Best of LinkedIn: Cloud Insights CW 39/ 40

00:00:00: This deep dive is provided by Thomas Allgeier and Frennis based on the most relevant LinkedIn posts about cloud in calendar weeks thirty nine and forty.

00:00:08: Frennis enables enterprises with market technology and competitive intelligence for portfolio and strategy development.

00:00:15: Welcome to the deep dive.

00:00:16: So if you've been tracking the digital transformation space, you'll know the cloud conversation has really shifted.

00:00:22: It's not just if anymore, is it?

00:00:24: No, definitely not.

00:00:25: It's all about the how.

00:00:26: How do we merge this incredible AI momentum we're seeing with, well, increasingly strict demands for sovereignty and Let's be honest, serious financial discipline.

00:00:35: Exactly.

00:00:36: And our sources for this deep dive, especially looking at calendar weeks, thirty nine and forty, they really highlight this.

00:00:42: There was a lot of buzz coming out of the Google Cloud Summit in Zurich in week thirty nine.

00:00:46: That's

00:00:46: right.

00:00:47: That summit plus ongoing global moves by the hyperscalers reacting to geopolitical stuff and financial pressures.

00:00:53: In week forty, all paints a picture of this tension triangle.

00:00:58: How do you get the power of AI?

00:01:00: satisfies sovereignty and keep costs under control.

00:01:03: It really

00:01:03: feels like the defining challenge for infrastructure right now.

00:01:06: So our mission today is to cut through some of that noise and pull out the key cloud trends we saw reflected in those LinkedIn posts.

00:01:13: We're talking the global race for sovereign cloud, agentic AI actually becoming operational, and this big financial reckoning.

00:01:20: that's driving strategy.

00:01:22: Okay, let's dive into that first theme.

00:01:24: Sovereignty, regulation, compliance.

00:01:27: The geopolitical cloud.

00:01:29: I mean, the core tension seems pretty clear.

00:01:31: Everyone wants to innovate fast.

00:01:32: But there's this growing need for jurisdictional control.

00:01:36: Absolutely.

00:01:36: And a big driver for that, let's be frank, is political distrust.

00:01:40: Anthony Moth made a really strong point about this.

00:01:42: He highlighted how distrust, particularly around the U.S.

00:01:44: Cloud Act and its potential reach, well, it's speeding up the shift towards local providers and specific sovereign initiatives, you know, like Blue and S-III and S in Europe.

00:01:54: Right.

00:01:54: It's not just a tick box exercise, then.

00:01:56: It's about fundamental control over national digital infrastructure.

00:02:00: Exactly.

00:02:01: And that makes defining true sovereignty really critical.

00:02:04: It's not enough to just say, oh, the data center is in Frankfurt.

00:02:08: Paul Seiler laid out a fantastic framework for this.

00:02:11: What did that involve?

00:02:12: He insisted you have to look deeper.

00:02:14: Who actually owns the company running the service?

00:02:16: Where do the profits go?

00:02:18: Critically, whose laws apply?

00:02:20: Are the people managing it local?

00:02:22: And maybe the most important point, who makes the final call if a foreign government comes asking for data?

00:02:29: Ah,

00:02:30: so if that decision ultimately rests in a boardroom overseas...

00:02:34: Then it's not truly sovereign, according to Seiler's framework.

00:02:37: And this is probably why some earlier attempts maybe didn't quite hit the mark.

00:02:40: Sven Treiflinger pointed out they often focus just on the legal or operational layers.

00:02:44: Okay,

00:02:44: so what's the missing piece?

00:02:46: The technical layer.

00:02:47: Layer three is, Treiflinger puts it, the real guarantee has to be baked into the technology itself, you know, regardless of trust or contracts.

00:02:54: And the key technology enabling that layer three.

00:02:57: Uh, it really comes down to confidential computing.

00:02:59: Explain that a bit.

00:03:00: Essentially, it's about encrypting data while it's being processed in the computer's memory.

00:03:05: Not just when it's stored on a disk or moving across the network.

00:03:08: It creates this secure, hardware-enforced space.

00:03:12: So even the cloud provider themselves can't peek inside.

00:03:15: Or a hacker who somehow got rude access.

00:03:18: Theoretically, they can't see the data or the code running inside that trusted environment.

00:03:23: It's using tech to solve a legal and trust problem.

00:03:26: Interesting.

00:03:27: Technology doing the lawyer's job, in a way, and meanwhile the actual lawyers and regulators aren't standing still either.

00:03:33: Walter Zerzavi mentioned Germany's evolving landscape.

00:03:36: Yeah, with things like Dora, the digital operational resilience act now being effective, that really raises the bar for financial services resilience and managing third-party risk, including cloud providers.

00:03:46: So the pressure is mounting on providers like, say, the AWS European Sovereign Cloud to get these truly independent EU setups fully operational.

00:03:55: Definitely.

00:03:56: And we're seeing the big players respond, but maybe in slightly different ways.

00:04:01: Sharif Tautik highlighted this contrast.

00:04:04: Some providers are focusing on data shield solutions or air-gapped environments, basically building digital walls.

00:04:11: Microsoft, on the other hand, seems to be pushing more of an end-to-end story, talking about architecture guardrails set in code, customer control keys, externally managed keys, and even making a formal legal pledge to challenge data requests.

00:04:26: It's a different posture.

00:04:27: That's the useful distinction.

00:04:29: And sometimes analogies help grasp these complex choices.

00:04:32: Varun Gupta offered one comparing sovereign cloud to, what was it, a regulated local restaurant?

00:04:38: Yeah,

00:04:38: exactly.

00:04:39: He said, think of it like that.

00:04:40: You get the benefits of a larger system of compliance, some agility, maybe better cost than doing it all yourself, but it operates strictly under local rules.

00:04:47: It's better than cooking everything at home, which is like on-prem with all its overheads.

00:04:51: Right.

00:04:51: Or hiring a super expensive personal chef, which he likened to a private cloud.

00:04:55: Precisely.

00:04:56: It helps frame that cost benefit, especially if you're in a highly regulated industry where just going public cloud isn't straightforward anymore.

00:05:05: OK, so this push for sovereignty often means modernizing infrastructure.

00:05:10: And what's demanding?

00:05:11: huge modernization budgets right now?

00:05:14: AI.

00:05:15: Let's shift to our second theme, AI and data momentum, specifically, agentic AI going operational.

00:05:23: This was apparently huge at the Zurich summit.

00:05:25: Oh,

00:05:25: it was definitely the buzzword.

00:05:26: Because it felt like a real shift.

00:05:28: AI isn't just about writing an email or making a picture anymore.

00:05:32: It's about taking action.

00:05:33: We should probably clarify what agentic AI means here.

00:05:36: So unlike your standard chatbot that just spits back an answer to your prompt, an agentic AI is designed to be more autonomous.

00:05:43: It can understand a goal, figure out the steps needed, take actions, observe the results, and even correct its course to achieve something complex.

00:05:50: often without structured data.

00:05:52: So it doesn't just answer the question, it actually goes and does the thing.

00:05:55: Exactly.

00:05:56: And Francesca Farah and Laurie B had this demo called superhuman speed of agentic AI.

00:06:01: They showed people who weren't coders using just natural language to build these agents to tackle really quite complex business problems.

00:06:09: Wow.

00:06:10: Not just theoretical problems?

00:06:11: No, real world stuff.

00:06:12: And apparently doing it super quickly, like under an hour in some cases.

00:06:16: That potential.

00:06:17: Well, it changes the game for automation.

00:06:19: And the compelling bit is that it's not just demos, there were actual customer stories.

00:06:23: This stuff is running now.

00:06:25: Like in finance, guesses Koopman shared how picked-at-wealth management is using something called investigator AI.

00:06:31: Right, using it for innovation in risk and compliance.

00:06:34: Think about that, using autonomous agents in such a high stakes, tightly regulated area.

00:06:39: Yeah, that's significant.

00:06:40: And on the consumer side.

00:06:42: Nestle, William Hart and Tomas Casab talked about embedding AI and machine learning deep into their marketing.

00:06:48: Moving beyond just basic ad targeting to using data insights for essentially a real competitive edge.

00:06:55: Okay, marketing compliance.

00:06:57: But maybe the most impactful example came from healthcare.

00:07:00: Dominic Steiner from Kinder Spital Zurich.

00:07:02: Absolutely incredible use case.

00:07:03: They're using generative AI on Vertex AI and Gemini, Google's platforms, to scan millions of scientific papers.

00:07:12: Literally millions.

00:07:13: That's you, what?

00:07:13: to pull out highly specific personalized insights in minutes.

00:07:17: The goal is accelerating how they develop personalized cancer therapies for children.

00:07:22: That's transformation with a capital T, immediate life changing potential.

00:07:26: Wow.

00:07:27: And these kinds of applications really underline the point Fundatilif Schermacher made, especially for leaders in financial services.

00:07:34: The pressure to adopt is immense, but the opportunity is arguably even greater.

00:07:40: What kind of opportunities?

00:07:41: Things like true hyper personalization for customers, serious automation of those complex back office or middle office tasks, and using predictive analytics to really enhance risk modeling.

00:07:51: It's becoming table stakes.

00:07:53: It feels like the perception is shifting fast.

00:07:55: Giza Malsock put it quite simply didn't she?

00:07:57: Something like, the future of AI isn't in the future.

00:07:59: Yeah, that captured it perfectly.

00:08:01: It's already here, running quietly under the surface in many operations, making decisions, maybe even writing code, driving research.

00:08:09: It's a new baseline you have to plan your infrastructure around.

00:08:12: Right, but that powerful AI infrastructure needs a place to live.

00:08:15: And that inevitably brings us back to the budget, which is our third theme.

00:08:19: infrastructure, modernization, and fin-ups, basically.

00:08:23: The cloud cost reckoning.

00:08:24: Uh-huh.

00:08:25: We've definitely hit a point, some are calling the cloud slow down.

00:08:28: Tom Gamali and James Gottry were talking about this, that old mantra of cloud first, no matter the cost, it's pretty much dead.

00:08:35: So what are people doing instead?

00:08:37: We're seeing actual workload repatriation.

00:08:40: Significant amounts, one figure suggested around twenty-one percent of enterprise workloads are moving back, either fully on-prem or more often to some form of private cloud.

00:08:49: Why the reversal?

00:08:50: Costs.

00:08:51: Specifically, the difficulty in governing public cloud costs.

00:08:54: Turns out it's really hard.

00:08:56: Nearly half of companies, when surveyed, admit they're wasting over a quarter of their cloud budget.

00:09:01: Wow, wasting twenty-five percent.

00:09:03: That's not just pocket change.

00:09:04: That's a major strategic issue.

00:09:07: So what's the fix?

00:09:08: You can't just abandon modernization, especially with agentic AI demanding resources.

00:09:13: I know, you can't retreat.

00:09:15: Shuresh Kakarla really stressed this.

00:09:17: The answer isn't stopping.

00:09:18: It's more disciplined modernization, simplifying and standardizing the infrastructure specifically for these new AI workloads.

00:09:26: How do you do that standardization?

00:09:28: A huge focus is on infrastructure as code, IAC.

00:09:31: Around forty percent of companies are prioritizing this.

00:09:34: Using IEC helps ensure consistency, better security, governance, and crucially it helps manage costs across potentially hybrid environments.

00:09:43: Any examples of that working?

00:09:44: Yeah, Skechers was cited.

00:09:46: They integrated service now with Terraform, a popular IAC tool, and managed to cut their provisioning time for infrastructure from weeks down to just hours.

00:09:54: Think about the cost savings and agility gain there.

00:09:55: Okay,

00:09:56: so better processes, better tools.

00:09:58: But if the core problem is complex spending, you need better measurement too, right?

00:10:01: That's where FinOps comes in.

00:10:03: Exactly.

00:10:04: But not just FinOps as a siloed cost cutting team.

00:10:07: Nicholas Fundrini made a great point about integrating phenops thinking directly with traditional project management like PMI standards.

00:10:13: How would that work?

00:10:13: It means bringing your cloud spend metrics right alongside your project delivery metrics like Earn Value, EVM, or cost performance index, CPI.

00:10:23: So you're not just tracking if you spent the cloud budget, but proving that the span is actually delivering measurable business value according to project goals.

00:10:31: That elevates fine-ups significantly.

00:10:33: Okay, that's the strategic view.

00:10:35: But let's talk brass tax, the actual costs.

00:10:38: Farouk Mobyan shared a comparison of basic ISS costs.

00:10:42: What was it for VCPUs, sixteen GB RAM and Frankfurt?

00:10:46: Yeah, on demand pricing.

00:10:47: And the numbers were... Well, eye-opening.

00:10:50: How so?

00:10:50: This is where you really see the different strategies playing out.

00:10:53: Oracle Cloud Infrastructure, OCI, came in remarkably low at about fifty-five dollars a month for that configuration.

00:10:58: Fifty-five dollars, okay, and the others.

00:11:00: The big

00:11:00: three AWS, Azure, Google Cloud, they were all clustered way higher.

00:11:04: Around a hundred and fifty to a hundred and fifty-eight dollars per month for essentially the same basic compute resource.

00:11:08: Wait nearly three times the cost for the same spec.

00:11:11: Roughly, yeah.

00:11:13: That kind of massive difference.

00:11:15: It forces you to ask really hard questions about your long-term strategy, and you know, what exactly you're paying that premium for with the bigger providers.

00:11:23: Hmm.

00:11:23: Suddenly that twenty-one percent repatriation figure makes a lot more sense, doesn't it?

00:11:27: If optimizing cost is paramount, maybe hybrid or multi-cloud isn't a compromise, it's just smart finance.

00:11:32: Precisely.

00:11:33: And we saw a great example of really mastering that hybrid complexity in a case study Liz Rice shared about... cloud era.

00:11:40: They were using advanced tools on AWS like EKS for Kubernetes, Carpenter for scaling nodes, Bottle Rocket OS to handle very dynamic workloads.

00:11:48: And the results?

00:11:49: Pretty stunning.

00:11:50: They cut their infrastructure costs by forty percent, they reduce the attack surface for developers by sixty percent, and they improve software build times by fifty percent.

00:11:58: Okay, so significant wins across costs, security and speed.

00:12:01: Exactly.

00:12:02: It demonstrates that if you approach hybrid with discipline and the right tools, it can be an incredibly powerful optimization strategy, not just a necessary evil.

00:12:13: So if we try to sort of pull it all together from these last couple of weeks, the cloud industry is really grappling with these three powerful, sometimes competing forces.

00:12:22: You've got the sheer speed and potential of agentic AI demanding new, flexible architectures.

00:12:27: Then you have these absolute non-negotiable requirements for digital sovereignty and real operational control.

00:12:35: And underpinning it all is this intense financial pressure for efficiency, for proving value.

00:12:40: companies that thrive are going to be the ones who can successfully balance all three of those things.

00:12:44: That really is the core challenge.

00:12:46: You know, the vibe at the Zurich Summit was apparently very much, why not?

00:12:49: Why not push the boundaries with AI?

00:12:51: Yeah.

00:12:51: But as we embrace all this amazing tech, we also need to remember the lessons of resilience.

00:12:57: YIO, ZA posted recently reflecting on an airport IT adage, reminding us that even when AI is writing code.

00:13:05: Still human resilience and adaptability that often saved the day.

00:13:10: So maybe the final thought for you listening is this.

00:13:12: As you build out your strategies for digital sovereignty and cost efficiency, how do you make sure those controls don't just constrain things, but actually empower that human ingenuity to deal with the unexpected?

00:13:24: How do you enable resilience?

00:13:26: If you enjoyed this deep dive, new episodes drop every two weeks.

00:13:29: Also check out our other editions on ICT and tech, digital products and services, artificial intelligence, sustainability and green ICT, defense tech and health.

00:13:38: Thank you for joining us and make sure to subscribe so you don't miss our next deep dive.