Best of LinkedIn: Sustainability & Green ICT CW 42/ 43

00:00:00: This episode is provided by Thomas Allgeier and Freanus, based on the most relevant LinkedIn posts about sustainability and green ICT in CW-四ty-two and forty-three.

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

00:00:22: Welcome to the deep dive.

00:00:25: If you're looking to get up to speed on green ICT trends, you know, without getting lost in all the noise, you're definitely in the right place.

00:00:32: Yeah, we've sifted through LinkedIn for the past couple of weeks, CW-Forty-Two and Forty-Three, really focusing on that tricky intersection of AI infrastructure and green software.

00:00:42: Our goal today is simple.

00:00:45: Cut through that noise, give you the key insights, and honestly, the big story we pulled out is pretty clear.

00:00:49: It

00:00:49: really is.

00:00:50: The whole industry seems to be shifting.

00:00:51: It's less about whether to be sustainable and much more about how to actually do it, execution.

00:00:56: Exactly.

00:00:56: But here's the kicker.

00:00:57: The scale of innovation, such with AI, it's huge.

00:01:00: And the cost, you know, energy resources, it's becoming almost impossible to ignore.

00:01:05: It feels like a real turning point.

00:01:07: Sustainability isn't just a report anymore.

00:01:09: It's becoming a core, technical, even financial constraint.

00:01:13: It dictates who can actually scale.

00:01:16: Which brings us straight into the first big theme that was all over the place, this AI energy paradox.

00:01:21: Right, the paradox.

00:01:23: Because AI needs so much power.

00:01:26: Yeah, you look at these large language models, generative AI, the energy cost is the first thing everyone mentions.

00:01:31: We saw Nate Nassauant put it really well.

00:01:33: He said the cloud is basically running out of breath.

00:01:35: That's a great way to phrase it.

00:01:36: Yeah,

00:01:37: and energy availability.

00:01:39: That's now shaping innovation speed.

00:01:41: It's not just about having the best algorithm anymore.

00:01:44: And the scale involved is just staggering.

00:01:46: Eva Cello pointed out Bite Dance, you know, TikTok's parent company.

00:01:50: Oh, yeah.

00:01:51: What about them?

00:01:52: They're data center plans, expanding from three hundred megawatts up to nine hundred megawatts in places like Brazil.

00:01:58: I mean, think about that.

00:01:59: Nine hundred megawatts.

00:02:01: Wow.

00:02:01: At that scale, power, water, they stop being just line items on a bill.

00:02:06: They become the scarce resource.

00:02:08: And that scale really throws up this awkward question, doesn't it?

00:02:12: Julio C. Serrano brought it up.

00:02:13: What's question

00:02:14: was that?

00:02:15: Well, AI is supposed to help save the energy sector, right?

00:02:19: Optimize grids, smart management, all that stuff.

00:02:21: That's the promise, yeah.

00:02:22: But is the AI industry itself creating an even bigger energy problem with its own consumption?

00:02:29: That's

00:02:29: the core tension.

00:02:30: And the numbers, frankly... They are a bit alarming.

00:02:34: We know AI training and running these models drives huge energy spikes.

00:02:38: Like how much?

00:02:39: Well, Microsoft's CO-II emissions up nearly thirty percent since twenty twenty.

00:02:44: Google's almost fifty percent higher than twenty nineteen.

00:02:47: Fifty

00:02:47: percent.

00:02:48: That's that's massive.

00:02:49: That's basically the cost of this gen AI race we're seeing.

00:02:52: It

00:02:52: feels pretty unsustainable, doesn't it?

00:02:53: Yeah.

00:02:54: If the big players are seeing increases like that, they must be pouring money into efficiency, right?

00:02:58: Otherwise, the whole model just breaks.

00:02:59: You'd think so.

00:03:00: They are.

00:03:01: but it's a constant battle.

00:03:02: Kevin Collins mentioned that AI, well, the way it's often done now, is incredibly wasteful.

00:03:07: Energy, water, even capital burn.

00:03:09: So it's inefficient across the board.

00:03:11: But that financial pressure, it's forcing what people are calling green AI.

00:03:16: We're hearing about scientists achieving like, forty x to sixty x efficiency gains.

00:03:21: Forty

00:03:22: to sixty times.

00:03:23: Okay, that's significant.

00:03:24: Yeah.

00:03:25: And it's not just about being eco-friendly, it's really about financial survival.

00:03:29: It's becoming a competitive need.

00:03:31: And

00:03:31: that efficiency focus, it's getting really granular right down into the models themselves.

00:03:36: Mart Arts gave a great explanation of tokens.

00:03:39: Ah yes, tokens.

00:03:41: Super important.

00:03:42: Explain

00:03:42: those a bit.

00:03:43: So, think of tokens as the basic building blocks of language for the AI.

00:03:47: Like smart chunks it processes.

00:03:50: Every single token needs energy to be processed.

00:03:52: Okay, makes sense.

00:03:53: So tracking energy per token, that becomes the key metric for efficiency inside the AI stack.

00:03:58: Optimize how you handle tokens, the model architecture, the hardware, and you cut costs and environmental impact.

00:04:04: It directly links code to the power

00:04:05: bill.

00:04:06: That linkage is critical.

00:04:07: But, you know, this drive for efficiency, you'd expect it to lead to more transparency.

00:04:12: You would, wouldn't you?

00:04:13: But that seems to be a speaking point.

00:04:15: Noah Edwards pointed out a pretty big gap in accountability.

00:04:18: How so?

00:04:19: Well, You've got companies like Anthropic, Google DeepMind.

00:04:24: They're making commitments, partnering with low-carbon data centers, publishing reports.

00:04:29: OK, some are stepping up.

00:04:30: But then you have OpenAI, you know, ChatGPT, the one everyone knows.

00:04:34: And they don't publicly share their environmental footprint.

00:04:37: Really?

00:04:38: That seems inconsistent.

00:04:41: It's

00:04:41: a major point of friction.

00:04:43: And that lag of transparency, it immediately creates risk.

00:04:46: Holly Alpine highlighted a specific example.

00:04:49: What was that?

00:04:49: A shareholder proposal, proposal ten.

00:04:52: Demanding Microsoft disclose the climate and financial risks of its AI tools being used in oil and gas.

00:04:59: Ah, so.

00:05:00: Investors are now connecting the dots, linking sustainability claims or lack theorem directly to financial risk.

00:05:06: Precisely.

00:05:07: So the takeaway here is, yeah, efficiency gains are happening, they're impressive even, but the sheer scale of AI growth might be outpacing them right now.

00:05:14: Which naturally leads to the next big question.

00:05:17: Where do we physically put all this demand?

00:05:19: Perfect segue.

00:05:20: Because the physical infrastructure, the data centers are under a huge spotlight, especially around resources that aren't just carbon.

00:05:26: We're seeing this global demand for more realistic reporting, aren't we?

00:05:30: Definitely.

00:05:31: Nino DuPlan mentioned Dr.

00:05:33: Kevin Gretch urging sustainability reports to properly account for digital water and energy footprints.

00:05:39: Digital

00:05:39: water, I like that term.

00:05:41: Yeah, treating these hidden impacts, you know, with the same seriousness as we treat, say, heavy industry emissions.

00:05:46: And the

00:05:47: water numbers themselves are just mind-boggling.

00:05:50: Doren Soren-Gigler cited a figure.

00:05:52: Data centers globally.

00:05:54: using about five hundred and sixty billion liters of water a year.

00:05:57: Five

00:05:58: hundred and sixty billion liters.

00:06:00: It's almost impossible to picture.

00:06:01: That's like what?

00:06:01: Two hundred and twenty-four thousand Olympic swimming pools?

00:06:04: Every year.

00:06:05: Every year.

00:06:05: And when consumption gets that high, it's not just the utility costs anymore.

00:06:09: It can become a real community issue.

00:06:11: Like the example Shristi Drollier raised about India's biggest data center project.

00:06:15: Exactly.

00:06:16: By twenty thirty, that one project could be using over three hundred and fifty billion liters annually.

00:06:21: That's incredible.

00:06:22: In areas where water is already scarce.

00:06:24: This demands total transparency on their cooling tech.

00:06:27: And frankly, obligations for water replenishment.

00:06:30: You can't just drain the local resources.

00:06:32: No, absolutely not.

00:06:33: And this resource pressure, it explains what Ash Allen termed the clean energy land grab.

00:06:38: Ah, yes.

00:06:39: Companies realizing securing clean power isn't optional anymore.

00:06:44: It's fundamental infrastructure for the AI age.

00:06:46: Like Meta.

00:06:47: Yep,

00:06:47: Meta.

00:06:48: Locking down huge amounts of renewables, seven hundred ninety-one megawatts of clean energy.

00:06:52: They're treating clean power sourcing as a competitive edge.

00:06:55: And it's not just companies, right?

00:06:57: Nations are doing it too.

00:06:58: For sure.

00:06:59: NASA or Albuci noted the UAE's massive six billion dollar plan.

00:07:03: They want to power their whole AI and data center ecosystem with one gigawatt of renewable energy.

00:07:09: So securing clean energy is the new battlefield.

00:07:12: Pretty much.

00:07:12: But in the meantime, while we build out renewables, we also saw discussion around bridging tech like carbon capture.

00:07:18: Himanshu Gaur had some interesting examples.

00:07:20: Go on.

00:07:21: Well, Google is backing a carbon capture and storage project in Illinois.

00:07:25: And Microsoft is exploring direct air capture, DAC.

00:07:28: DAC.

00:07:28: Yeah, direct air capture.

00:07:30: And they're looking at cleverly using the waste heat from the data center itself to power it, kind of a symbiotic system.

00:07:36: That's clever, using waste heat.

00:07:38: But while new builds grab headlines, we can't forget existing infrastructure.

00:07:44: Thomas M. made that point.

00:07:45: Very important point.

00:07:47: Optimizing what we already have.

00:07:48: Smart retrofitting of current data centers offers huge, immediate efficiency wins.

00:07:54: And Liam Papel highlighted circularity too, right?

00:07:56: Using secondhand buildings, focusing on server life cycles.

00:07:59: Exactly.

00:08:00: This whole idea of integration, of not just building new, but optimizing and reusing.

00:08:05: It's driving new ways of measuring things too.

00:08:08: Chad McCarthy introduced this initiative.

00:08:10: The Enlighten Feme one with the ICFN scores.

00:08:13: That's

00:08:13: the one.

00:08:13: Integrated carbon-free energy scores.

00:08:16: Tell me more about that.

00:08:16: What makes ICFN different?

00:08:18: It's a really fundamental shift in thinking.

00:08:21: Traditional metrics.

00:08:22: They look at a data center as just a consumer.

00:08:24: An isolated island drawing power.

00:08:26: Okay.

00:08:27: ICFN changes that.

00:08:28: It formally includes things like heat recovery and the score.

00:08:31: It looks at how well the facility integrates with the community's energy system.

00:08:34: So it's not just about what it takes, but what it potentially gives back, like heat for local buildings?

00:08:40: Precisely.

00:08:41: It reframes the data center from just being a drain on resources to potentially being a productive community asset.

00:08:49: Treating data centers like integrated community assets, I really like that framing.

00:08:54: It's a powerful idea.

00:08:55: And it leads us nicely up the stack, doesn't it, from the physical building to the actual code running on it.

00:09:00: Yep,

00:09:01: moving from hardware to software.

00:09:03: And we're seeing real maturity here in standards for sustainable software engineering.

00:09:09: Rafa Nadim defined it pretty clearly.

00:09:11: He did.

00:09:11: Basically, minimizing environmental impact through efficient code, smart resource use, green tech, standard engineering principles, really, but with an environmental lens.

00:09:21: And Ann Curry made that crucial point about co-benefits, which is why this isn't just a nice to have.

00:09:26: Co-benefits,

00:09:27: yes.

00:09:28: Her point was simple.

00:09:29: A greener system is usually just a better system.

00:09:31: How so?

00:09:32: Because building green forces you to be efficient.

00:09:35: An efficient code is cheaper to run, it's faster, it's often more resilient, more secure.

00:09:39: Inefficiency costs money in every direction.

00:09:41: So it creates this really compelling business case, not just a moral one.

00:09:46: Exactly.

00:09:46: Which is why we're seeing it translate into actual procurement rules.

00:09:50: Rachel Keilenberg mentioned the Bayer Group ICT software developing practical tools.

00:09:55: Like a schedule of requirements.

00:09:56: A KPI tool.

00:09:58: Tools to structurally include sustainability criteria in public sector software tenders.

00:10:03: That forces developers to consider it from day one.

00:10:06: That makes sense.

00:10:07: Embed it in the buying process, but holding vendors accountable.

00:10:11: We need to look beyond just the electricity used when the software runs, right?

00:10:14: Oh,

00:10:15: absolutely.

00:10:15: Anita Shetler tackled that misconception head on.

00:10:19: The idea that IT's emissions are minimal.

00:10:21: That idea only works if you ignore most of the picture.

00:10:25: Precisely.

00:10:25: If you only look at electricity used during operation, you miss the huge footprint from Scope III missions.

00:10:31: And Scope III is everything else, basically.

00:10:34: Yeah,

00:10:34: the tricky stuff.

00:10:35: Hardware manufacturing, building the data centers, even the energy used by the software on our devices, our laptops, our phones.

00:10:42: If you ignore Scope III, you're really kidding yourself about the total impact.

00:10:46: It's good to see the industry starting to respond though.

00:10:48: Charles Humble confirmed AWS is adding scope three data to its carbon footprint

00:10:53: tool.

00:10:53: That's a big deal.

00:10:55: Covering embodied carbon from hardware, data centers, transport.

00:10:59: That allows for much more accurate accounting by their customer.

00:11:02: And that transparency needs to filter down, right, to individual products.

00:11:06: Pablo Jose Gomez or SOSMO stressed transparency as fundamental.

00:11:10: He did.

00:11:11: Arguing for interoperable systems like a digital product information system, or DPIS, to track sustainability data across the lifecycle.

00:11:20: The current landscape is too fragmented.

00:11:22: Okay, so transparency across the supply chain.

00:11:24: What about the design itself, the user experience?

00:11:27: Torsten

00:11:27: Jonas made a great point about sustainable UX, his argument.

00:11:31: User convenience often highs an environmental cost.

00:11:33: Like what?

00:11:34: Autoplaying videos.

00:11:35: Exactly.

00:11:36: Or... constantly fetching data we don't need.

00:11:38: Designing to reduce that kind of digital waste.

00:11:41: It saves costs and gets ahead of potential future regulations.

00:11:44: It's just smart design.

00:11:45: Smart design.

00:11:46: And finally, standards seem to be favoring certain approaches.

00:11:49: Philip Kirsting noted open source has an edge.

00:11:52: Yeah, particularly for certifications like the German blower angle, the Blue Angel eco label.

00:11:57: Because the code is open, it's just easier to provide the evidence needed to prove it meets the sustainability criteria.

00:12:04: Transparency helps verification.

00:12:08: Okay, so wrapping this deep dive up, it's really clear that green ICT isn't just aspirational talk anymore.

00:12:13: It's about professionalizing the field.

00:12:16: We've gone from vague goals to, like, measuring the energy cost of a single software token.

00:12:23: Yeah,

00:12:23: to serious discussions about water stewardship and these new metrics like ICP, that C-TECH is part of the community grid.

00:12:30: But we also need that dose of realism that James Martin Wilco Bergeroff called for.

00:12:35: The projected impact of Gen AI is enormous.

00:12:38: We can't just rely on blind optimism about progress.

00:12:41: No.

00:12:41: Being realistic about where we actually are is crucial if we want to hit climate targets.

00:12:46: The stakes feel incredibly high right now.

00:12:48: So maybe the thought to leave everyone with comes from Anna Lerner-Nesbitt's observation about how organizations are structured.

00:12:54: She noted that most companies treat their AI teams as strategic investments, pouring resources into them.

00:13:00: OK,

00:13:00: that makes sense.

00:13:01: But the sustainability teams, they're often treated purely as a cost

00:13:04: center.

00:13:05: And critically, these two teams, they rarely talk to each other internally.

00:13:09: Even though their issues, energy, resources, efficiency are fundamentally linked, that's quite a disconnect.

00:13:15: It is.

00:13:15: So the real question for you listening is probably this.

00:13:19: What will it take inside your organization?

00:13:22: Yeah, what steps can you take to get sustainability seen, not just as a cost to be managed, but as a core strategic investment?

00:13:30: An investment in resilience, efficiency, and frankly, future viability.

00:13:34: A lot to think about there.

00:13:35: If you enjoyed this deep dive, new episodes drop every two weeks.

00:13:39: Also, check out our other editions on cloud, digital products and services, artificial intelligence, and ICT and tech insights, health tech, defense tech.

00:13:48: Thanks for diving deep with us today.

00:13:49: And make sure you subscribe so you don't miss our next analysis.