WEBVTT
00:00:00.320 --> 00:00:06.080
The data center industry is moving faster than ever, and behind the scenes some seriously smart people are making it all happen.
00:00:06.240 --> 00:00:13.039
Welcome to the Inside Data Centre Podcast, where we dive into the stories, insights, and innovations driving real change across the industry.
00:00:13.279 --> 00:00:19.920
From hyperscale campuses to game-changing technologies, we're talking to the people shaping the future of digital infrastructure.
00:00:20.160 --> 00:00:25.839
Whether you live in Brieve Data Centres or just want to understand what's powering our connected world, you're in the right place.
00:00:26.320 --> 00:00:27.920
Welcome to the Inside Data Centre Podcast.
00:00:28.000 --> 00:00:35.679
We're back again with another edition from the Accelerated Compute Show in London, joined by my good friend and co-host Rory Flashman Wells.
00:00:36.159 --> 00:00:49.759
We watched the keynote this morning, so we managed to secure a couple of the uh the panelists to talk about a bit more detail around AI infrastructure and specifically within the UK and kind of what can we do to be a competitive player in the market.
00:00:49.920 --> 00:00:51.840
So I'll let my guests introduce themselves.
00:00:52.240 --> 00:00:52.719
Thank you very much.
00:00:52.799 --> 00:00:54.479
Um is Nick Reid.
00:00:54.560 --> 00:00:57.439
I am uh the public policy lead for Coreweave.
00:00:57.600 --> 00:01:01.119
Um Core Weave is an AI cloud computing provider.
00:01:01.280 --> 00:01:15.280
Uh we have a technology platform which uh helps train, develop, deploy AI workloads vertically integrated with a range of data centres, predominantly in the US but increasingly in Europe, including the UK, and I cover all of our government engagement outside of the US.
00:01:15.519 --> 00:01:17.040
My name is Adam Neversol.
00:01:17.200 --> 00:01:21.120
I am the SVP of sales and marketing at KO Data.
00:01:21.359 --> 00:01:31.120
KO is a uh UK-based um developer, uh builder, operator of data centers uh engineered for AI.
00:01:31.359 --> 00:01:31.920
Thank you, Adam.
00:01:32.000 --> 00:01:33.040
Thank you, Nick, for being here.
00:01:33.120 --> 00:01:39.920
I am not going to introduce myself, we're just gonna get straight into it and uh and and probably ask you, Adam, the most difficult question yet.
00:01:40.159 --> 00:01:41.040
Sovereign AI.
00:01:41.120 --> 00:01:44.560
It's a big theme for the event, or sovereign AI, sovereign compute.
00:01:44.799 --> 00:01:48.400
How do we define that and where are the challenges and opportunities?
00:01:48.640 --> 00:01:50.480
And great to get that question first, isn't it?
00:01:50.560 --> 00:01:51.519
I mean that's marvellous.
00:01:51.760 --> 00:02:01.840
Uh so I would say uh probably somewhat of an obvious statement, but um to me, sovereign AI obviously means lots of different things to lots of different people.
00:02:02.000 --> 00:02:23.199
Uh and I think um probably it's maybe maybe we're the only ones in the in the world that have this kind of debate because we're British, but it tends to kind of like go back down to this kind of like almost like a commodity in a supermarket kind of angle, whereby unless it's grown in Britain, canned in Britain, made in Britain, eaten in Britain like a can of beans, you know, it's not sovereign AI.
00:02:23.360 --> 00:02:24.719
I don't necessarily believe that.
00:02:24.800 --> 00:02:27.840
I think it's a bit more multifarious and complex to that.
00:02:28.080 --> 00:02:34.960
To me, sovereign AI means um that we have the ability to act independently.
00:02:35.120 --> 00:02:37.120
Uh, I think that's the key element to it.
00:02:37.280 --> 00:02:54.560
Uh, we don't necessarily need to own uh each of the layers of the stack, we might not own any of the layers of the stack, but we have to basically be at a situation where we are either not reliant on somebody else, uh we can act independently, and we've got trusted access to the layers of compute that we need to.
00:02:54.800 --> 00:02:58.479
I think for us for that to me is sovereign AI.
00:02:58.800 --> 00:03:01.840
Yeah, define sovereignty in a couple of minutes.
00:03:02.000 --> 00:03:02.960
Basically, it's impossible.
00:03:03.039 --> 00:03:20.639
I spend 90% of my time talking to, particularly to policymakers around sovereignty, and I've yet to find two of them who probably agree on what it what sovereignty actually is, um, which I think poses a lot of challenges to this, to the debate.
00:03:20.800 --> 00:03:26.080
Um, and I think Adam's absolutely right, like sovereignty means different things to different people.
00:03:26.319 --> 00:03:40.560
Our from our perspective, sovereignty is very much it's about control, um, and we look at it through the lens of a user, specifically think about AI workloads, what does the user want in terms of sovereignty?
00:03:40.639 --> 00:04:07.680
And from our perspective, that user needs to have a range of options, a range of services technologies that is capable of using to deliver whatever it is they define as sovereignty and has the ability to uh has the optionality, the interoperability to select a number of those services or move between any one without friction and is able to control the data, control the workloads that they are running.
00:04:07.919 --> 00:04:10.800
I'll go off piste as always, but sovereignty as well.
00:04:11.039 --> 00:04:16.319
When your point resonated with me, Adam, around that like British made, you know, it's a very British thing, isn't it?
00:04:16.399 --> 00:04:20.560
Like, you know, buying British eggs from the supermarket, you know, it's like it is very British.
00:04:20.800 --> 00:04:28.560
So do you think, like both of you, that uh having that definition of sovereignty or having sovereign AI is important to the community as well?
00:04:28.639 --> 00:04:39.920
Because when you look at that conversation around the British, you know, becoming an AI power and community kickback, is it important to the consumer that it's British made?
00:04:40.079 --> 00:04:49.759
That they can understand that this is not um supporting industry elsewhere, it's actually supporting industry in the UK because it's British-made AI.
00:04:49.839 --> 00:04:50.959
That's the term I'm going to use.
00:04:51.199 --> 00:04:54.160
I think for some people it absolutely will be important.
00:04:54.319 --> 00:05:00.639
They will want to use and develop AI that's been made and um brought to market in the UK.
00:05:00.800 --> 00:05:08.480
For other people, that won't be necessarily as applicable, so which is why I think there needs to be a range of uh options on the table.
00:05:08.560 --> 00:05:13.439
But I do think there needs to be more AI made, built, and grown in Britain.
00:05:13.600 --> 00:05:28.319
I think we sort of we are certainly, so I'm sure we'll get into some of the reasons for that, we are certainly behind in terms of the uh development of AI in the UK and that ability, as Adam says, to like control a certain part of the AI stack.
00:05:28.639 --> 00:05:30.319
Uh it's a really good point, Andy.
00:05:30.639 --> 00:05:47.920
Um this reminds me of uh a discussion I had with one of my relatives the other day, which he he got quite annoyed about the fact that things weren't built in Britain and therefore that artificial intelligence wasn't British, and so he then typed into Chat GPT why is you know why can't we have British AI?
00:05:48.000 --> 00:05:52.319
And the uh the irony of watching him do that was uh not lost to me at the time, to be perfectly honest with you.
00:05:52.480 --> 00:06:00.399
Uh I think it it again uh to echo what Nick said, it will depend largely on the application or the services being used.
00:06:00.560 --> 00:06:05.920
I think people are quite happy that that that we use lots of stuff which has originated from other countries, you know.
00:06:06.079 --> 00:06:19.199
I mean we've got a culture whereby our our power stations are owned by France and China, our national grid is owned by China, uh most of the train services that we run on are owned by Germany or Japan or whatever.
00:06:19.279 --> 00:06:33.040
So I think in our while while we get quite passionate about it, I think actually British people are quite happy to use services owned by other nations or companies in other nations, so long as they work and as long as they're good.
00:06:33.120 --> 00:06:35.759
You know, we tend to get we less worried about them when they're working.
00:06:35.920 --> 00:06:49.600
But I think things like, for example, NHS data, I think majority of people in the UK would want to make sure that stays in the UK and is be and is being managed in the UK without it having a little sticker on it saying buy British.
00:06:49.759 --> 00:06:50.879
I think that would probably be quite key.
00:06:51.040 --> 00:06:51.279
Rory?
00:06:51.600 --> 00:06:52.720
Yeah, it's an interesting concept.
00:06:52.800 --> 00:07:02.480
I mean, to your point about things being made in Britain, you've got this big made in America campaign happening right now, and that's changing the dynamics of supply chains and all sorts of things happening there.
00:07:02.639 --> 00:07:12.319
So you can certainly see the arguments for, and maybe yeah, to your point, Adam, arguments against the whether or not a sovereign there raises another question, like whether or not sovereignty can really be achieved.
00:07:12.560 --> 00:07:15.279
On that point, we are 20 months.
00:07:15.360 --> 00:07:26.480
We talked about the AI Opportunities Action Plan earlier, 20 months post-AI Opportunities Action Plan, and the topic of frontier AI has been kind of a theme on the podcast today.
00:07:26.800 --> 00:07:29.759
From your perspective, as a developer and operator, what are we seeing?
00:07:29.839 --> 00:07:32.000
Is it demand for training and inference here?
00:07:32.160 --> 00:07:36.160
And I suppose the big question is what needs to change?
00:07:36.399 --> 00:07:48.079
I would say from as a UK operator, um, we are seeing, in terms of the uh international interest that we have in the UK, that it's uh a lot more inference-based than training based.
00:07:48.240 --> 00:08:00.959
And I think that's for very logical reasons, which is that if you are a loud, large cloud provider that are doing generative AI workloads, you know, very heavy, intensive workloads, why would you put that in the UK at the moment?
00:08:01.120 --> 00:08:08.000
Um, you've got energy costs which are four times America, double France, I think eight times Norway.
00:08:08.160 --> 00:08:15.759
So are you willing to pay eight times the power cost of Norway to train uh workloads which don't need to be close to the eyeballs?
00:08:16.000 --> 00:08:16.800
It makes no sense.
00:08:16.959 --> 00:08:21.439
And at the same time, there are complications around AI copyright in the UK.
00:08:21.519 --> 00:08:31.920
Um that's an area which I'm not it's not my specialised subject, but it from what I'm told uh there are error there are issues there if you're wanting to scrape data in the UK, which you can't, which you could do in the EU.
00:08:32.240 --> 00:08:53.120
So, with those things in in part, most of the compute that we're seeing is essentially the inference side of it, which I think is a shame for the UK because as an operator, uh the money in data centers to me is in the training element to it, because that's the high-intensity industrial scale data centers which comes with all the construction jobs, comes of all the supply chain and everything else.
00:08:53.360 --> 00:09:00.159
And it means that as a country, uh we're in the situation whereby we have got some of the brightest talents and brains in the world.
00:09:00.240 --> 00:09:18.320
Uh, we've got the academic and research talent, the ideas, the AI invention, but that will eventually be drip-fed, I think, to countries like the Middle on the Middle East region, Singapore, um, other nations, which I think is a is um an own goal for the UK.
00:09:18.559 --> 00:09:20.799
Yeah, completely agree with all of that.
00:09:21.039 --> 00:09:26.879
Um we certainly don't see any um large-scale training taking place in the UK.
00:09:26.960 --> 00:09:50.080
There are some, so there's COSIN who are training a uh uh an LLM on the Isambard computer in Bristol, which you know has potential, but in terms of the actual GPU hours that they are able to deploy, it's minimal compared to your um Frontier, you know, your your Chat GPT and your um Claude and whatnot, it's just it's not in the same ballpark.
00:09:50.159 --> 00:09:56.159
Um, and for the reasons Adam you know um neatly articulates, I can't see that changing.
00:09:56.399 --> 00:09:59.679
We don't have the compute clusters to do it.
00:09:59.919 --> 00:10:01.919
If we did, it would be too expensive.
00:10:02.000 --> 00:10:17.919
And the startups and scale-ups in the AI world, when they have sort of fairly limited runways in terms of their funding, are not going to be paying the premium that it would take to train in the UK when they can get it for you know significantly less if they were using GPs than the Nordics.
00:10:18.000 --> 00:10:24.720
So at the moment, um it's definitely inference-based in terms of the services that are supplied here.
00:10:24.960 --> 00:10:34.320
Government have, to be fair, set out to make AI here, but they haven't as yet pulled the levers that will be needed in order to make that happen.
00:10:34.639 --> 00:10:41.039
Yeah, we talked earlier about the importance of jobs, has become a key component of the data centre story now, hasn't it?
00:10:41.120 --> 00:10:56.879
Like communities want to see employment, but you also, as I said earlier, you want to retain those skills in the UK because a lot of what we're going to end up doing if we're not careful is training the skills here, and then the skills are gonna be in demand elsewhere, and we're gonna lose the skills that we trained, which would be a very, very British thing to do.
00:10:56.960 --> 00:11:03.840
But it is, I think, really important that it's not just looked at as keeping the you know the industry here, it's the skills here as well.
00:11:04.000 --> 00:11:05.600
Like, well, you know, these people need to work.
00:11:05.759 --> 00:11:10.639
I wanted to touch on Scotland because I know you're obviously involved in a in a development in that region.
00:11:10.720 --> 00:11:18.559
It's probably when I do my weekly newsletter, I probably see about 20 different articles about community uh engagement in in Scotland.
00:11:18.639 --> 00:11:26.399
So I just wanted to see what from your perspective, that as a region is a strong region for what we're talking about, for AI training.
00:11:26.480 --> 00:11:30.080
Um, but is it feasible that it will ever become a competitive market?
00:11:30.320 --> 00:11:45.360
Yeah, I think Scotland has a very unique opportunity to be you know uh a sort of an industrial leader in AI, it has pretty much everything you would you would want in terms of being able to deploy a data centre and to run AI workloads.
00:11:45.440 --> 00:11:54.799
It has, in theory, the most accessible, abundant, affordable sources of renewable energy, which are incredibly important to us.
00:11:54.960 --> 00:12:16.240
The issue in part is because the way the UK grid is set up, that the sector cannot take full advantage of that, and the fact you are having sort of wind farms in Scotland being paid not to run, whereas you could, in theory, put a data centre right next to them and take the power power down at a sort of relatively discounted rate, run AI workloads for the for the benefit of the UK as a whole.
00:12:16.399 --> 00:12:28.320
I think Scotland has yeah, massive potential, and um you know the cooler climate again, the from an academic point of view, the University of Edinburgh, Glasgow, Strathclyde, there's a huge amount going on up there.
00:12:28.480 --> 00:12:37.759
Like Scotland has an incredible history in terms of like innovation, and so and I I see that as incredibly attractive for a you know, vibrant startup and scale-up scene.
00:12:37.840 --> 00:12:43.600
We work with a company called Cobase to try and support um a number of the startups and scale-ups that are operating there.
00:12:43.759 --> 00:12:57.440
So that I think the few there is a lot of positivity up there, but again, it this is one of those things that needs a whole UK approach, and Scotland is not going to be able to operate just as an island within that, and so there is more I think that needs to be done to help us realise that potential.
00:12:57.600 --> 00:13:03.919
But yeah, for sure, it's there, um, and you know, which is why we're developing a site up there, and we're obviously keen to do more.
00:13:04.159 --> 00:13:07.039
Uh yeah, I agree with everything that Nick said, uh Andy.
00:13:07.279 --> 00:13:25.120
Um I think that I think the situation in Scotland uh is utterly nonsensical and and is just crazy, and it's it's almost um a perfect example of how ludicrous our strategy around compute in the UK has got.
00:13:25.519 --> 00:13:33.360
Um, you know, I as I know I took my family up there uh last year for a thoroughly miserable week of rain.
00:13:33.519 --> 00:13:38.639
Uh Scotland is notoriously wet and is notoriously cold.
00:13:38.879 --> 00:13:41.840
Um it has a very similar climate to Norway.
00:13:42.000 --> 00:13:50.240
It currently has about 32 to 35 megawatts of compute, which is under utilisation in Scotland, and they're talking about having a moratorium on data centres.
00:13:50.799 --> 00:14:09.519
There is less compute in Scotland than there is in Harlow in Essex, and yet with the amount of potential that Scotland has got to be um a real engine room for compute in the UK and in Europe, they want to pull the plug on it uh because they're worried that data centres will drain Loch Lomond.
00:14:09.919 --> 00:14:13.200
I mean, it's it's I almost can't get my head around it.
00:14:13.360 --> 00:14:25.279
Um we've now got to a situation in the UK where we're looking at uh there was a data centre development which is um muted for Bedfordshire, which is 1.8 gigawatts of gas uh fed uh compute.
00:14:25.440 --> 00:14:33.120
So in the middle of a climate emergency, we're contemplating building 1.8 gigawatts of gas-fired fossil fuel compute.
00:14:33.360 --> 00:14:43.200
When, as Nick said, if you put that compute up into Scotland, uh one, it would reduce the curtailment fees that we all pay uh to turn off wind turbines.
00:14:43.360 --> 00:14:47.360
It would be using 100% green electrons, so it'd be great for the environment.
00:14:47.519 --> 00:14:58.240
The inward investment that would come into that from having an enormous data centre up there would be phenomenal, and you've got the perfect climate uh to actually run that data centre as they as they do in Norway.
00:14:58.399 --> 00:15:02.559
I mean, if it can be done in Norway, uh I've no idea what you can't do in Scotland.
00:15:02.799 --> 00:15:16.159
Yeah, it's an interesting conversation, right, because well one of the big challenges at the moment is misinformation, huge amounts of let's say not necessarily misinformation, but misunderstanding in terms of the general public, the role of data centres, what AI will do for the economy.
00:15:16.240 --> 00:15:24.159
And I think a lot of people are beginning to be scared of AI and don't really understand how it could or would benefit them.
00:15:24.399 --> 00:15:25.679
I'm interested from you both.
00:15:25.759 --> 00:15:27.120
This is super left field.
00:15:27.440 --> 00:15:39.919
What kind of messages are you trying to uh convey, certainly to the general public, and more broadly, like Nick, from your perspective as well, like how do you approach that in Scotland where there is so much opposition to data centres?
00:15:40.159 --> 00:15:41.759
There's no quick fix for it, Rory.
00:15:42.240 --> 00:15:52.960
Um I think how how we are approaching it is by being open, transparent, and uh zoning in on the facts, what is actually accurate facts.
00:15:53.120 --> 00:16:01.039
There's a real need for data centers to be explained uh to so that the society can understand really what it what it entails.
00:16:01.279 --> 00:16:14.080
Um at the end of the day, you know, we we as an industry, uh one, we haven't been great at being transparent ourselves, you know, but we're an industry that's based on secrecy, uh, has always been the case.
00:16:14.240 --> 00:16:16.799
So trying to get rid of that mentality, I think, is the first thing.
00:16:16.960 --> 00:16:22.320
The second thing is we're not great at uh being transparent on the facts of in terms of how we operate.
00:16:22.399 --> 00:16:24.799
Um that's something we we need to get better at.
00:16:25.039 --> 00:16:30.799
And the other issue is that we need to get away from the idea that the public don't understand what data centers are.
00:16:30.960 --> 00:16:36.480
I mean, I've spoken to so many residents around data centers, they they do understand that the data center underpins what they do.
00:16:36.559 --> 00:16:40.879
You know, we don't need to educate them that you're booking a flight on EasyJet, you're using a data center.
00:16:40.960 --> 00:16:41.440
They get that.
00:16:41.519 --> 00:16:43.360
I mean, they're not stupid, they get that.
00:16:43.519 --> 00:16:48.080
The bit I think they don't get is the environmental impact of the data center and how it operates.
00:16:48.159 --> 00:16:56.559
And that's largely because I think they, you know, they read the newspapers, the newspapers are quoting figures from 20 years ago based on data centres that are in Arizona and Nevada.
00:16:56.720 --> 00:17:02.000
Um, we're operating um our new data centre is a closed loop system, for example, on water.
00:17:02.159 --> 00:17:11.680
We're using the same amount of water as one family of four users in one year uh across 20 years of operation for a data centre that powers 32,000 GPUs.
00:17:11.839 --> 00:17:13.839
So the water element there is negligible.
00:17:14.000 --> 00:17:25.039
So if you if you're building that type of infrastructure and at the same time you're being told that you might drain Loch Lomond of water, it there's a real juxtaposition there.
00:17:25.200 --> 00:17:32.160
So I think it's just by a case-by-case basis, taking each fact as it comes and trying to provide the accuracy on it.
00:17:32.400 --> 00:17:34.960
Yeah, I think I'm again agree completely with that.
00:17:35.039 --> 00:17:48.559
I think as an industry we haven't done a particularly good job in terms of like tackling the misinformation out is that is out there and articulating the value story around both data centers and in our specific case AI data centers.
00:17:48.960 --> 00:18:02.480
Um it is a challenge because these things are very complex structures and systems, and to and so to like counterbalance relatively simple arguments like use a lot of energy, use a lot of water makes noise, those are quite simple.
00:18:02.559 --> 00:18:06.240
And as Anna says, there's a load of like just really bad data out there.
00:18:06.480 --> 00:18:29.519
Um from our perspective, that community engagement piece at the very local level is absolutely key, and also understanding that the communities that we operate in, they're all very different, whether they're in the US or in Europe, you know, communities in Finland have different expectations of what a data center company would offer versus Norway or Sweden or the UK.
00:18:29.680 --> 00:18:59.680
So it's about having that kind of like on the ground, being being part of the community on the ground, showing like an actual human face, and then working out like how do we contribute viably to the local economies and making sure that these data centers are sustainable, not in terms of an environmental perspective, but a sustainable and long-term part of these communities, and so that's something we have to take very seriously and is a you know critical part of the engagement we do very much at the at the local level.
00:18:59.920 --> 00:19:02.400
Yeah, I think I've I've been on a bit of a journey with this myself.
00:19:02.480 --> 00:19:06.640
I think you you kind of look at it as around you need to share the good news, but it's not true, is it?
00:19:06.799 --> 00:19:07.759
You need to share the truth.
00:19:07.839 --> 00:19:13.519
Like I mean, you can look at it like, oh, there's some positive stories about the industry, and don't get me wrong, there's some amazing positive stories.
00:19:13.599 --> 00:19:19.279
Like we were talking about the individual positive stories about people that have developed their career and gone on to be really successful.
00:19:19.440 --> 00:19:30.079
There's a positive story about companies that have you know totally transformed because of the data center industry, but when it comes to the data behind it, I no pun intended, um, it's being honest, isn't it?
00:19:30.160 --> 00:19:44.400
Like this is how much water it uses, this is how much power it uses, this is how big it is, you know, and and this is how large it will become, not might become, like I always talk about the you know, it's not a one gigawatt data center when only 20 megawatts have been approved.
00:19:44.480 --> 00:19:46.079
It's it potentially could be.
00:19:46.160 --> 00:19:50.000
So I think how do we get the truth out there is is the way I'm looking at it now.
00:19:50.079 --> 00:19:52.799
Like, what can I do personally to share more true stories?
00:19:52.880 --> 00:19:54.079
I think that's really important.
00:19:54.160 --> 00:20:13.519
But it's it's a big challenge because you are now competing with the other side of the debate, which is around uh we don't like data centres, we don't want data centres, and I'm finding it more and more difficult to find none, I don't want to say negative, but stories from that side of the debate when I'm doing my newsletter.
00:20:13.680 --> 00:20:14.720
It's harder and harder.
00:20:14.799 --> 00:20:27.359
If you search data centre, or you at home, or search data centre, the first two pages on Google, I guarantee you will include at least 50% of articles from media around negative impact of data centres.
00:20:27.440 --> 00:20:29.920
So it's a it's a big noise we're we're fighting again.
00:20:30.079 --> 00:20:31.119
Yeah, I couldn't agree more.
00:20:31.279 --> 00:20:37.200
I mean, I think the the there are so many, many challenges um to your point about the narrative.
00:20:37.440 --> 00:20:43.759
I I don't think it's the job of the general public to have to understand what we do.
00:20:43.839 --> 00:20:46.079
I think the industry needs to do a better job of communicating.
00:20:46.160 --> 00:20:47.920
I think it needs multiple initiatives.
00:20:48.000 --> 00:20:51.680
That idea of explaining data centers, data centers explained is a good one, right?
00:20:51.759 --> 00:20:52.720
And it's really important.
00:20:52.880 --> 00:21:04.000
But also I think like you need to um, as an industry, you need to appeal to Joe Bloggs on the street and you need to make sense to, you know, like the plumbers, electricians, what will AI do for them, right?
00:21:04.160 --> 00:21:08.640
People who don't really understand this technology and work with their hands all the day, how can they get a career?
00:21:08.720 --> 00:21:10.559
Um, how can they improve their lives?
00:21:10.720 --> 00:21:19.839
How could they go from working underneath kitchen sinks or in cupboards to working on large-scale industrial sites and improving the future for not only them but for the people that are around them?
00:21:19.920 --> 00:21:23.200
I think there's there's so many other positive stories and different angles we could take.
00:21:23.279 --> 00:21:26.559
Um, but it's multifaceted and increasingly complex.
00:21:26.720 --> 00:21:28.720
So uh yeah, more to come.
00:21:28.799 --> 00:21:30.640
So look, guys, we are at time.
00:21:30.880 --> 00:21:33.359
Appreciate you guys taking the time to speak with us today.
00:21:33.440 --> 00:21:36.160
And I'm sure we could sit around talking about this all day long.
00:21:36.240 --> 00:21:41.119
So maybe we do a follow-up and uh and get into kind of the press side of it in the future.
00:21:41.359 --> 00:21:42.160
Cheers, guys.
00:21:42.400 --> 00:21:44.960
That's it for this episode of the Inside Data Centre Podcast.
00:21:45.119 --> 00:21:46.079
Thanks for tuning in.
00:21:46.240 --> 00:21:54.079
This show is brought to you by DataX and Data Exec, the recruitment partners powering global talent across the data centre and digital infrastructure world.
00:21:54.400 --> 00:21:58.240
If you enjoyed today's conversations, don't forget to follow, like, and subscribe.
00:21:58.400 --> 00:22:00.720
It really helps us grow the podcast community.
00:22:01.119 --> 00:22:06.000
Stay tuned for more stories, insights, and conversations with the people shaping the future of our industry.
00:00:00.320 --> 00:00:06.080
The data center industry is moving faster than ever, and behind the scenes some seriously smart people are making it all happen.
00:00:06.240 --> 00:00:13.039
Welcome to the Inside Data Centre Podcast, where we dive into the stories, insights, and innovations driving real change across the industry.
00:00:13.279 --> 00:00:19.920
From hyperscale campuses to game-changing technologies, we're talking to the people shaping the future of digital infrastructure.
00:00:20.160 --> 00:00:25.839
Whether you live in Brieve Data Centres or just want to understand what's powering our connected world, you're in the right place.
00:00:26.320 --> 00:00:27.920
Welcome to the Inside Data Centre Podcast.
00:00:28.000 --> 00:00:35.679
We're back again with another edition from the Accelerated Compute Show in London, joined by my good friend and co-host Rory Flashman Wells.
00:00:36.159 --> 00:00:49.759
We watched the keynote this morning, so we managed to secure a couple of the uh the panelists to talk about a bit more detail around AI infrastructure and specifically within the UK and kind of what can we do to be a competitive player in the market.
00:00:49.920 --> 00:00:51.840
So I'll let my guests introduce themselves.
00:00:52.240 --> 00:00:52.719
Thank you very much.
00:00:52.799 --> 00:00:54.479
Um is Nick Reid.
00:00:54.560 --> 00:00:57.439
I am uh the public policy lead for Coreweave.
00:00:57.600 --> 00:01:01.119
Um Core Weave is an AI cloud computing provider.
00:01:01.280 --> 00:01:15.280
Uh we have a technology platform which uh helps train, develop, deploy AI workloads vertically integrated with a range of data centres, predominantly in the US but increasingly in Europe, including the UK, and I cover all of our government engagement outside of the US.
00:01:15.519 --> 00:01:17.040
My name is Adam Neversol.
00:01:17.200 --> 00:01:21.120
I am the SVP of sales and marketing at KO Data.
00:01:21.359 --> 00:01:31.120
KO is a uh UK-based um developer, uh builder, operator of data centers uh engineered for AI.
00:01:31.359 --> 00:01:31.920
Thank you, Adam.
00:01:32.000 --> 00:01:33.040
Thank you, Nick, for being here.
00:01:33.120 --> 00:01:39.920
I am not going to introduce myself, we're just gonna get straight into it and uh and and probably ask you, Adam, the most difficult question yet.
00:01:40.159 --> 00:01:41.040
Sovereign AI.
00:01:41.120 --> 00:01:44.560
It's a big theme for the event, or sovereign AI, sovereign compute.
00:01:44.799 --> 00:01:48.400
How do we define that and where are the challenges and opportunities?
00:01:48.640 --> 00:01:50.480
And great to get that question first, isn't it?
00:01:50.560 --> 00:01:51.519
I mean that's marvellous.
00:01:51.760 --> 00:02:01.840
Uh so I would say uh probably somewhat of an obvious statement, but um to me, sovereign AI obviously means lots of different things to lots of different people.
00:02:02.000 --> 00:02:23.199
Uh and I think um probably it's maybe maybe we're the only ones in the in the world that have this kind of debate because we're British, but it tends to kind of like go back down to this kind of like almost like a commodity in a supermarket kind of angle, whereby unless it's grown in Britain, canned in Britain, made in Britain, eaten in Britain like a can of beans, you know, it's not sovereign AI.
00:02:23.360 --> 00:02:24.719
I don't necessarily believe that.
00:02:24.800 --> 00:02:27.840
I think it's a bit more multifarious and complex to that.
00:02:28.080 --> 00:02:34.960
To me, sovereign AI means um that we have the ability to act independently.
00:02:35.120 --> 00:02:37.120
Uh, I think that's the key element to it.
00:02:37.280 --> 00:02:54.560
Uh, we don't necessarily need to own uh each of the layers of the stack, we might not own any of the layers of the stack, but we have to basically be at a situation where we are either not reliant on somebody else, uh we can act independently, and we've got trusted access to the layers of compute that we need to.
00:02:54.800 --> 00:02:58.479
I think for us for that to me is sovereign AI.
00:02:58.800 --> 00:03:01.840
Yeah, define sovereignty in a couple of minutes.
00:03:02.000 --> 00:03:02.960
Basically, it's impossible.
00:03:03.039 --> 00:03:20.639
I spend 90% of my time talking to, particularly to policymakers around sovereignty, and I've yet to find two of them who probably agree on what it what sovereignty actually is, um, which I think poses a lot of challenges to this, to the debate.
00:03:20.800 --> 00:03:26.080
Um, and I think Adam's absolutely right, like sovereignty means different things to different people.
00:03:26.319 --> 00:03:40.560
Our from our perspective, sovereignty is very much it's about control, um, and we look at it through the lens of a user, specifically think about AI workloads, what does the user want in terms of sovereignty?
00:03:40.639 --> 00:04:07.680
And from our perspective, that user needs to have a range of options, a range of services technologies that is capable of using to deliver whatever it is they define as sovereignty and has the ability to uh has the optionality, the interoperability to select a number of those services or move between any one without friction and is able to control the data, control the workloads that they are running.
00:04:07.919 --> 00:04:10.800
I'll go off piste as always, but sovereignty as well.
00:04:11.039 --> 00:04:16.319
When your point resonated with me, Adam, around that like British made, you know, it's a very British thing, isn't it?
00:04:16.399 --> 00:04:20.560
Like, you know, buying British eggs from the supermarket, you know, it's like it is very British.
00:04:20.800 --> 00:04:28.560
So do you think, like both of you, that uh having that definition of sovereignty or having sovereign AI is important to the community as well?
00:04:28.639 --> 00:04:39.920
Because when you look at that conversation around the British, you know, becoming an AI power and community kickback, is it important to the consumer that it's British made?
00:04:40.079 --> 00:04:49.759
That they can understand that this is not um supporting industry elsewhere, it's actually supporting industry in the UK because it's British-made AI.
00:04:49.839 --> 00:04:50.959
That's the term I'm going to use.
00:04:51.199 --> 00:04:54.160
I think for some people it absolutely will be important.
00:04:54.319 --> 00:05:00.639
They will want to use and develop AI that's been made and um brought to market in the UK.
00:05:00.800 --> 00:05:08.480
For other people, that won't be necessarily as applicable, so which is why I think there needs to be a range of uh options on the table.
00:05:08.560 --> 00:05:13.439
But I do think there needs to be more AI made, built, and grown in Britain.
00:05:13.600 --> 00:05:28.319
I think we sort of we are certainly, so I'm sure we'll get into some of the reasons for that, we are certainly behind in terms of the uh development of AI in the UK and that ability, as Adam says, to like control a certain part of the AI stack.
00:05:28.639 --> 00:05:30.319
Uh it's a really good point, Andy.
00:05:30.639 --> 00:05:47.920
Um this reminds me of uh a discussion I had with one of my relatives the other day, which he he got quite annoyed about the fact that things weren't built in Britain and therefore that artificial intelligence wasn't British, and so he then typed into Chat GPT why is you know why can't we have British AI?
00:05:48.000 --> 00:05:52.319
And the uh the irony of watching him do that was uh not lost to me at the time, to be perfectly honest with you.
00:05:52.480 --> 00:06:00.399
Uh I think it it again uh to echo what Nick said, it will depend largely on the application or the services being used.
00:06:00.560 --> 00:06:05.920
I think people are quite happy that that that we use lots of stuff which has originated from other countries, you know.
00:06:06.079 --> 00:06:19.199
I mean we've got a culture whereby our our power stations are owned by France and China, our national grid is owned by China, uh most of the train services that we run on are owned by Germany or Japan or whatever.
00:06:19.279 --> 00:06:33.040
So I think in our while while we get quite passionate about it, I think actually British people are quite happy to use services owned by other nations or companies in other nations, so long as they work and as long as they're good.
00:06:33.120 --> 00:06:35.759
You know, we tend to get we less worried about them when they're working.
00:06:35.920 --> 00:06:49.600
But I think things like, for example, NHS data, I think majority of people in the UK would want to make sure that stays in the UK and is be and is being managed in the UK without it having a little sticker on it saying buy British.
00:06:49.759 --> 00:06:50.879
I think that would probably be quite key.
00:06:51.040 --> 00:06:51.279
Rory?
00:06:51.600 --> 00:06:52.720
Yeah, it's an interesting concept.
00:06:52.800 --> 00:07:02.480
I mean, to your point about things being made in Britain, you've got this big made in America campaign happening right now, and that's changing the dynamics of supply chains and all sorts of things happening there.
00:07:02.639 --> 00:07:12.319
So you can certainly see the arguments for, and maybe yeah, to your point, Adam, arguments against the whether or not a sovereign there raises another question, like whether or not sovereignty can really be achieved.
00:07:12.560 --> 00:07:15.279
On that point, we are 20 months.
00:07:15.360 --> 00:07:26.480
We talked about the AI Opportunities Action Plan earlier, 20 months post-AI Opportunities Action Plan, and the topic of frontier AI has been kind of a theme on the podcast today.
00:07:26.800 --> 00:07:29.759
From your perspective, as a developer and operator, what are we seeing?
00:07:29.839 --> 00:07:32.000
Is it demand for training and inference here?
00:07:32.160 --> 00:07:36.160
And I suppose the big question is what needs to change?
00:07:36.399 --> 00:07:48.079
I would say from as a UK operator, um, we are seeing, in terms of the uh international interest that we have in the UK, that it's uh a lot more inference-based than training based.
00:07:48.240 --> 00:08:00.959
And I think that's for very logical reasons, which is that if you are a loud, large cloud provider that are doing generative AI workloads, you know, very heavy, intensive workloads, why would you put that in the UK at the moment?
00:08:01.120 --> 00:08:08.000
Um, you've got energy costs which are four times America, double France, I think eight times Norway.
00:08:08.160 --> 00:08:15.759
So are you willing to pay eight times the power cost of Norway to train uh workloads which don't need to be close to the eyeballs?
00:08:16.000 --> 00:08:16.800
It makes no sense.
00:08:16.959 --> 00:08:21.439
And at the same time, there are complications around AI copyright in the UK.
00:08:21.519 --> 00:08:31.920
Um that's an area which I'm not it's not my specialised subject, but it from what I'm told uh there are error there are issues there if you're wanting to scrape data in the UK, which you can't, which you could do in the EU.
00:08:32.240 --> 00:08:53.120
So, with those things in in part, most of the compute that we're seeing is essentially the inference side of it, which I think is a shame for the UK because as an operator, uh the money in data centers to me is in the training element to it, because that's the high-intensity industrial scale data centers which comes with all the construction jobs, comes of all the supply chain and everything else.
00:08:53.360 --> 00:09:00.159
And it means that as a country, uh we're in the situation whereby we have got some of the brightest talents and brains in the world.
00:09:00.240 --> 00:09:18.320
Uh, we've got the academic and research talent, the ideas, the AI invention, but that will eventually be drip-fed, I think, to countries like the Middle on the Middle East region, Singapore, um, other nations, which I think is a is um an own goal for the UK.
00:09:18.559 --> 00:09:20.799
Yeah, completely agree with all of that.
00:09:21.039 --> 00:09:26.879
Um we certainly don't see any um large-scale training taking place in the UK.
00:09:26.960 --> 00:09:50.080
There are some, so there's COSIN who are training a uh uh an LLM on the Isambard computer in Bristol, which you know has potential, but in terms of the actual GPU hours that they are able to deploy, it's minimal compared to your um Frontier, you know, your your Chat GPT and your um Claude and whatnot, it's just it's not in the same ballpark.
00:09:50.159 --> 00:09:56.159
Um, and for the reasons Adam you know um neatly articulates, I can't see that changing.
00:09:56.399 --> 00:09:59.679
We don't have the compute clusters to do it.
00:09:59.919 --> 00:10:01.919
If we did, it would be too expensive.
00:10:02.000 --> 00:10:17.919
And the startups and scale-ups in the AI world, when they have sort of fairly limited runways in terms of their funding, are not going to be paying the premium that it would take to train in the UK when they can get it for you know significantly less if they were using GPs than the Nordics.
00:10:18.000 --> 00:10:24.720
So at the moment, um it's definitely inference-based in terms of the services that are supplied here.
00:10:24.960 --> 00:10:34.320
Government have, to be fair, set out to make AI here, but they haven't as yet pulled the levers that will be needed in order to make that happen.
00:10:34.639 --> 00:10:41.039
Yeah, we talked earlier about the importance of jobs, has become a key component of the data centre story now, hasn't it?
00:10:41.120 --> 00:10:56.879
Like communities want to see employment, but you also, as I said earlier, you want to retain those skills in the UK because a lot of what we're going to end up doing if we're not careful is training the skills here, and then the skills are gonna be in demand elsewhere, and we're gonna lose the skills that we trained, which would be a very, very British thing to do.
00:10:56.960 --> 00:11:03.840
But it is, I think, really important that it's not just looked at as keeping the you know the industry here, it's the skills here as well.
00:11:04.000 --> 00:11:05.600
Like, well, you know, these people need to work.
00:11:05.759 --> 00:11:10.639
I wanted to touch on Scotland because I know you're obviously involved in a in a development in that region.
00:11:10.720 --> 00:11:18.559
It's probably when I do my weekly newsletter, I probably see about 20 different articles about community uh engagement in in Scotland.
00:11:18.639 --> 00:11:26.399
So I just wanted to see what from your perspective, that as a region is a strong region for what we're talking about, for AI training.
00:11:26.480 --> 00:11:30.080
Um, but is it feasible that it will ever become a competitive market?
00:11:30.320 --> 00:11:45.360
Yeah, I think Scotland has a very unique opportunity to be you know uh a sort of an industrial leader in AI, it has pretty much everything you would you would want in terms of being able to deploy a data centre and to run AI workloads.
00:11:45.440 --> 00:11:54.799
It has, in theory, the most accessible, abundant, affordable sources of renewable energy, which are incredibly important to us.
00:11:54.960 --> 00:12:16.240
The issue in part is because the way the UK grid is set up, that the sector cannot take full advantage of that, and the fact you are having sort of wind farms in Scotland being paid not to run, whereas you could, in theory, put a data centre right next to them and take the power power down at a sort of relatively discounted rate, run AI workloads for the for the benefit of the UK as a whole.
00:12:16.399 --> 00:12:28.320
I think Scotland has yeah, massive potential, and um you know the cooler climate again, the from an academic point of view, the University of Edinburgh, Glasgow, Strathclyde, there's a huge amount going on up there.
00:12:28.480 --> 00:12:37.759
Like Scotland has an incredible history in terms of like innovation, and so and I I see that as incredibly attractive for a you know, vibrant startup and scale-up scene.
00:12:37.840 --> 00:12:43.600
We work with a company called Cobase to try and support um a number of the startups and scale-ups that are operating there.
00:12:43.759 --> 00:12:57.440
So that I think the few there is a lot of positivity up there, but again, it this is one of those things that needs a whole UK approach, and Scotland is not going to be able to operate just as an island within that, and so there is more I think that needs to be done to help us realise that potential.
00:12:57.600 --> 00:13:03.919
But yeah, for sure, it's there, um, and you know, which is why we're developing a site up there, and we're obviously keen to do more.
00:13:04.159 --> 00:13:07.039
Uh yeah, I agree with everything that Nick said, uh Andy.
00:13:07.279 --> 00:13:25.120
Um I think that I think the situation in Scotland uh is utterly nonsensical and and is just crazy, and it's it's almost um a perfect example of how ludicrous our strategy around compute in the UK has got.
00:13:25.519 --> 00:13:33.360
Um, you know, I as I know I took my family up there uh last year for a thoroughly miserable week of rain.
00:13:33.519 --> 00:13:38.639
Uh Scotland is notoriously wet and is notoriously cold.
00:13:38.879 --> 00:13:41.840
Um it has a very similar climate to Norway.
00:13:42.000 --> 00:13:50.240
It currently has about 32 to 35 megawatts of compute, which is under utilisation in Scotland, and they're talking about having a moratorium on data centres.
00:13:50.799 --> 00:14:09.519
There is less compute in Scotland than there is in Harlow in Essex, and yet with the amount of potential that Scotland has got to be um a real engine room for compute in the UK and in Europe, they want to pull the plug on it uh because they're worried that data centres will drain Loch Lomond.
00:14:09.919 --> 00:14:13.200
I mean, it's it's I almost can't get my head around it.
00:14:13.360 --> 00:14:25.279
Um we've now got to a situation in the UK where we're looking at uh there was a data centre development which is um muted for Bedfordshire, which is 1.8 gigawatts of gas uh fed uh compute.
00:14:25.440 --> 00:14:33.120
So in the middle of a climate emergency, we're contemplating building 1.8 gigawatts of gas-fired fossil fuel compute.
00:14:33.360 --> 00:14:43.200
When, as Nick said, if you put that compute up into Scotland, uh one, it would reduce the curtailment fees that we all pay uh to turn off wind turbines.
00:14:43.360 --> 00:14:47.360
It would be using 100% green electrons, so it'd be great for the environment.
00:14:47.519 --> 00:14:58.240
The inward investment that would come into that from having an enormous data centre up there would be phenomenal, and you've got the perfect climate uh to actually run that data centre as they as they do in Norway.
00:14:58.399 --> 00:15:02.559
I mean, if it can be done in Norway, uh I've no idea what you can't do in Scotland.
00:15:02.799 --> 00:15:16.159
Yeah, it's an interesting conversation, right, because well one of the big challenges at the moment is misinformation, huge amounts of let's say not necessarily misinformation, but misunderstanding in terms of the general public, the role of data centres, what AI will do for the economy.
00:15:16.240 --> 00:15:24.159
And I think a lot of people are beginning to be scared of AI and don't really understand how it could or would benefit them.
00:15:24.399 --> 00:15:25.679
I'm interested from you both.
00:15:25.759 --> 00:15:27.120
This is super left field.
00:15:27.440 --> 00:15:39.919
What kind of messages are you trying to uh convey, certainly to the general public, and more broadly, like Nick, from your perspective as well, like how do you approach that in Scotland where there is so much opposition to data centres?
00:15:40.159 --> 00:15:41.759
There's no quick fix for it, Rory.
00:15:42.240 --> 00:15:52.960
Um I think how how we are approaching it is by being open, transparent, and uh zoning in on the facts, what is actually accurate facts.
00:15:53.120 --> 00:16:01.039
There's a real need for data centers to be explained uh to so that the society can understand really what it what it entails.
00:16:01.279 --> 00:16:14.080
Um at the end of the day, you know, we we as an industry, uh one, we haven't been great at being transparent ourselves, you know, but we're an industry that's based on secrecy, uh, has always been the case.
00:16:14.240 --> 00:16:16.799
So trying to get rid of that mentality, I think, is the first thing.
00:16:16.960 --> 00:16:22.320
The second thing is we're not great at uh being transparent on the facts of in terms of how we operate.
00:16:22.399 --> 00:16:24.799
Um that's something we we need to get better at.
00:16:25.039 --> 00:16:30.799
And the other issue is that we need to get away from the idea that the public don't understand what data centers are.
00:16:30.960 --> 00:16:36.480
I mean, I've spoken to so many residents around data centers, they they do understand that the data center underpins what they do.
00:16:36.559 --> 00:16:40.879
You know, we don't need to educate them that you're booking a flight on EasyJet, you're using a data center.
00:16:40.960 --> 00:16:41.440
They get that.
00:16:41.519 --> 00:16:43.360
I mean, they're not stupid, they get that.
00:16:43.519 --> 00:16:48.080
The bit I think they don't get is the environmental impact of the data center and how it operates.
00:16:48.159 --> 00:16:56.559
And that's largely because I think they, you know, they read the newspapers, the newspapers are quoting figures from 20 years ago based on data centres that are in Arizona and Nevada.
00:16:56.720 --> 00:17:02.000
Um, we're operating um our new data centre is a closed loop system, for example, on water.
00:17:02.159 --> 00:17:11.680
We're using the same amount of water as one family of four users in one year uh across 20 years of operation for a data centre that powers 32,000 GPUs.
00:17:11.839 --> 00:17:13.839
So the water element there is negligible.
00:17:14.000 --> 00:17:25.039
So if you if you're building that type of infrastructure and at the same time you're being told that you might drain Loch Lomond of water, it there's a real juxtaposition there.
00:17:25.200 --> 00:17:32.160
So I think it's just by a case-by-case basis, taking each fact as it comes and trying to provide the accuracy on it.
00:17:32.400 --> 00:17:34.960
Yeah, I think I'm again agree completely with that.
00:17:35.039 --> 00:17:48.559
I think as an industry we haven't done a particularly good job in terms of like tackling the misinformation out is that is out there and articulating the value story around both data centers and in our specific case AI data centers.
00:17:48.960 --> 00:18:02.480
Um it is a challenge because these things are very complex structures and systems, and to and so to like counterbalance relatively simple arguments like use a lot of energy, use a lot of water makes noise, those are quite simple.
00:18:02.559 --> 00:18:06.240
And as Anna says, there's a load of like just really bad data out there.
00:18:06.480 --> 00:18:29.519
Um from our perspective, that community engagement piece at the very local level is absolutely key, and also understanding that the communities that we operate in, they're all very different, whether they're in the US or in Europe, you know, communities in Finland have different expectations of what a data center company would offer versus Norway or Sweden or the UK.
00:18:29.680 --> 00:18:59.680
So it's about having that kind of like on the ground, being being part of the community on the ground, showing like an actual human face, and then working out like how do we contribute viably to the local economies and making sure that these data centers are sustainable, not in terms of an environmental perspective, but a sustainable and long-term part of these communities, and so that's something we have to take very seriously and is a you know critical part of the engagement we do very much at the at the local level.
00:18:59.920 --> 00:19:02.400
Yeah, I think I've I've been on a bit of a journey with this myself.
00:19:02.480 --> 00:19:06.640
I think you you kind of look at it as around you need to share the good news, but it's not true, is it?
00:19:06.799 --> 00:19:07.759
You need to share the truth.
00:19:07.839 --> 00:19:13.519
Like I mean, you can look at it like, oh, there's some positive stories about the industry, and don't get me wrong, there's some amazing positive stories.
00:19:13.599 --> 00:19:19.279
Like we were talking about the individual positive stories about people that have developed their career and gone on to be really successful.
00:19:19.440 --> 00:19:30.079
There's a positive story about companies that have you know totally transformed because of the data center industry, but when it comes to the data behind it, I no pun intended, um, it's being honest, isn't it?
00:19:30.160 --> 00:19:44.400
Like this is how much water it uses, this is how much power it uses, this is how big it is, you know, and and this is how large it will become, not might become, like I always talk about the you know, it's not a one gigawatt data center when only 20 megawatts have been approved.
00:19:44.480 --> 00:19:46.079
It's it potentially could be.
00:19:46.160 --> 00:19:50.000
So I think how do we get the truth out there is is the way I'm looking at it now.
00:19:50.079 --> 00:19:52.799
Like, what can I do personally to share more true stories?
00:19:52.880 --> 00:19:54.079
I think that's really important.
00:19:54.160 --> 00:20:13.519
But it's it's a big challenge because you are now competing with the other side of the debate, which is around uh we don't like data centres, we don't want data centres, and I'm finding it more and more difficult to find none, I don't want to say negative, but stories from that side of the debate when I'm doing my newsletter.
00:20:13.680 --> 00:20:14.720
It's harder and harder.
00:20:14.799 --> 00:20:27.359
If you search data centre, or you at home, or search data centre, the first two pages on Google, I guarantee you will include at least 50% of articles from media around negative impact of data centres.
00:20:27.440 --> 00:20:29.920
So it's a it's a big noise we're we're fighting again.
00:20:30.079 --> 00:20:31.119
Yeah, I couldn't agree more.
00:20:31.279 --> 00:20:37.200
I mean, I think the the there are so many, many challenges um to your point about the narrative.
00:20:37.440 --> 00:20:43.759
I I don't think it's the job of the general public to have to understand what we do.
00:20:43.839 --> 00:20:46.079
I think the industry needs to do a better job of communicating.
00:20:46.160 --> 00:20:47.920
I think it needs multiple initiatives.
00:20:48.000 --> 00:20:51.680
That idea of explaining data centers, data centers explained is a good one, right?
00:20:51.759 --> 00:20:52.720
And it's really important.
00:20:52.880 --> 00:21:04.000
But also I think like you need to um, as an industry, you need to appeal to Joe Bloggs on the street and you need to make sense to, you know, like the plumbers, electricians, what will AI do for them, right?
00:21:04.160 --> 00:21:08.640
People who don't really understand this technology and work with their hands all the day, how can they get a career?
00:21:08.720 --> 00:21:10.559
Um, how can they improve their lives?
00:21:10.720 --> 00:21:19.839
How could they go from working underneath kitchen sinks or in cupboards to working on large-scale industrial sites and improving the future for not only them but for the people that are around them?
00:21:19.920 --> 00:21:23.200
I think there's there's so many other positive stories and different angles we could take.
00:21:23.279 --> 00:21:26.559
Um, but it's multifaceted and increasingly complex.
00:21:26.720 --> 00:21:28.720
So uh yeah, more to come.
00:21:28.799 --> 00:21:30.640
So look, guys, we are at time.
00:21:30.880 --> 00:21:33.359
Appreciate you guys taking the time to speak with us today.
00:21:33.440 --> 00:21:36.160
And I'm sure we could sit around talking about this all day long.
00:21:36.240 --> 00:21:41.119
So maybe we do a follow-up and uh and get into kind of the press side of it in the future.
00:21:41.359 --> 00:21:42.160
Cheers, guys.
00:21:42.400 --> 00:21:44.960
That's it for this episode of the Inside Data Centre Podcast.
00:21:45.119 --> 00:21:46.079
Thanks for tuning in.
00:21:46.240 --> 00:21:54.079
This show is brought to you by DataX and Data Exec, the recruitment partners powering global talent across the data centre and digital infrastructure world.
00:21:54.400 --> 00:21:58.240
If you enjoyed today's conversations, don't forget to follow, like, and subscribe.
00:21:58.400 --> 00:22:00.720
It really helps us grow the podcast community.
00:22:01.119 --> 00:22:06.000
Stay tuned for more stories, insights, and conversations with the people shaping the future of our industry.
