Best of LinkedIn: Sustainability & Green ICT CW 44/ 45

00:00:00: This episode is provided by Thomas Allgaier and Freeness, based on the most relevant LinkedIn post about sustainability and green ICT in CW-IV and IV-V.

00:00:10: Freeness supports ICT enterprises with marketing competitive intelligence, decoding green software developments, benchmarking emerging standards, tracking regulatory shifts, and analyzing competitor strategies.

00:00:23: Welcome back to the deep dive.

00:00:24: So you've sent us a massive stack of curated LinkedIn content from the last couple of weeks.

00:00:29: We have, and the message is pretty loud and clear.

00:00:32: Green ICT is no longer this fringe concern, right?

00:00:35: It's really starting to dictate fundamental business strategy.

00:00:38: Exactly.

00:00:39: What's fascinating here and what we're seeing in the source material is this critical shift away from just generic sustainability claims.

00:00:45: The greenwashing.

00:00:46: Right.

00:00:47: It's moving to concrete delivery.

00:00:48: We're seeing a focus on hard efficiency, total transparency, and demanding measurable real world impact.

00:00:55: I mean, the talking phase is over.

00:00:56: So our mission today is to unpack those critical insights for you.

00:00:59: We've clustered the top sustainability and green ICT trends into four big areas.

00:01:04: First

00:01:04: up, the big one.

00:01:06: the inescapable AI energy collision.

00:01:08: Then we'll get into the infrastructure and the physical limits technology is running into.

00:01:13: After that, green software is a potential solution.

00:01:16: And finally, we'll look at the essential policy and strategy frameworks that are starting to define the future.

00:01:22: Okay,

00:01:22: so let's jump in.

00:01:23: We really have to start with the biggest resource drain of the moment.

00:01:26: The explosive growth of artificial intelligence.

00:01:29: Yeah.

00:01:30: Dr.

00:01:30: Shahid Massoud described the situation perfectly.

00:01:32: He called it a collision.

00:01:33: A collision, yeah.

00:01:35: AI's rapid ascent is just, you know, running headfirst into these hard resource limits.

00:01:40: And

00:01:40: it's turning energy and I think far more urgently than most people realize, water availability into a core constraint for innovation.

00:01:48: It's a fundamental conflict between digital scaling and physical capacity.

00:01:52: And that water challenge is the one that's truly startling.

00:01:55: It is.

00:01:55: Data centers rely so heavily on water for cooling, often pulling from local supplies.

00:02:00: Lexicido and Alexander Sachikov both emphasize this.

00:02:04: What

00:02:04: are the numbers like?

00:02:05: Well, a large data center can consume one point seven million gallons of water a day.

00:02:11: Each day.

00:02:11: Wow.

00:02:12: Yeah.

00:02:13: It's an astonishing amount.

00:02:14: And it means every single AI query you run.

00:02:17: has a tangible water footprint.

00:02:20: So that immediately raises these massive questions about ethical scaling, doesn't it?

00:02:24: Yeah.

00:02:24: If an AI model needs the water of a small town.

00:02:27: Where's

00:02:28: the limit?

00:02:29: Exactly.

00:02:30: This isn't just an engineering problem anymore.

00:02:32: It's a political and social one.

00:02:33: It needs innovative countermeasures.

00:02:35: And that's why this idea of circularity is so urgent.

00:02:38: Absolutely.

00:02:40: Prakash Govindan highlighted work from a company called Gradient.

00:02:43: They're supporting new data centers to recover up to ninety nine percent of their process water.

00:02:48: Ninety nine percent.

00:02:49: That kind of maximum efficiency, it really has to become the standard.

00:02:52: Otherwise, AI development is going to create some serious ecological problems.

00:02:56: And while that kind of innovation is essential, the sheer scale of the problem just demands total transparency.

00:03:01: Which is where we see a crucial governance failure emerging.

00:03:04: It's the whole controversy that Chris Adams and Holly Alpine raised.

00:03:08: Right, around Microsoft.

00:03:09: Yeah.

00:03:10: They're urging investors to vote for greater transparency.

00:03:13: The argument is really direct.

00:03:15: How can a company claim climate leadership?

00:03:17: While

00:03:17: using that exact same AI technology to accelerate fossil fuel production.

00:03:22: Precisely.

00:03:23: That contradiction introduces these huge undisclosed financial and ethical risks for shareholders.

00:03:30: It just doesn't add up.

00:03:31: And that really highlights a fundamental misalignment in ESG reporting today, right?

00:03:35: Yeah.

00:03:36: If the metrics let you expand carbon production over here while claiming you're optimizing sustainability over there.

00:03:42: Then the framework is broken.

00:03:43: It's fundamentally broken.

00:03:45: And Joanna Myler reinforced this point brilliantly.

00:03:47: She said, executives have to prioritize measurable outcomes.

00:03:50: An AI pilot that doesn't scale or doesn't deliver real value, like cash flow or risk reduction.

00:03:56: It's not just a waste of money.

00:03:57: It's environmentally unsustainable.

00:03:58: Every watt burned for a failed pilot is, you know, a pointless emission.

00:04:03: So the conversation has to mature.

00:04:05: We need to distinguish, as Phil Nordberg pointed out, between responsible AI, which is about ethics, bias, people.

00:04:12: And sustainable AI, which is laser focused on the planet, on energy, and the carbon footprint.

00:04:19: You need both to build systems that are actually viable long term.

00:04:23: Right.

00:04:23: And all that demand for resources, it leads us straight into the next theme.

00:04:27: Infrastructure.

00:04:29: Because the scale of this demand, from energy to water, isn't just challenging supply chains.

00:04:34: It's physically breaking the infrastructure that was designed just a few years ago.

00:04:38: That's the core realization here.

00:04:40: Karen van der Zandern shared insights from Guy Massey showing just how much the power density is exploding inside data centers.

00:04:46: We're

00:04:46: moving so quickly from racks that pull maybe ten kilowatts.

00:04:50: To needing a hundred kilowatts.

00:04:51: Or even more for these modern AI compute loads.

00:04:55: And ten times jump and the immediate implication of that is just massive technical debt and obsolescence.

00:05:00: So an air-cooled data center built just five years ago.

00:05:03: It's obsolete.

00:05:04: It's basically obsolete for handling next-gen AO workloads.

00:05:07: You just can't cool it efficiently anymore.

00:05:08: Which makes the shift to liquid cooling mandatory.

00:05:11: Not just a nice to have.

00:05:13: We're seeing all the hyperscalers, Meta, Google, Microsoft, they're all embracing liquid first designs.

00:05:19: For sure.

00:05:20: Ali Forsythe even noted a company called Alloy Enterprises, Inc.

00:05:24: and their innovation with something called stack-sorging technology.

00:05:28: And what's that?

00:05:28: It's basically about creating these ultra-efficient thermal management parts, like next-gen heat sinks, to manage those extreme temperatures and help centers scale sustainably.

00:05:40: But

00:05:40: cooling is just one piece of a much bigger strategic puzzle.

00:05:44: It's the only way forward, but you're right.

00:05:46: PSLE outlined ten key elements for sustainable data centers, and it all points to a fundamental paradigm shift.

00:05:53: Which is what?

00:05:53: That data centers have to be planned as integrated grid assets.

00:05:57: They're not just passive loads anymore.

00:05:58: That means integrating twenty-fourty-seven carbon-free energy and planning for heat reuse from day one.

00:06:04: That

00:06:04: grid integration is already driving where companies are choosing to build.

00:06:08: Oh, absolutely.

00:06:08: We're seeing these excellent examples of strategic greensighting.

00:06:12: Charles Henderson and Mark Wilson highlighted Scotland.

00:06:15: They're aiming to be a green AI hub.

00:06:17: Exactly.

00:06:17: By linking AI growth directly to their clean energy goals, and it's attracting huge investment.

00:06:23: like CoreWeave's one point five billion pound commitment, they're using their clean energy as a competitive advantage.

00:06:30: And this isn't just for the hyperscalers.

00:06:32: Not at all.

00:06:33: Jonathan Pepper gave this fantastic local example from Stoke-on-Trent.

00:06:37: A historic pottery site is being converted into the UK's first green AI data center, Josiah One.

00:06:44: And it has waste heat capture and reuse design.

00:06:46: right into its

00:06:47: core.

00:06:47: That's technology rooted in local industrial circularity.

00:06:51: I love that.

00:06:52: And this strategic connection to the grid, it's also about flexibility.

00:06:56: and Curry pointed out, the UK's commitment to fifty-five gigawatts of flexibility by twenty-thirty.

00:07:01: Right, which means data centers become grid supporters.

00:07:03: They can shift non-urgent tasks, like training a huge AI model, to times when grid demand is lower.

00:07:09: They actively help balance the load.

00:07:11: We can't forget the carbon footprint of the building itself, either.

00:07:14: The embodied carbon.

00:07:15: A huge point.

00:07:16: Lexicido noted the growing use of mass timber in projects like EcoDataCenter and also Microsoft's new Hyperscale Centers.

00:07:22: So

00:07:23: using timber.

00:07:24: which actually stores CO-II.

00:07:26: It dramatically reduces the structural emissions compared to relying on traditional carbon-intensive concrete and steel.

00:07:32: Okay, so let's shift from the concrete limits of hardware to the operational side of things.

00:07:38: To the

00:07:38: code itself, green software.

00:07:40: As Heidi Gonzalez-Doria noted, software is no longer just about functionality.

00:07:45: It's now a direct operational lever for cutting energy consumption and, at the same time, improving performance.

00:07:52: This whole discipline is about tackling the massive inefficiency that's just inherent in traditional computing.

00:07:58: Shabab Kuhi detailed the core principles.

00:08:01: You've got operational efficiency, so maximizing how you use resources.

00:08:06: Then card and awareness, which is about shifting workloads to run when and where the grid is cleanest.

00:08:10: Plus

00:08:11: hardware efficiency and code efficiency.

00:08:12: Right.

00:08:13: Extending equipment life and just writing better algorithms.

00:08:16: It's about being lean and carbon aware at every single layer of the stack.

00:08:21: implications for cloud architecture design.

00:08:24: Francesca C talked about this.

00:08:25: Green IT principles have to be architected in from the start.

00:08:29: Which means strategically choosing your cloud regions not just based on latency but on the carbon intensity of their local grid.

00:08:35: And adopting designs like serverless or event-driven architectures that scale down to zero when they're idle so they consume almost no power.

00:08:45: This attention to detail can deliver some real measurable wins.

00:08:49: For sure.

00:08:50: RLO and shared a great example.

00:08:52: SAP clients on Azure can now consolidate multiple H&N instances onto a single virtual machine,

00:08:58: which simplifies backup, improves resource use,

00:09:01: and how uncrucially cuts both their cloud spend and their carbon emissions.

00:09:06: That's the true win-win that green software is designed to provide.

00:09:10: And we're finally seeing the measurement discipline start to mature.

00:09:12: Because if you can't measure it, you can't optimize it.

00:09:15: Exactly.

00:09:16: Tommy Tlazma, for instance, launched WaterLyze, which is a practical tool for measuring a website's QO and energy consumption.

00:09:22: It makes that invisible carbon cost instantly visible.

00:09:25: And Martin Michielic noted that the recent EcoCompute conference showcased tools like EcoCI.

00:09:31: Which are designed to be integrated directly into your CI CD pipelines.

00:09:34: Think of it like a quality gate.

00:09:36: So code that's too carbon-intensive gets flagged before it even gets deployed.

00:09:40: Right.

00:09:41: It embeds sustainability as a continuous metric for deaf teams, just like security or latency.

00:09:46: Okay, so all these technical capabilities, they're pretty useless without supportive frameworks.

00:09:52: Which brings us to our final theme, how governments and expert groups are trying to publish guidance and define a common sustainable digital path.

00:10:01: That's the theory at least.

00:10:03: But the reality seems to be more about regulatory fragmentation and tension.

00:10:08: Yeah, Vincent Barrow highlighted a bitcom report that was urging Germany to simplify its planning and approval procedures for data centers.

00:10:15: And align them with broader European standards.

00:10:18: The warning they issued was pretty striking.

00:10:20: They said without legislative change to streamline these processes, Germany risks becoming a data colony.

00:10:27: A

00:10:27: data colony.

00:10:28: Because the investment appetite for advanced sustainable infrastructure is being held back by impractical local regulations.

00:10:34: It's a race against time to standardize.

00:10:36: And that's why these policy frameworks are so essential for creating some predictability.

00:10:39: Absolutely.

00:10:41: Boris Kamazachikov shared updated, sustainable AI policy principles that explicitly emphasize collaboration between governments and industry.

00:10:49: So not just regulation.

00:10:50: Not just regulation.

00:10:51: It's focused on accelerating clean energy infrastructure and mandating AI transparency.

00:10:57: They recognize that forcing sustainability requires shared accountability.

00:11:01: This all rolls up into a really high level strategic takeaway from Fernanda Tori.

00:11:06: Which is that the convergence of AI and sustainability doesn't just mitigate risk.

00:11:11: It creates a strategic moat.

00:11:13: A strategic mode.

00:11:14: I like that.

00:11:14: It's the key distinction for executives.

00:11:17: The real hard to copy advantage isn't in using some replicable AI tool.

00:11:22: It's in the unique locally rooted ecosystems and specialized data value chains that a company creates.

00:11:27: Exactly.

00:11:27: If you can build infrastructure that is fundamentally linked to a flexible, clean, local energy grid, you create an advantage no competitor operating on older, less sustainable infrastructure can possibly match.

00:11:39: It requires executives to realize that AI and climate change aren't these parallel disruptions.

00:11:45: They are converging forces that will define competitive advantage for the next decade.

00:11:50: So if we synthesize all the sources we've covered, I think the core takeaway is this massive, tangible shift towards accountability and integrated sustainability.

00:11:59: We're moving past the abstract net zero targets.

00:12:02: We are.

00:12:02: We're viewing all ICT assets from the smallest AI model to the highest density data center as an intrinsic, accountable part of the overall energy system, one that has to contribute to decarbonization.

00:12:15: If you enjoyed this deep dive, new episodes drop every two weeks.

00:12:18: Also check out our other editions on cloud, digital products and services, artificial intelligence and ICT and tech insights.

00:12:25: health tech, defense tech.

00:12:26: Thanks for tuning in and subscribing.

00:12:28: We'll catch you on the next deep dive.