Latest / Best of LinkedIn: Digital Transformation & Tech / Best of LinkedIn: Sustainability & Green ICT CW 18/ 19

Transcript

Best of LinkedIn: Sustainability & Green ICT CW 18/ 19 · Best of LinkedIn: Digital Transformation & Tech

• 0:00
  Speaker 1: This episode is provided by Thomas Allgeier and Frennis, based on the most relevant LinkedIn posts about sustainability in green ICT in CW-E eighteen and nineteen.
• 0:09
  Speaker 1: Frennes supports ICT enterprises in the form of delivering precise ICT market and pricing intelligence that analysts subscriptions and existing databases cannot provide.
• 0:19
  Speaker 1: you can find more info in the description.
• 0:21
  Speaker 1: fine.
• 0:21
  Speaker 1: so imagine buying a diet soda specifically to justify eating like three extra large pizzas
• 0:28
  Speaker 0: stomach my weekend.
• 0:29
  Speaker 1: Yeah, well that is exactly the kind of math the tech industry's doing right now when it comes to AI and energy efficiency.
• 0:35
  Speaker 1: Today we're really tearing down the greenwashing.
• 0:38
  Speaker 0: We absolutely are!
• 0:39
  Speaker 0: We're looking at top sustainability in Green ICT trends.
• 0:42
  Speaker 0: It really dominated conversation across LinkedIn over these past two weeks.
• 0:47
  Speaker 0: The underlying theme here is just a massive collision between digital ambition and physical reality.
• 0:52
  Speaker 1: It's huge, we've got a massive stack of insights to get through.
• 0:55
  Speaker 1: We're gonna cover everything from the hidden energy costs of AI To you know...the deep mechanics of green software engineering
• 1:02
  Speaker 0: And sustainability reporting landscape which is honestly bit of wild west right now
• 1:06
  Speaker 1: Totally!
• 1:07
  Speaker 1: The goal for you listening right now Is cut-through noise.
• 1:11
  Speaker 1: We want extract actionable evidence based insight that can take back to your own IT infrastructure.
• 1:18
  Speaker 0: Exactly, understanding not just what is happening but you know the actual mechanics of why?
• 1:23
  Speaker 1: Right.
• 1:24
  Speaker 1: so let's start with The Elephant in the Room which is AI's explosive growth.
• 1:28
  Speaker 1: Mark Bradley shared some numbers that really set the baseline for this physical cost.
• 1:33
  Speaker 0: yeah his post was eye-opening.
• 1:35
  Speaker 1: we hear a lot about AI taking energy.
• 1:40
  Speaker 1: It's hard to grasp.
• 1:41
  Speaker 1: Bradley points out that a single chat GPT query takes about ten times the electricity of a standard Google search.
• 1:47
  Speaker 0: Ten times, just for one query?
• 1:49
  Speaker 0: Right.
• 1:49
  Speaker 1: and when you zoom-out at the macro level A single new data center in Illinois needs seven hundred megawatts power.
• 1:55
  Speaker 1: And get this three million gallons water per day.
• 1:58
  Speaker 0: That is a day!
• 1:59
  Speaker 0: Its staggering.
• 2:00
  Speaker 1: Bradley argues we have apply phenops discipline here and quote treat carbon like cash.
• 2:06
  Speaker 0: Treating Carbon Like Cash requires understanding how the spending actually works, right?
• 2:11
  Speaker 0: Because a traditional Google search is essentially a retrieval operation.
• 2:14
  Speaker 1: Right it's just looking something up
• 2:16
  Speaker 0: exactly.
• 2:17
  Speaker 0: It's looking up information in an incredibly well organized pre-existing index.
• 2:22
  Speaker 0: But generative AI query Is fundamentally different.
• 2:26
  Speaker 0: its calculating probabilities for every single word it generates in real time.
• 2:32
  Speaker 0: Its
• 2:32
  Speaker 1: firing billions of parameters to synthesize and answer yeah
• 2:36
  Speaker 0: The compute requirement is orders of magnitude higher.
• 2:39
  Speaker 1: Which makes sense on a per-query basis, but the defense you always hear from the tech giants... ...is that models are getting vastly more efficient.
• 2:46
  Speaker 0: Sure they say so
• 2:47
  Speaker 1: And they ARE right!
• 2:48
  Speaker 1: Ben Hillier highlighted Google reduced energy required for a Gemini query by thirty three times in just one year.
• 2:55
  Speaker 0: They absolutely did.
• 2:56
  Speaker 0: But Hillier also points out this brilliant counterintuitive reality here known as the Jevons Paradox.
• 3:02
  Speaker 1: Oh right, the economic principle
• 3:04
  Speaker 0: Yeah.
• 3:05
  Speaker 0: It says that as technological progress increases the efficiency of a resource, demand increases.
• 3:15
  Speaker 1: It's like adding a lane to a congested highway, it doesn't actually reduce the traffic...it just makes driving slightly easier for a minute which invites thousands of new cars until the wider highway is completely jammed again.
• 3:26
  Speaker 1: That is
• 3:26
  Speaker 0: the perfect analogy making inference The actual running-of-the AI cheaper and more efficient Just integrates into thousands of news software products Which drives up aggregate demand right?
• 3:38
  Speaker 0: And we have to look at the other side of the ledger, which is the training phase.
• 3:42
  Speaker 0: Hillier notes that the energy required to train frontier models has grown roughly thirty-seven thousand times in six years.
• 3:48
  Speaker 0: Thirty
• 3:49
  Speaker 1: seven thousand times,
• 3:50
  Speaker 0: wow!
• 3:51
  Speaker 0: Training GPT-IV alone took an estimated fifty-six thousand megawatt hours.
• 3:56
  Speaker 1: Okay so diet soda is the inference.
• 3:57
  Speaker 1: efficiency Yeah.
• 3:58
  Speaker 1: And three large pizzas are training costs and massive spike in aggregate demand.
• 4:02
  Speaker 0: Exactly The hangover from those pizzas lasts a very long time.
• 4:06
  Speaker 0: James Martin highlighted new MIT paper by Jennifer Turlich & John Sturman.
• 4:11
  Speaker 0: What
• 4:11
  Speaker 1: did they find?
• 4:12
  Speaker 0: They look at the actual atmospheric mechanics.
• 4:41
  Speaker 0: simply doesn't work that way.
• 4:42
  Speaker 1: Okay, let's unpack this.
• 4:44
  Speaker 1: if we're training models that require fifty six thousand megawatt hours isn't chasing per query efficiency just a massive distraction from the real problem?
• 4:53
  Speaker 0: It certainly masking larger structural issue.
• 4:55
  Speaker 0: Naveen Balani brought up this exact tension.
• 4:58
  Speaker 0: He argued that the current token economics model is completely breaking down,
• 5:01
  Speaker 1: and for you listening a token as essentially A piece of word That AI processes right?
• 5:05
  Speaker 0: And Balani's point Is at The next decade Of ai isn't going to be one by whoever Spends the most on compute.
• 5:11
  Speaker 0: it's gonna Be won By Whoever Waste the least.
• 5:13
  Speaker 1: that Makes total sense.
• 5:14
  Speaker 0: This Shear Velocity If This Growth is hitting a wall.
• 5:17
  Speaker 0: Karen VanderZand pointed out that Microsoft is currently having to deeply rethink their twenty-thirty climate goals.
• 5:24
  Speaker 1: Because they're sustainability commitments are directly threatened by their own AI deployment velocity, which brings us the physical reality of all this compute.
• 5:33
  Speaker 1: because software doesn't run in a vacuum.
• 5:36
  Speaker 1: Nope!
• 5:37
  Speaker 0: It has a physical footprint.
• 5:38
  Speaker 1: If we can't stop the AI train We have to re think where and how it runs in the physical world.
• 5:45
  Speaker 1: I want bring up Mathieu François.
• 5:47
  Speaker 1: he shared some mind-blowing research with Professor Xiaolei Ren regarding water stress.
• 5:52
  Speaker 0: Water is basically the primary mechanism we use to pull heat away from these massive GPU clusters, without it that chips melt!
• 5:59
  Speaker 1: Right and Francois and Ren found that a Lemma A-B query consumes eight point nine six milliliters of water in India versus three point two six millimeters in Germany.
• 6:08
  Speaker 0: Okay so on the surface That's less than a Three X difference.
• 6:11
  Speaker 0: volume
• 6:12
  Speaker 1: Yeah, you might think.
• 6:13
  Speaker 1: well the servers in India are just a bit less efficient at cooling.
• 6:15
  Speaker 0: But volume isn't the metric that matters here?
• 6:18
  Speaker 1: Exactly!
• 6:18
  Speaker 1: The metric that matter is regional capacity to replenish water.
• 6:23
  Speaker 1: Because of regional scarcity... ...the actual water stress impact is twenty-one times higher in India.
• 6:29
  Speaker 0: Twenty one times for exact same query.
• 6:31
  Speaker 1: Geography's primary variable not just volume.
• 6:34
  Speaker 0: This creates an incredibly difficult infrastructure bottleneck.
• 6:39
  Speaker 0: The power grid and the local water supply simply cannot move as fast, is a software company deploying new AI features.
• 6:46
  Speaker 0: Yeah And we're seeing two wildly different responses to this bottleneck popping up in our sources.
• 6:52
  Speaker 0: on one hand We have the brute force high friction approach.
• 6:54
  Speaker 1: right then negative one.
• 6:56
  Speaker 1: yeah
• 6:56
  Speaker 0: Anna Lerner-Nesbitt and Chris Adams brought attention to reports about anthropic leasing capacity at XAI's Colossus Facility in Memphis.
• 7:03
  Speaker 1: And just to be clear, we are purely reporting what was highlighted in these LinkedIn posts.
• 7:08
  Speaker 1: Yes!
• 7:09
  Speaker 0: The concern Nesbett & Adams raised is that this facility has reportedly operated dozens of unpermitted gas turbines bringing massive pollution into a frontline community.
• 7:17
  Speaker 1: But think the mechanics.
• 7:19
  Speaker 1: why a tech company would resort to burning gas locally.
• 7:22
  Speaker 1: A traditional power grid requires years of planning right?
• 7:25
  Speaker 0: It's exactly.
• 7:26
  Speaker 0: Yeah, permitting substation construction all to deliver five hundred megawatts safely and AI arms race demands that compute power now so you bypass the grid.
• 7:37
  Speaker 1: But then there's the other side of the coin, which is innovative infrastructure approaches.
• 7:44
  Speaker 1: Experts across our sources are tracking ocean-based solutions.
• 7:48
  Speaker 1: Amir Olajuan highlighted China testing underwater data centers cooled by seawater.
• 7:54
  Speaker 1: Stella Su discussed Pantherlassa building offshore AI datacenters powered entirely by wave energy.
• 8:00
  Speaker 1: And Sean Moran noted at Jacobson Start Campus Project in Portugal, achieving a water usage effectiveness of exactly zero.
• 8:07
  Speaker 1: Zero?
• 8:08
  Speaker 1: Zero!
• 8:08
  Speaker 1: By using Atlantic seawater for cooling.
• 8:10
  Speaker 0: Okay but I have to push back here.
• 8:12
  Speaker 0: Putting data centers underwater sounds like James Bond villain's lair.
• 8:15
  Speaker 1: It totally does.
• 8:16
  Speaker 1: Isn't
• 8:16
  Speaker 0: the maintenance, corrosion and connectivity of saltwater immersed tech just a logistical nightmare that completely offsets the cooling benefits?
• 8:24
  Speaker 1: The physical trade-offs are definitely high.
• 8:26
  Speaker 1: Salt Water is notoriously destructive But you have to understand the sheer desperation of thermal management right now.
• 8:33
  Speaker 1: James Hall observed that an Amazon data center in Virginia recently went down simply because it got too hot.
• 8:40
  Speaker 0: Wait, It didn't get hacked?
• 8:41
  Speaker 1: No!
• 8:42
  Speaker 1: A backhoe didn't cut a fiber cable, it just overheated.
• 8:46
  Speaker 1: Because AI chips run radically hotter than traditional cloud servers.
• 8:50
  Speaker 1: GreenOps isn't about saving carbon anymore
• 8:53
  Speaker 0: It's the sheer resilience and survival of technology.
• 8:56
  Speaker 0: Exactly
• 8:57
  Speaker 1: We're being forced into extreme environments because the traditional grid cannot handle thermal load.
• 9:03
  Speaker 0: If hardware is fighting for its life thermally we need to look at what's commanding that hardware so hard in first place.
• 9:10
  Speaker 0: The software Exactly.
• 9:12
  Speaker 0: The software, hardware only uses energy when software tells it to.
• 9:15
  Speaker 0: which brings us to our third theme Green Software and Engineering.
• 9:19
  Speaker 1: This is where we shift from physical geography To digital frugality.
• 9:23
  Speaker 1: Yeah
• 9:23
  Speaker 0: And Paul Young shared an incredible win From GitHub that illustrates this perfectly.
• 9:28
  Speaker 0: They managed to cut their agentic workflow tokens by forty three to sixty two percent.
• 9:32
  Speaker 1: That's massive!
• 9:33
  Speaker 1: For context An agent work flow.
• 9:36
  Speaker 1: When you give AI autonomy break down a big task, use tools and figure out the steps on its own.
• 9:42
  Speaker 1: Right!
• 9:43
  Speaker 0: And every time the AI thinks about step it processes tokens which burns energy.
• 9:49
  Speaker 0: GitHub achieved this massive reduction by simply moving basic data gathering at of LLM reasoning loop.
• 9:55
  Speaker 1: Oh interesting.
• 9:56
  Speaker 1: so what did they used instead?
• 9:58
  Speaker 0: Just simple command line interface commands CLI commands.
• 10:01
  Speaker 0: Using a billion parameter LLM to figure out what files are in a folder is like chartering a Boeing seven forty-seven To drive across the street to pick up groceries.
• 10:09
  Speaker 1: That's a great way to put it.
• 10:10
  Speaker 1: They also stripped out unused tool schemas didn't they?
• 10:13
  Speaker 0: Yeah, those were adding ten to fifteen kilobytes of overhead per API call.
• 10:16
  Speaker 1: So what does this all mean for the people writing the code?
• 10:19
  Speaker 1: Because it feels like we've been asking a chef to cook as zero waste meal without giving them a compost bin or proper knives.
• 10:25
  Speaker 1: We keep giving developers guilt trips instead of actual tools.
• 10:28
  Speaker 0: I love that framing and David DeCarvalho, and Johanser brought up this exact necessity.
• 10:35
  Speaker 0: Eco-design has to move past just sensitizing developers.
• 10:39
  Speaker 1: They need concrete tools.
• 10:41
  Speaker 0: Exactly, they specifically pointed to the API green score which gives a measurable metric on efficiency and they highlighted frameworks like Spring Native To make digital frugality A daily measureable reality.
• 10:54
  Speaker 1: And education is catching up too.
• 10:55
  Speaker 1: Rhett Palmera shared that The Green Software Foundation now has free self-paced courses.
• 11:01
  Speaker 1: Always great!
• 11:02
  Speaker 1: Anita Shutler discussed Blue Angel certification for sustainable software
• 11:06
  Speaker 0: Which perfectly aligns with Purvivarma's core argument.
• 11:09
  Speaker 0: She emphasizes that sustainability cannot be a side initiative.
• 11:13
  Speaker 0: It must be a core cloud architecture principle
• 11:15
  Speaker 1: right alongside cost and performance.
• 11:17
  Speaker 0: Exactly, but here's the catch to prove to your leadership That using spring native actually saved energy you need bulletproof data.
• 11:25
  Speaker 1: And now leads directly into our final theme The fragmented world of sustainability report.
• 11:30
  Speaker 0: yeah You can't reduce what you haven't measured.
• 11:33
  Speaker 0: Pascal Jolie shared a post detailing at baseline AI emissions assessment He just completed for octa.
• 11:38
  Speaker 1: He was mapping the carbon footprint of their LLM chatbots, right?
• 11:42
  Speaker 1: Yeah.
• 11:43
  Speaker 0: Yes and his biggest takeaway is that most sustainability teams have zero visibility into enterprise AI usage.
• 11:51
  Speaker 1: They are completely flying blind
• 11:53
  Speaker 0: Totally flying blind And part reason because the methodology still so raw.
• 11:57
  Speaker 0: Dr Sasha Lucione's literature review found that AI environmental reporting Is currently relying heavily on course proxies
• 12:05
  Speaker 1: Because the actual life cycle of AI is still poorly defined, so they're just guessing emissions based on money spent.
• 12:12
  Speaker 0: Exactly!
• 12:13
  Speaker 0: And when a metric is poorly-defined it leaves massive loopholes.
• 12:17
  Speaker 0: Vinayak Setpute and Shreya Gupta broke down.
• 12:20
  Speaker 0: nine tactics companies use beyond traditional greenwashing.
• 12:23
  Speaker 1: Tactics like greenhushing?
• 12:24
  Speaker 1: Yes
• 12:25
  Speaker 0: staying silent about genuine wins to avoid audits on other initiatives...and green crowding
• 12:30
  Speaker 1: Right, where you hide in slow-moving industry coalitions.
• 12:33
  Speaker 1: You sign a pledge to be carbon neutral by twenty fifty put the badge on your website and then do absolutely nothing.
• 12:38
  Speaker 0: Exactly!
• 12:39
  Speaker 0: The only antidote of these tactics is hard evidence.
• 12:42
  Speaker 0: And the market's responding?
• 12:44
  Speaker 0: Bjorn Stengel and Brianna Weiss mentioned.
• 12:46
  Speaker 0: IDC just mapped over thirty sustainability software vendors.
• 12:49
  Speaker 1: Charlie Appleton highlighted partnership between Greenpixie and Generate Zero To combine AI usage data with scope one through three carbon tracking
• 12:57
  Speaker 0: And Jay Ruckelshaus pointed out that the IFRS is shifting ESG disclosure rules to focus on substance over
• 13:04
  Speaker 1: form,
• 13:05
  Speaker 0: meaning disclosures must connect directly to financial performance and value creation.
• 13:10
  Speaker 1: Right now though it sounds like everyone is grading their own homework using entirely different rubrics
• 13:16
  Speaker 0: Pretty much.
• 13:17
  Speaker 1: When do we finally get a standardized test for carbon accounting so you can actually compare company A to company B?
• 13:22
  Speaker 0: That's exactly what Kate Goldenring or Svia Segment are pushing shared emissions standards.
• 13:28
  Speaker 0: Until we have interoperable data, as Bjorn Soaring-Gigler mentioned climate intelligence won't be truly actionable.
• 13:35
  Speaker 1: We're basically building the airplane while flying it!
• 13:37
  Speaker 0: We really are.
• 13:38
  Speaker 0: and that brings me to a final thought.
• 13:39
  Speaker 0: I want to leave the listener with today weaving together Marosa Zalabak's point about AI ethics
• 13:44
  Speaker 1: The question of who benefits?
• 13:45
  Speaker 1: And Who is harmed?
• 13:47
  Speaker 0: Yes pair that with Nelson Switzer rule of forty four regarding climate tech scaling.
• 13:52
  Speaker 0: The challenge for you listening is this, as you look at your own organization's digital transformation don't just ask if your new AI tool or cloud architecture is highly efficient.
• 14:02
  Speaker 0: Ask if the compute power is genuinely creating enough human-or business value to justify its physical footprint on earth?
• 14:11
  Speaker 0: Are we building efficient systems of things that we do not need in first place?
• 14:15
  Speaker 1: That is a phenomenal question to end on.
• 14:17
  Speaker 1: Yeah, if you enjoyed this episode new episodes drop every two weeks.
• 14:22
  Speaker 1: also check out our other editions on cloud digital products and services artificial intelligence an ICT in tech insights health tech defense
• 14:30
  Speaker 0: Tech.
• 14:31
  Speaker 0: thanks for listening everyone.
• 14:32
  Speaker 1: thank You so much for taking the steep dive with us today.
• 14:34
  Speaker 1: don't forget To subscribe And we'll catch On The next one.