Navigating Sustainability in the Age of AI

[MUSIC PLAYING] Welcome to the Data Edit, the podcast for data forward thinkers. Hosted by Agiles Ben-Harris, head of AI, and Oli Seger, account director, we explore the challenges, innovations, and trends shaping the data landscape. From governance in ESG to AI ethics, cloud migration, and analytics, discover how data is driving real business value. So hi there, and welcome back to episode two in our series of podcasts. In the last one, we talked pretty much-- it would be remiss if it was not to have discussed AI. So we did that. Obviously, that leads to various conversations. We talked last time about organizations really trying to think about using AI to generate positive outcomes and address business challenges, rather than using AI for the sake of using AI. Interested, we had a couple of comments on that, which we'll come to, and we may have applied by LinkedIn. But what we wanted to talk about today was, obviously, AI is only of value based on the data that feeds it, and the requests that go through to the data itself, and the impact that the massive demand and increase of data requirements is having on the actual data centers that are storing the data and the compute power required to analyze the data and provide the responses by the various AI type tooling. So where do you want to start, Ben? Have you got any interesting stats or figures around this kind of topic? Yeah, so hi, and welcome back to everybody. So I think, for me, I think we've almost got two topics here and probably two separate podcasts. So there is that whole shell of what the actual data center looks like, how that works, how that operates, and then be what we're putting in there and why we've been in there in the first place. So let's tackle the first one. So there is a-- and I think we need to recognize it, to be perfectly honest. I think there is a new sector or a new industry, which is now starting to pop up. We are seeing in the UK, so the UK is growing in its AI and data center kind of usage. And obviously, as we talked about previously, more AI means more data centers, more requirements, more power consumption, et cetera, et cetera. And the UK is responding. We are building data centers now in the UK. There is a demand at the moment of about 40 to 50,000 employed people looking after these data centers. And approximately, it takes about 40,000 people to build one. Now, that to me seems absolutely crazy. But actually, when you go back into the base of it, it's from the guy, I guess, who does the initial survey of land all the way through to those guys that put all that wiring in, all the safety systems, everything else. I mean, that's how many people actually get touched in that overall process. It's a massive stat. So are we actually now starting to build a new ecosystem, a new sector within AI? Well, not within AI, but within UK business, around AI, around the use of data. And specifically, I guess, looking at the UK, considering that's where I'm from and you are too, it's expected to be 10% year-on-year growth. So at the moment, we're spending something like $4 billion. By 2035, it's expected to be $44 billion. That to me is, yeah, that's nuts. Yeah, that's a significant increase, isn't it? It's a massive market, yeah. I guess, whilst absolutely that creates potentially a new silo within the construction industry, I guess, which is what you're saying kind of related to data center build and ethical build, the challenge, I guess, which is being addressed, I know, from various of the sort of hyperscalers in terms of their data center build program, the environmental and sustainability piece around their builds. But it's not just that, is it? It's the actual impact upon in the UK specifically here. Our own power for you, me, everyone, the individuals that reside in this country. How is the increased demand for the power consumption on those data centers that we're building, 10% year-on-year, affecting the availability of some of that power for you and me? And us being able to access sustainable power, electricity, carbon-neutral electricity, because the hyperscalers are demanding that to make their own operations sustainable if you call net zero, carbon zero, whatever. So I guess there's various kind of things that could impact this, aren't there? Have we got enough power? Are we getting it from sustainable sources? And if we are, does that mean that Joe Bloggs is unable to access that green energy for our own benefit? Yeah, and I guess from my point of view, and not trying to get too political at all there, but what actually gets turned to green energy? Just because nuclear today means we don't have to dig a hole in the ground and extract coal. Actually, we're still digging holes in other parts of the country to store the spend fuel, because we have no other way of getting rid of it. So we're now trying to-- I guess there's an ongoing conversation that needs to happen. Are we actually just borrowing? Are we using state borrowing from the future? Are we making it somebody else's problem by calling it? Renewable energy. There's all sorts of stats and whether you want to believe them or not around how long it takes for solar panels to degrade. How useful they are before, how long can you actually get a valid power supply from them? And as I understand it currently today, happy to be challenged on this, but welcome comments as well from anybody listening to the podcast. How do we dispose of them? Because my opinion is at the moment, there isn't a way to actually take them apart, reuse them, recycle any parts from them, the actual chemical aspects once they're used. They're used and they are potentially toxic. So I don't want to make it political. By any means the imagination, but I think there is this concern about from society, we're all told electric cars are the way for you forward. And you're going to touch on that. We had a conversation about this a while back, didn't we? Yeah. We're all relevant. It's all relevant, right? Because the consumer world is replicated either before or after the business world. And we should absolutely be making decisions that we would take sensibly in our own personal lives and they should be translated into decisions that are made in the business world when it comes to ethically building sustainable futures for not just my kids, but their kids, their kids, et cetera. And are we too late? Or are we just in time? We just need to think about it. Maybe this podcast is the catalyst for a various vote. Just kind of sit up and have a think about this. Because not that we're going to change the world by any stretch, but these are things that we should all be thinking about, right? Yeah. And I think, again, it kind of goes back to I can't have an electric car in my area at the moment. Because, alluding to your earlier point, the National Grid isn't capable of delivering enough power into my area, me to be able to have an electric charge point at home. Now, as you rightly said, we've touched on this a few times. There are people who just don't put them in and come what may. There have been unfortunate incidents and fires, et cetera, et cetera. I know people in my area who just dangle a cord out of their upstairs window. And it's just a 240-volt power cord that just goes out. And they can't charge their car if it's raining. Yeah. You know, all those kind of things. And so people are obviously trying to find ways around it. We're told that we have to go down that. So I'm interested to know what, where, when-- tell me more about your kind of your mini-nuclear reactors. Because I think this is potentially a way forward. But I do wonder if we're still borrowing a little bit from the future. Yeah. It's an issue. It was an article I read recently about Rolls-Royce basically wanting to position themselves as one of the-- well, the global leader of nuclear energy for good, I guess. There are various use cases where they call these SMR small modular reactors available for-- one of the things they actually state specifically is for data centers. Now, that feels like it could be an answer. There's no right answers or wrong answer or whatever. But from the UK perspective, with the UK company, would we not be wise to support that? You know, the likes of the hyperscalers in other parts of the world when they're building data centers are addressing and trying where they've got North America, et cetera, vast amounts of space and land. And they can use different types of energy. We don't have that in the UK, right? So maybe this is a problem that could solve our UK and some of Northern Europe problems. Would it be translatable to the US? I'm not sure. But it feels like it's a good opportunity. A company like Rolls-Royce would not be investing millions in this technology if they didn't think it had legs. And to call out data centers, they've obviously recognized that this was a problem brewing a few years back or however many years back. They decided to start working on this or translating the use of the nuclear reactors that they've been using for years into more of a commercial business model. Yeah, it might be an answer. Again, let's see what people's thoughts are. Absolutely. I think there's an interesting aspect in there as well about it kind of shows when money's been spent in the past. As a country, all countries promote defence. And that's obviously where Rolls-Royce have got this from. We all know that boats and ships and submarines are all running on nuclear reactors these days. So they're out there. They're already in use. They've already proved that they're safe. I haven't heard of a UK touch a lot of woods. I don't want to see a radio. There's now today, thanks. But you don't hear about any incidents or accidents. They're obviously safe. They've obviously developed and they've been working on them for a number of years. So they must be a commoditised product. I think there's still some other aspects that are interesting and potentially need to be thought through about. We can solve potentially the electrical issue using those SMRs, but can we solve the distribution of heat? So I appreciate your kind of systems and bits of pieces. You've talked to me three before that you've seen and read about. But that still has to be dissipated into the atmosphere somewhere. So, you know, it's Norfolk where I live. Is that going to become the new cost they'll solve with record temperatures every year of 35 degrees on the beach? And in which case, I need to make an investment. Or in some new flip-flops. To have the beach out. Yeah, yeah. And that's a really interesting thing as well, isn't it? The location of these things, you know, being near water sources, for example, and sustainable water supplies for cooling and that sort of stuff. You know, there's logic that needs to be applied as well, right? Yeah, look, this is we talked at the start of the beginning. This is one of two parts. We don't have the answers. You know, this is a chit chat to see what other people think. You know, I'd love to get some insight into, you know, some slightly more informed brains, perhaps, as to, you know, what is being done and what the answers might or might not be. But it is a sure, sure five consideration that feeding in from the AI. AI needs data. Data needs to be housed somewhere. And it needs to be computed, if that's such a word. Then, you know, the increased demand from all follows all the way through. And, you know, the energy, the energy requirement is super important. Yeah, like I say, interested to hear anyone's thoughts on this. I think we'd probably keep this is a pretty short and snappy podcast today because we want to move into the second part of this kind of potential data center issue around the actual amount of data that we're storing in these data centers that may or as you might find out in the next one, may not be required. So unless you've got any other comments, Ben, I'm happy to wrap this one up so short and sweet because it's really kind of just what's people's opinions and thoughts. And we touch on this. Again, I guess the one big one, and again, it's kind of like, I don't have an answer to it, it's just, you know, it goes all the way back to that very first question. Are we trying to plan for something that may never happen? I would love, you know, to hear other people's kind of views, the societal impact of AI on the world. Are we trying to just, it's a big thing right now, yeah? We said it in the podcast. But are we actually trying to run before we can walk? And are we trying to set ourselves up for a demand just because people are going, let's have some AI, rather than, you know, maybe in a couple of years once that initial wave of anticipation has gone, will the demand for all of this reduce a little bit or is it here to stay? I'm sure it's here to stay in this volume. I had a couple of looks at a couple of stats, and on average, SkyFi films are about 30 to 50 years ahead of where we are societally. Yeah, yeah, yeah. No, it's kind of like... I know you're going with this, yeah. Interesting. It is interesting. Are we being driven by, are we being driven a little bit by our past as well from the wants and the promises from SkyFi era of this computer and technology and AI and thought-provoking stuff will actually, yeah, allow us to do all these other things, is that just misfortune, forwards? Go on then, give us some examples. Oh, no, I don't want to scare people. Obviously Terminator's going to come through. Yeah, yeah. What's their name? Sarah Connor. Sarah Connor. Yeah. It's interesting to say, okay, a couple of examples, just because I think probably the first film where I really saw some type of AI, 2001 Space Odyssey. Uh-huh. Uh-huh. Yeah. Showing our age, yeah. Within 50 years, Hal became Siri, and now most people have a version of it on your mobile phone. Yeah, yeah. Yeah. We can watch. Yeah. I actually looked like someone, I had someone phone me yesterday, my phone was inside, I did answer it on my watch. I've done it a few times, yeah. Yeah. I was that starter. Yeah. Manu Nanu. Wargames, which was one of my favorite films when I was growing up, 83 that was released. Yeah. Yeah. Teaching a computer how to play chess and the benefits and stuff. Yeah. I've been using AI for the past all machine learning parts of AI for the past 10, 15 years now. Yeah. Yeah. So that one's come true, Terminator, not fully seen yet, but we do have AI drains. We do have drains that can actually operate as forming effects in real life. Unfortunately, at the moment, I've got one here around Jurassic Park, but I'm not really 100% sure I want to be around the world of AI systems and genetics, that's a step too far. Interesting. Do you know what, just back on that drone thing, I was watching, gosh, it was a couple of years ago, and I remember watching the New Year's Eve celebrations thing on TV, you know, they do all around the world at different times and got to London and I was thinking, how on earth are they, and I'm a bit simple sometimes, so I was thinking, how on earth are they doing some of these displays where pictures are up in the sky, et cetera, et cetera, and then not long ago, I watched a documentary on TV about how they've set these massive displays which are all drones, you know, talking to each other, distance apart, et cetera, et cetera. Fascinating. And yeah, my word, it's clever stuff. And is that, I guess, back on topic, is that as sustainable and ethical use of data and AI, just so we can see some fantastic, what's wrong with a good old-fashioned firework? Yeah. You know, it's clever stuff, but it's not, yeah. There's lots, I think there's lots of, I think some, there's some real value ads potential uses for all of this, so, yeah, they train doctors who AI tools to do really complicated and delicate surgeries, and they can have multiple goes at it. And yeah, the worst that happens is they press reset, and that to me is quite a comforting thought, yeah, yeah, yeah, the life to know that, yeah, and they can scan you, they can actually rebuild your body in a virtual world, et cetera, et cetera, et cetera, so maybe some of these AI films, these sci-fi films have a benefit. They are being realised for good, I think, to some of them, yeah, sci-fi films always come in a bit of terror about them, they're known as well, so, yeah, yeah, and in fact, they've been made, and we're now starting to see some of these things come to life, makes you revert back to the negative effects of some of those films, but you're absolutely right, you know, this is, again, back to similar topic of using data and AI for good and for a valuable reason, the right reasons, and, you know, housing all the data in the computer, in the right way as well, so, agreed, well, let's call it, and let's move on to the next podcast where we're going to talk about the actual data itself, and some interesting statistics as to, do we actually need what we've got and what we're paying for, anyway, thanks for, thanks for checking out the podcast for those that jumped on, and look forward to catching up with you in the next one, yeah, thanks all, thank you, thanks for listening to the Data Edit, if you enjoyed today's conversation, be sure to follow the show and share it with your network, you'll find more resources, insights, and upcoming episodes at agile.co.uk [Music]