a manifesto where they are actively demanding European jurisdiction across the board in

00:02:38: price and transparency too right?

00:02:40: yeah there really drawing on line in the sand against traditional, more opaque hyperscaler models.

00:02:46: Okay but let me push back for a second here.

00:02:48: because you build data centers this size of furniture warehouse You restrict staff to very specific citizens and publish these aggressive manifestos.

00:02:56: It kinda sounds like we're just building massive walled fortresses into cloud.

00:03:01: So bringing that CSIS research.

00:03:03: I have to ask critical question Do these EU sovereignty policies, these sovereign walls actually undermine innovation without genuinely improving security?

00:03:14: Well that right.

00:03:15: there is the defining debate of the Sovereign Cloud movement.

00:03:18: I mean building a fortress sounds great until you realize you've locked yourself inside and can't trade with the outside world.

00:03:24: Right!

00:03:24: You just cannot operate an AI ecosystem in a vacuum

00:03:27: Exactly.

00:03:28: but if look at the forestry research they address this dynamic beautifully.

00:03:32: They argue that true sovereignty isn't really about achieving total independence, it's actually about managing dependence.

00:03:39: Oh!

00:03:39: That is an interesting way to frame it.

00:03:41: so you are managing the risk rather than isolating entirely.

00:03:44: Right

00:03:44: You want know exactly who holds master keys and have legal right change locks but even then proving incredibly difficult.

00:03:52: I mean European Commission recently ran a sovereign cloud tender process Ironically revealed that literally no provider has achieved full sovereignty yet.

00:04:01: Nobody has the perfect fortress

00:04:03: nobody.

00:04:04: The dependencies are just too deep.

00:04:05: Yeah, just look at France.

00:04:07: they've been struggling in this highly publicized effort to replace Palantir's intelligence software with a homegrown alternative.

00:04:15: It just proves how incredibly sticky foreign software dependency really is.

00:04:19: Oh absolutely ripping out intelligent software That's woven into the daily habits of thousands of analysts.

00:04:25: it's like Well, it's like performing a heart transplant on someone running a marathon.

00:04:30: Right!

00:04:31: Which perfectly validates the commentary coming out of the UK.

00:04:35: They're arguing that before we dump billions into sovereign infrastructure We desperately need a clearer vocabulary and much better workload classification.

00:04:43: Yeah you can't just slap a sovereign sticker on everything.

00:04:46: You have to classify what actually demands extreme isolation

00:04:49: Exactly, and you know managing that classification actually transitions us perfectly to the second big shift in the market.

00:04:55: Right?

00:04:56: The cost side of things?

00:04:57: Yeah

00:04:57: If sovereign cloud is about controlling the physical location of your infrastructure AI Finops is about Controlling the wildly unpredictable costs running on it!

00:05:06: The AI Cost Curve Is Completely Rewriting Cloud Economics.

00:05:10: right now

00:05:11: It really is.

00:05:13: Shreedhar Ramaswamy shared a staggering snowflake stat That blew my mind.

00:05:18: In just two years, the share of Phenops teams managing AI spend jumped from thirty one percent to ninety eight percent.

00:05:26: That is a massive leap in twenty four months.

00:05:29: but you have look at mechanism behind it.

00:05:32: AI token spend is no longer just a standard line item.

00:05:35: No, it's totally variable

00:05:37: right?

00:05:37: It has mutated into a full-blown engineering control problem.

00:05:41: You need a formal governance operating model because reports are consistently showing that AI token costs are running way above initial projections.

00:05:50: And since generative AI models are kind of commoditizing the baseline intelligence, The real differentiation between vendors is narrowing.

00:05:57: So execution is actually the real competitive advantage now.

00:06:01: Oh

00:06:01: one hundred percent?

00:06:02: Yeah!

00:06:02: How efficiently you govern that spend is everything.

00:06:05: By the way if you are navigating these exact AI governance challenges in your own stack right Now make sure you hit subscribe so don't miss our future deep dives.

00:06:12: We track this stuff constantly.

00:06:14: Definitely You're gonna need those strategies.

00:06:16: Yeah, because honestly traditional cloud spend was like a predictable utility bill.

00:06:20: But AI token spend it's giving your engineering team an open corporate card at a casino.

00:06:26: That is the perfect analogy.

00:06:28: every prompt Is a pull on the slot machine and The market is desperately trying to respond to that.

00:06:33: yeah Neiman Jane brought up a fascinating new tool this phenops copilot

00:06:38: though one that generates cost estimates.

00:06:40: yes It literally turns a developer's conversational description of workload into an instant cost estimate.

00:06:47: So you intercept the spend before it even happens!

00:06:50: That is brilliant, so your basically pushing that find-ups discipline left right in to CICD pipelines?

00:06:55: Exactly – You force costs.

00:06:57: checks happen automatically before code ever goes live.

00:07:00: A small engineering team can actually govern massive compute budgets autonomously Right

00:07:05: But there is fierce debate here.

00:07:06: Are these rising prices purely because AI is inherently expensive?

00:07:10: Or, this is a big question.

00:07:12: Is outdated cloud architecture and economic technical debt actually the real constraint on financial transparency?

00:07:19: Oh wow!

00:07:20: Blaming the AI for problem that started ten years ago.

00:07:23: Exactly If you deploy cutting-edge AI On top of highly inefficient legacy infrastructure Your spend will spiral no matter how good your Fedops copilot is.

00:07:32: It's like putting Ferrari engine in a rusty minivan.

00:07:35: Yeah

00:07:36: And this cost pressure is causing huge ripples among the hyperscalars themselves.

00:07:39: as they backdrop to all of these.

00:07:41: Look at Google Cloud's Q-to growth outpacing Microsoft and AWS, or IBM running aggressive limited time promos with seventy percent savings!

00:07:50: Everyone knows organizations are desperate for better economics...

00:07:54: Which honestly creates a massive headache for people actually trying build that stuff….

00:07:59: You have a mandate data in a localized sovereign fortress, but you also need the absolute cheapest public cloud compute for your AI models.

00:08:09: Right those are conflicting demands and

00:08:10: trying to optimize both of those pushes organizations right into

00:08:20: where cloud theory goes to die and reality takes over.

00:08:23: Yeah, Dave McCarthy used that exact phrasing when he was talking about IBM Fusion targeting platform sprawl.

00:08:29: in this messy middle

00:08:30: It's a huge problem.

00:08:31: You're also seeing tools like HPE's Morpheus software trying to offer unified hybrid management.

00:08:37: just so IT teams don't lose their minds managing fourteen different dashboards.

00:08:42: Right, and for highly regulated sectors one insight noted that multi-cloud banking only succeeds through really deliberate workload and data placement decisions.

00:08:52: you can't just spin up databases randomly anymore.

00:08:54: No the calculus is way too complex now.

00:08:56: And hovering over all of that placement logic Is The Security.

00:09:00: Angle Mandits in Mtrend's report dropped a pretty chilling insight about this messy middle.

00:09:05: Threat actors are moving at machine speed now,

00:09:08: which means human security analysts physically cannot type fast enough to stop them

00:09:13: exactly.

00:09:13: so it requires completely new agentic Security models and strict micro segmentation.

00:09:19: just To stop that lateral movement

00:09:20: you have to seal off the bulkheads.

00:09:22: if an attacker breaches one server In that hybrid environment they can't be allowed to Just walk over to The sovereign database next door.

00:09:28: right but wait Managing all that complex placement makes me question the whole AI deployment life cycle.

00:09:35: We keep hearing about cloud first, but one poster noted that public Cloud is actually just for early AI experimentation while on-prem wins at scale.

00:09:45: Why The Sudden Reversal?

00:09:46: It really comes down to data gravity and those token economics we talked about.

00:09:50: Public Cloud is great to start because you rent GPUs, run tests shut them down.

00:09:56: Right very flexible.

00:09:57: but when that model scales and needs to ingest petabytes of data moving that data back-and forth to a public cloud will absolutely bankrupt you with egress fees.

00:10:06: oh so bringing the compute physically closer to the data like on-prem or in a regional sovereign data center becomes much cheaper

00:10:14: exactly.

00:10:15: But that means your software architecture has to be fluid.

00:10:17: You can't rewrite millions of lines of code When you move from Azure to his sovereign box in Frankfurt.

00:10:22: Which is why tools like Radius and Dapp are getting so much traction, right?

00:10:25: Right.

00:10:25: They abstract the underlying cloud APIs.

00:10:28: So developers create these portable applications that span both sovereign and public clouds seamlessly.

00:10:34: You

00:10:34: can just shift the workload wherever the economics or laws dictate.

00:10:38: Yes But to govern a fluid system like that you need total visibility.

00:10:42: That's why Gartner's Observability Magic Quadrant Is heavily shaping these hybrid decisions even for government buyers in Australia.

00:10:51: If you can't observe the data across the messy middle, You can secure it or

00:10:55: govern.

00:10:55: The cost exactly?

00:10:57: well, you know Managing that messy middle securing against machine speed attacks governing all those AI budgets It's not just a software routing problem is completely rewiring the talent stack and careers.

00:11:08: Oh, the human element is shifting so fast.

00:11:10: AI has now deeply woven into every single tier of cloud infrastructure careers from technical to management.

00:11:17: But despite all that AI hype, Ali Khaja Hussaini shared a really grounding reminder.

00:11:21: Mastering core virtualization concepts is still the absolute foundation for AWS skills.

00:11:27: Yeah because at the end of day The cloud is just someone else's computer.

00:11:31: AI can write a script but if an architect doesn't understand underlying plumbing They won't be able to troubleshoot when the AI inevitably hallucinates.

00:11:39: bad template

00:11:40: Right...but it sounds like Just being a technical whiz in the basement isn't enough anymore.

00:11:46: You actually need to speak the language of the boardroom and interview process now?

00:11:50: You really do!

00:11:51: That's why we're seeing so much rising traction for the STAR method...in interviews.

00:11:55: Situation, task action result

00:11:58: Exactly.

00:11:59: Hiring managers are forcing.

00:12:00: cloud engineers connect their technical work directly to measurable business outcomes.

00:12:05: So they don't just want hear I deploy dapper.

00:12:07: They wanna hear I deployed Dapper to meet sovereign compliance and reduced our fine-op spend by twenty percent.

00:12:14: Right, cloud architects are basically being forced to evolve from just managing tech stacks... ...to leading what the industry is now calling Intelligent Knowledge Ecosystems.

00:12:24: Intelligent knowledge ecosystems?

00:12:26: That's a heavy title for one person.

00:12:31: They're orchestrating autonomous AI agents, human engineers financial guardrails and sovereign laws all at the same time.

00:12:38: It's incredible!

00:12:39: And honestly that leads to a pretty provocative final thought to ponder as we wrap up.

00:12:43: Okay let's hear it.

00:12:44: As our architecture shifts through these intelligent knowledge ecosystems and security moves into agentic models The biggest bottleneck in your digital transformation won't be cloud hardware or software.

00:12:56: The real bottleneck is going to be finding the human talent capable of orchestrating machines that are increasingly capable managing

00:13:16: themselves.