TRANSCRIPT 🔗
WEBVTT
00:00:00.239 --> 00:00:06.000
The data centre 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.
00:00:20.079 --> 00:00:25.920
Whether you live in Breathe Data Centres or just want to understand what's powering our connected world, you're in the right place.
00:00:26.160 --> 00:00:27.839
Welcome to the Inside Data Centre Podcast.
00:00:27.920 --> 00:00:32.159
We're back again with another episode from the Accelerated Compute Show in London.
00:00:32.320 --> 00:00:34.640
Another discussion around AI infrastructure.
00:00:34.719 --> 00:00:36.399
Um, very hot topic at the moment.
00:00:36.479 --> 00:00:42.320
There's uh a lot of uh chat in the media around this subject, so looking forward to getting some more expert views on it.
00:00:42.479 --> 00:00:44.240
I'll let my guests introduce themselves.
00:00:44.560 --> 00:00:46.960
Matthew Baines, I work for Schneider Electric.
00:00:47.039 --> 00:00:51.359
I'm the VP of our data centre and secure power operations for the UK and Ireland.
00:00:51.600 --> 00:00:57.359
Thanks, I'm Carl Havard, Chief Commercial Officer of ERA 4, and we're building out AI infrastructure across the UK.
00:00:57.600 --> 00:01:00.560
That was a uh a very quick intro, Carl, to say the least.
00:01:00.640 --> 00:01:03.359
I'm not going to introduce myself, but it's good to be back, Andy.
00:01:03.759 --> 00:01:10.159
We are around about I'll say 20, 21 months on from the launch of the AI Opportunities Action Plan, right?
00:01:10.239 --> 00:01:17.760
And there are still lots of questions being asked around it, lots of divided opinions around what it's going to be to kind of make the UK success.
00:01:17.920 --> 00:01:18.560
Where are we now?
00:01:18.640 --> 00:01:19.920
Let's start with a punchy one.
00:01:20.000 --> 00:01:23.439
Where are we now in terms of the UK's AI ambitions?
00:01:23.680 --> 00:01:28.000
And what is it realistically going to take for the UK to become an AI maker?
00:01:28.159 --> 00:01:29.280
How long have we got?
00:01:29.840 --> 00:01:30.640
20 minutes.
00:01:30.879 --> 00:01:31.200
Okay.
00:01:31.359 --> 00:01:32.879
Um, so yeah, you're right.
00:01:33.040 --> 00:01:40.239
Uh Matt Clifford AI Opportunity Action Plan, 50 points launched, announcement of AI Growth Zones, Keir Salmer standing up.
00:01:40.400 --> 00:01:43.120
I think it was January 13, 2025.
00:01:43.519 --> 00:01:49.359
Yeah, so the AI Opportunity Action Plan, I think, was put out there with all the great intentions.
00:01:50.000 --> 00:02:03.680
Um, but the world has moved at such a pace that um there is a strong argument to be had that a lot of what was put out there, specifically around the AI growth zones for the UK, may be now outdated.
00:02:03.840 --> 00:02:07.200
The reason I say that is the AI market's moved on.
00:02:07.359 --> 00:02:22.960
The the need for massive clusters of 500 megawatts and 100 uh acres for an AI growth zone to train big frontier models at the time seemed a good idea and the right intent was behind it.
00:02:23.199 --> 00:02:27.280
The reality now has moved on where that market has sort of leveled out now.
00:02:27.360 --> 00:02:30.319
We know who the big frontier models uh makers are.
00:02:30.479 --> 00:02:44.639
Um thankfully, there's a few now emerging out of the UK, like Cosine, which is great, but it's the usual suspects that we expected who are making these models, um, who are deploying on big clusters internationally because the energy is cheaper.
00:02:44.879 --> 00:02:48.000
So the market's now moved more to inference and tokenization.
00:02:48.080 --> 00:03:05.599
So people are deploying the AI models, tuning their own data with it, and offering it to their companies, to the citizens of the UK to create videos of cats, whatever, whatever it might be, which means you don't need significant sized AI growth zones.
00:03:05.759 --> 00:03:11.120
You can actually have a distributed mesh of AI data centers across the the UK.
00:03:11.360 --> 00:03:15.039
Now, also since the AI Action Plan, we've DCIT's gone.
00:03:15.280 --> 00:03:25.360
Um, we've got a new Prime Minister, and uh the new Prime Minister Andy Burnham, I think, has got absolutely the right idea about devolution, you know, good growth in every postcode.
00:03:25.680 --> 00:03:40.719
And um I think that in itself may see, I'm being optimistic here, the decline in AI growth zones and actually supporting of AI data centers in harmony with local communities and being built that way.
00:03:41.120 --> 00:03:44.639
Can't top that, but that's uh very good insight, I think.
00:03:44.719 --> 00:03:46.240
Um and I I'm aligned with it.
00:03:46.319 --> 00:03:53.919
I think um, you know, we were very excited about the launch of the AI growth zones in the in the UK and Ireland, um, as you said, a couple of years ago.
00:03:54.400 --> 00:03:57.919
Um, but nothing massively has come from that, really.
00:03:58.080 --> 00:04:03.759
Um at the time, I think the UK still struggled to attract the uh the companies to deploy in the UK.
00:04:03.840 --> 00:04:07.199
And I look there's still uh there's a big challenge with the cost of energy, really.
00:04:07.280 --> 00:04:16.800
I think was one of the biggest prohibitors in the UK and still remains so, um, particularly when you talk about AI training models and uh the frontier models, as you as you say.
00:04:17.040 --> 00:04:20.560
Where the where the zones are located, you know, people have their opinions on that.
00:04:20.639 --> 00:04:28.000
Maybe I I don't have a the a massively strong opinion, but maybe some people felt that wasn't maybe not the ideal
Load full transcript
locations, perhaps.
00:04:28.160 --> 00:04:36.480
But if you're training models, then I guess in essence they can be located, well the location that is not as important as uh enterprise um models.
00:04:36.720 --> 00:04:42.399
And the topic about um the uh uh devolution and the the regionalization I think is is going to be key as well.
00:04:42.560 --> 00:05:03.920
Um, if we're moving more away from training, let's say, because as you said, that market's moved on, we know who the players are, we know where they're deploying, and uh it's uh the economics of it mean the deployment of that goes in low-cost, low energy cost areas, of course, um, and uh access to really renewable, uh available energy, then we're gonna move more towards the inference, as you said.
00:05:04.000 --> 00:05:12.000
And the UK, in terms of the inference market, if we call it inference or enterprise AI, is primed for that.
00:05:12.240 --> 00:05:14.800
It's a huge enterprise market in the UK.
00:05:14.959 --> 00:05:20.079
Enterprise cloud is still very strong, enterprise AI will be strong in the in the UK and Ireland.
00:05:20.160 --> 00:05:30.959
The uh the the corporations we have um embedded into the country and the communities are strong in terms of uptake of IT and innovation and uh and uh early adopters of that.
00:05:31.040 --> 00:05:36.480
So I think that'll be strong for us and will will provide good growth outside of the training.
00:05:36.639 --> 00:05:39.040
But that that one challenge still remains, I think.
00:05:39.120 --> 00:05:52.000
They'll be located in metropolitan areas and maybe a few other areas, but we still have the power problem, I think, in terms of the access to energy and the speed at which we can get that energy to really um really for that to take off and lift off, I think.
00:05:52.160 --> 00:06:07.920
It's interesting you talk about inference in the in the UK as you know an inference leader, shall we say, because I've been talking to capital for probably about three to four years around potentially creating platforms or organizations within that kind of model and they've never really moved ahead.
00:06:08.079 --> 00:06:20.959
So, like, is it now is now the time for that capital to be mobilised because you the capital for training is so significant now that it kind of limits the amount the the ability of certain capital partners to enter the market.
00:06:21.120 --> 00:06:27.759
They can't come in with two, three billion to invest, whereas a few hundred million is is much more feasible to get a return.
00:06:27.839 --> 00:06:30.879
So be interested in your views on whether that will happen now.
00:06:31.040 --> 00:06:33.839
Will we see that capital flow through into an inference market?
00:06:34.000 --> 00:06:48.560
And equally on the energy point, and and Carl, you touched on some of this in the keynote earlier this morning around the UK being competitive in the energy market, and can data centres become energy assets, which I know is is is one of your players.
00:06:48.639 --> 00:06:53.519
So, two questions there, but you know, can can the market become can data centres become an energy asset?
00:06:53.680 --> 00:06:57.759
And will the capital come in to create those inference organisations?
00:06:58.079 --> 00:07:01.759
You know, capital goes where there's a strong return, of course.
00:07:02.079 --> 00:07:08.720
Um the uh AI training, the capital funding, the AI training is uh is an interesting one.
00:07:08.959 --> 00:07:21.920
I think uh depends on who you talk to in the capital markets, whether it's private equity or you know, if you go senior debt, they have a different appetite for for these types of applications and the um the contract times and the way they look at the uh the returns.
00:07:22.079 --> 00:07:27.839
Um we talk about inference, let's say we could say that's an enterprise play.
00:07:28.079 --> 00:07:32.720
I think the uh the capital markets, the debt markets in the UK are very experienced in that.
00:07:32.879 --> 00:07:38.480
It's typically more uh, you know, there's a big co-location element when you talk about enterprise AI.
00:07:38.639 --> 00:07:41.040
And in the UK we're very well structured around that.
00:07:41.120 --> 00:07:50.800
We've got a good track record, we've got some very, very large players who are very well funded to house that enterprise or inference um around the country as well.
00:07:51.040 --> 00:08:01.839
I think for them, those organizations, raising capital for for that application is um is BAU a bit, and we know money wants to flow into this market, well is and has been and will remain so.
00:08:02.000 --> 00:08:08.560
There's a strong yield, there's a good return, uh uh AAA rated customers, and I think it's that that's healthy.
00:08:08.720 --> 00:08:13.920
So I think, I mean, I'm not a markets man myself particularly, but I try and navigate it.
00:08:14.079 --> 00:08:18.800
I think um I think that will be um the money will flow for inference, I think, for that.
00:08:18.959 --> 00:08:23.120
Can data centers be an active asset on the grid?
00:08:23.839 --> 00:08:29.279
I think for the the license to operate, I think there's something, a direction of travel that data centers need to go down.
00:08:29.439 --> 00:08:39.039
If we talk, I mean there's big numbers being floated around, of course, in the UK, but also Europe and and and the world about how many more gigawatts we're gonna we need to deploy in terms of data centers.
00:08:39.200 --> 00:08:42.399
And the flip side of that, of course, there's a lot of press around them.
00:08:42.559 --> 00:08:53.120
That's that's front and centre, and then we have a lot of pushback, we could say, um activist activity, you could say, whatever you want to call it.
00:08:53.279 --> 00:08:57.600
And I think we all understand that, but I don't think the narrative is quite right on that.
00:08:57.679 --> 00:09:01.759
So we have a bit of work to do as an industry to try and change that that
narrative.
00:09:01.840 --> 00:09:14.960
But part of that, of course, is um some of the fears I think people have, individuals have on the street, is it's taking all the energy, our energy costs are going to go up, it's taking all the water, and you know, it's fair enough if we haven't sent the right message out there.
00:09:15.279 --> 00:09:31.360
If we can send couple that message with the fact that data centers are are um accelerating energy technology, they are accelerating grid innovation, which typically is is uh is normally a long haul in terms of uh research and innovation on the grid, let's say.
00:09:31.600 --> 00:09:35.679
They're accelerating that and they're starting to become an active asset on the grid.
00:09:35.840 --> 00:09:40.639
There are some technical challenges to some of that that needs to be um looked at.
00:09:40.720 --> 00:09:52.080
It's happened in some areas, but I think the direction of travel needs to be in that direction, then that can help the narrative around data centers in terms of uh the pushback that we're getting from from society.
00:09:52.399 --> 00:09:56.399
So um capital's not a problem, it's the it's a type of capital.
00:09:56.559 --> 00:10:05.360
So uh overseas capital is sort of not easy to get, but it's it's uh if you've got a solid foundation on what to build and what you're deploying, it it's available.
00:10:05.679 --> 00:10:07.360
UK capital, less so.
00:10:07.600 --> 00:10:11.120
Um so there's the discussion about sovereign money there, which could come into play.
00:10:11.360 --> 00:10:17.360
Um when it comes to data centers as an asset to the UK, then I think you're right.
00:10:17.440 --> 00:10:33.279
You know, there's a lot of um press and act activists out there talking about uh not in my backyard, quite rightly so, um, sustainability issues which some are well informed, some are less so, like water usage, etc.
00:10:33.600 --> 00:10:47.600
And that that flows through into the House of Commons in the media, and I think there's a an education piece there, but absolutely the UK feels a little bit like the US, and data centres are being done to people, and that's not and that and that's not the right way to approach this.
00:10:47.919 --> 00:10:54.399
Having a big building pop up in the site of your back garden is not is not a not a nice thing to have.
00:10:54.639 --> 00:11:15.120
And um, you know, you can take in South Wales where there's activation going on there, where there's a reason for that, and the the people deploying data centres in in that area on there the west of the estuary, yeah, get that right, um, is uh is sort of smoke and mirrors.
00:11:15.200 --> 00:11:18.399
There's a lack of transparency on on what's being done there.
00:11:18.639 --> 00:11:22.080
Buildings popping up, people driving around in fancy cars, etc.
00:11:22.399 --> 00:11:24.399
Local community don't really know.
00:11:24.639 --> 00:11:35.519
Whereas on the other side of the estuary, the east side, uh, then the old steelworks there um you could say I would say this, but we've acquired the old steel works and we're cleaning that up.
00:11:35.679 --> 00:11:46.480
We're working in harmony with the Newport City Council there and working with them to um, to your point, create an asset that's gonna bring life back into that that that area.
00:11:46.639 --> 00:11:56.799
So, not just about prosperity, but could we um re-industrialise the part of the steelworks from the coiling point of view that brings people back to work?
00:11:56.960 --> 00:11:57.840
Yes, we can.
00:11:58.080 --> 00:11:59.519
Can we build a community centre?
00:11:59.600 --> 00:12:00.799
Are we gonna build a tech college?
00:12:00.879 --> 00:12:02.639
Are we gonna build sports facilities?
00:12:02.799 --> 00:12:04.799
Are we gonna preserve the conservation area?
00:12:04.960 --> 00:12:05.679
Yes, we are.
00:12:05.840 --> 00:12:12.639
Absolutely do all of that, create a destination again, only possible through the deployment of AI and the investment that brings.
00:12:12.879 --> 00:12:18.879
I mean, it's really interesting to hear you say that capital's not the problem, energy is, and more importantly, the cost of energy.
00:12:18.960 --> 00:12:39.120
And to your points about uh building on Brownfield land, uh I firmly believe you don't need to build data centres on the green belt, and very much aligned with kind of this vision for growth in every postcode, and uh and and and yeah, as I say, you know, I think there's uh a lot that can be done to kind of level up um level up lots of communities here before I say anything else.
00:12:39.200 --> 00:12:41.360
I'm just gonna throw them right back to Andy quickly.
00:12:41.519 --> 00:12:48.559
It's just a quick question on your point, Carl, because I've been to Newport to visit one of the data centres in that in that area, and I've seen the development coming from.
00:12:48.960 --> 00:12:52.000
The the great stuff that's happening there, is it coming through?
00:12:52.240 --> 00:13:03.279
Because you know, is is it are you able to get your story across to the local communities, or do you find that challenging because there's other voices that are kind of louder, shall we say?
00:13:03.519 --> 00:13:04.879
So it's a really interesting point.
00:13:04.960 --> 00:13:18.159
I mean it's it's like um a Petri dish uh uh of the Newport region where on one side you've got data centres being built in one way, and on the other side, I think I'd like to think we're doing it the right way.
00:13:18.240 --> 00:13:30.879
So we are engaging with the council, with the local community, um, having forums, you know, discussing what this could do, listening to the feedback and saying, you know, what is it you would like as part of this uh this area?
00:13:31.120 --> 00:13:44.399
Um even interviewing generations that were there previously in the steelworks to find out some of the stories they have, the lifelong friendships they've made in the area, how that's
changed, but what they would love to see come back into the region.
00:13:44.559 --> 00:13:52.399
So we're actively doing that, and wherever we deploy in the future, even you know, what where we generate our energy through waste, we engage in it in a similar way.
00:13:52.639 --> 00:13:54.000
Others don't do that.
00:13:54.159 --> 00:14:13.039
Uh, and and and and and that's my concern is the data centers industry as a whole is being tarnished with it, and some some quite rightly, but on some areas this could be a real benefit to the local community, not just to bring life back, but could we build something around um a startup community in the region?
00:14:13.120 --> 00:14:14.799
Could we do something with energy tariffs?
00:14:14.960 --> 00:14:20.240
Because we could we're bringing energy behind the meter, could we help the local community reduce their energy bills?
00:14:20.399 --> 00:14:22.799
So I think there are the right ways and wrong ways to do it.
00:14:22.879 --> 00:14:24.720
It just so happens that Newport has both.
00:14:24.879 --> 00:14:26.720
So it's an interesting point, right?
00:14:26.799 --> 00:14:29.200
We'll we'll come on to behind the meter energy in a moment.
00:14:29.360 --> 00:14:36.720
But given your experience then, what would you advise the data centre industry to do in terms of engaging with local communities?
00:14:36.799 --> 00:14:51.519
You know, if you're if you're seeing positive impacts in the Newport area, if it if the local MPs and the local government are aligned with your vision and what you're going to do, what would be the single piece of advice you would say to other AI infrastructure operators, other co-location data center developers as well?
00:14:51.919 --> 00:14:53.279
I think I think it's two things.
00:14:53.440 --> 00:15:04.159
Um, first one is which I've just sort of said, that's that's absolutely genuinely engaged with the local community, but through your engagement, be completely transparent.
00:15:04.320 --> 00:15:08.879
Don't try and um pull the wool over people's eyes because people are smart.
00:15:09.440 --> 00:15:15.679
They're not going to be, hang on a minute, you said the energy was coming from there when the reality is it's it's actually not 100% renewal.
00:15:16.320 --> 00:15:21.360
Be honest about it, and if there's a pathway to being more carbon neutral, then then talk about that.
00:15:21.519 --> 00:15:27.840
But also talk about the genuine benefit of building in their area and deliver.
00:15:28.159 --> 00:15:30.159
Don't say things and not do it.
00:15:30.240 --> 00:15:31.440
That's the other thing as well.
00:15:31.679 --> 00:15:35.840
Listening to what you're doing in Newport is quite unique, I think, and it's it's great to see.
00:15:35.919 --> 00:15:39.039
And I think um the industry needs to do more of that.
00:15:39.200 --> 00:15:45.200
Not everybody has the opportunity to do it because when you talk about you're taking over the steelworks, what part of it, and you can leverage some of that.
00:15:45.360 --> 00:15:48.960
That's not always the case, of course, if you're building in Slough, maybe.
00:15:49.039 --> 00:15:52.720
So if that's the best example, but you know what I mean in a in a more urban area.
00:15:52.879 --> 00:15:56.720
But definitely the industry needs to um portray itself a little bit better.
00:15:56.960 --> 00:16:02.879
Um I think it has possibly has uh it's been portrayed incorrectly, maybe.
00:16:03.039 --> 00:16:08.559
Again, you could argue that, but there's work, there's work to do, and I think your topic around, but what are people concerned about?
00:16:08.799 --> 00:16:11.360
Are they concerned about the building being close to them?
00:16:12.399 --> 00:16:15.919
From a from an eyesore perspective, maybe.
00:16:16.080 --> 00:16:24.480
Um they're concerned about it's gonna increase energy bills, they're concerned about um uh the noise, perhaps as well.
00:16:24.639 --> 00:16:29.519
But what they really want to know about is job creation, of course, and particularly in areas like that, of course.
00:16:29.600 --> 00:16:34.000
So, and I think it's not clear the job creation for data center actually today.
00:16:34.159 --> 00:16:35.360
So there's some work to be done there.
00:16:35.440 --> 00:16:36.720
That's uh that would be a very good one.
00:16:36.879 --> 00:16:48.320
The topic around uh alignment with the grid and the and and the energy side of the equation and putting to rest perhaps the concern around raising energy prices because of having data center in the area.
00:16:48.559 --> 00:16:53.919
Maybe we can feed you can feed back in with your your waste of energy, even if that can help to reduce at certain times.
00:16:54.080 --> 00:16:56.879
But then we talk about heat reuse or waste waste heat as well.
00:16:56.960 --> 00:17:07.359
I was at a data center in that was in um Sweden, where the or Sweden or Finland, where they're building, they're gonna feed um waste heat to 100,000 homes.
00:17:07.519 --> 00:17:16.799
And the people in those homes are gonna benefit massively by something of the order, actually I won't say a figure, but a hugely reduced energy or heating bill as a result of that.
00:17:16.880 --> 00:17:19.200
Now, do you want your house next to a data center?
00:17:19.440 --> 00:17:30.720
Necessarily, because you have to have there's a distance challenge, but these are some of the discussions we just need to amplify a bit more, I think, as well, as an industry collectively, um, to put maybe some of the concerns to rest.
00:17:30.799 --> 00:17:33.839
But maybe the last thing is what's actually going on in the data center?
00:17:33.920 --> 00:17:39.039
Because data center is just a building that's housing something, it's uh it's an enabler.
00:17:39.200 --> 00:17:53.440
It's data center is only the enemy if they're designed badly, but otherwise they are housing infrastructure that's really underpinning the digital economy and what everybody's doing in their day-to-day
lives, and we probably have to amplify that a bit more.
00:17:53.519 --> 00:17:55.359
What's happening inside the data center?
00:17:55.440 --> 00:18:00.880
What's the engine in there and what's it providing to the communities and to the nation and to the world?
00:18:01.039 --> 00:18:11.200
That's a bit of a big but you know what I mean, I think, in terms of uh what we do every day goes through a data center, and that's probably not clear to perhaps you know every man on the street.
00:18:11.519 --> 00:18:26.319
The conversations across the water are certainly not happening from uh helping from a perception perspective right now around AI being the end of the world, and you know, and certain conversations that have a it I couldn't agree more with your point around what happens inside is really important.
00:18:26.480 --> 00:18:28.000
What what's the data centre for?
00:18:28.079 --> 00:18:31.519
You know, is it providing a cure for an illness?
00:18:31.680 --> 00:18:33.839
Is it you know, sir is it cat videos?
00:18:33.920 --> 00:18:35.759
I think it be transparent to a degree.
00:18:35.839 --> 00:18:45.920
I know you can't be completely transparent about what is actually doing will help communities understand why we need them so much, but equally it's on us as an industry to communicate that much better.
00:18:46.160 --> 00:18:53.119
I know we're short on time, but behind the meter power was something we wanted to quickly cover um because of both of the your positions within that.
00:18:53.200 --> 00:18:55.440
But is that a solution for the UK market?
00:18:55.519 --> 00:18:58.720
You know, is that how we overcome this challenge around the grid?
00:18:58.799 --> 00:19:05.680
You know, we're seeing it in certain regions now, but is it is it feasible for that to become a solution as we move forward?
00:19:05.920 --> 00:19:06.559
We're having to.
00:19:06.640 --> 00:19:11.200
I mean, we're being forced into that position uh because the grid's not um fit for purpose.
00:19:11.279 --> 00:19:15.279
There's a massive NISO queue of uh zombie projects that are being worked through, etc.
00:19:15.599 --> 00:19:19.279
Um, of people saying, Yeah, I've got power, give me the ability.
00:19:19.599 --> 00:19:22.160
Oh well, so um there's a whole bunch of things.
00:19:22.240 --> 00:19:26.000
So ER4 generates its own uh energy from waste.
00:19:26.160 --> 00:19:28.079
You mentioned I think cost of energy earlier.
00:19:28.240 --> 00:19:32.720
I think the the national grid charge about 23 pence per kilowatt hour.
00:19:33.039 --> 00:19:39.039
We generate about 50 megawatts in total through waste management, which is a a byproduct of consumism.
00:19:39.279 --> 00:19:49.200
You know, we just sort of circular economy where we're pulling that out, returning it, converting it into methane, bagging it up, putting that and converting it into electricity.
00:19:49.359 --> 00:19:52.400
We sell back to the grid for five pence per kilowatt.
00:19:52.640 --> 00:19:56.480
So we plug it in at five pence, the grid sell it on for 23 pence.
00:19:56.720 --> 00:19:57.839
Big gap there, right?
00:19:58.079 --> 00:20:18.000
So so what we what we are are now doing is saying, well, we could use that energy as a foundation to stand up AI infrastructure, and then that gives us um you know a real lower base to actually maybe produce AI infrastructure for inference or tuning and attract overseas or internal revenue onto the platform.
00:20:18.160 --> 00:20:21.200
Um but we're limited to 50 megawatts.
00:20:22.079 --> 00:20:46.480
Um the other thing we're doing on the reindustrialization of sites is we're looking for behind-the-meter sources of energy, and um there's an abundance of offshore power, which I think I mentioned this morning actually, which um like the wind farms in the in in Scotland and the North Sea are being told and even paid to not put their energy on mainland because the national grid cannot take it.
00:20:46.640 --> 00:20:49.680
Um so they're looking at batteries and storing up batteries, etc.
00:20:50.000 --> 00:20:54.799
So that can be managed in a way where when the grid needs demand, it pulls from the batteries, etc.
00:20:54.960 --> 00:20:55.200
etc.
00:20:55.519 --> 00:20:56.960
But it's still a bit clunky.
00:20:57.119 --> 00:21:14.559
I think I think if the grid can be rejuvenated, and you mentioned Sweden, localized energy tariffs in the UK, the other benefit for behind the meter energy is it reduces the bills of the of the local community, and ideally the grid can sort of update itself to be able to be ready to accept it.
00:21:14.799 --> 00:21:16.000
But that's my view.
00:21:16.480 --> 00:21:20.960
I think it is a it is a um a bona fide solution or option for the UK market.
00:21:21.119 --> 00:21:24.960
I think it has to be because access to energy is challenging for sure.
00:21:25.119 --> 00:21:45.519
We've seen a lot of uh schemes that have been delayed by seven years or seven years plus, and if you want to really accelerate AI, tech, innovation, and for the UK and Ireland to be a maker, not a taker, and be a leader in this market, you have you do need data centers to do that, and you need energy to power those data centers.
00:21:45.920 --> 00:21:52.640
If you have uh two heads clashing there where we need that, we need to accelerate, but the energy is not available, we have to look at other solutions.
00:21:52.720 --> 00:21:54.079
So, this is definitely a solution.
00:21:54.240 --> 00:21:58.000
I am aware of uh uh you know several people looking at that right now.
00:21:58.160 --> 00:21:59.759
Um, I think there's a lot of benefit to.
00:22:00.240 --> 00:22:02.880
It because of obviously the speed and access.
00:22:03.119 --> 00:22:12.160
The cost of it when you look at a micro big microgrid, for example, I think it works out roughly the same as uh just taking it off the grid.
00:22:12.240 --> 00:22:17.200
The cost of your energy is lower, but you've got the capital to put the grid in, of course, the microgrid in.
00:22:17.279 --> 00:22:23.519
Um, but once you balance it all out over 10-15 years, I think it's roughly equates to the cost of uh power on the grid today.
00:22:23.599 --> 00:22:25.279
So not but you get it much quicker.
00:22:25.359 --> 00:22:27.200
So definitely an option, definitely a solution.
00:22:27.279 --> 00:22:32.640
I think we'll see we'll see more of them um come to life in the in the next few years uh in the UK.
00:22:32.960 --> 00:22:39.440
Before we close, one question on the on the zombie projects, I guess, because it is a problem, isn't it?
00:22:39.519 --> 00:22:47.519
Like you say, we've got this massive queue of projects that the industry knows are not feasible and are not going to happen.
00:22:47.680 --> 00:22:55.839
Is there a solution to this problem, or is it just a case of the industry just needs to kind of highlight this challenge and hope that there becomes a solution?
00:22:56.319 --> 00:23:16.640
So it's a self-created problem, and that was when the AI Opportunities Action Plan was launched and uh applications or expression of interest for AI growth zones were requested by the government, they got over 300 applications through everything from real organizations that had genuine power and land to farmers that thought, I've got a windmill in my farm.
00:23:16.720 --> 00:23:25.440
I can generate so all of a sudden all the NISO Q flooded, and therefore the AI Growth Zone had the criteria around 500 megawatts, 100 acres.
00:23:25.519 --> 00:23:27.519
So the aim was to cut that down.
00:23:28.000 --> 00:23:37.440
I think the reality is that on peak demand, the UK is at 65 gigawatts of consumer industrial need, 65 gigawatts cross-the-country, UK demand.
00:23:37.599 --> 00:23:45.039
I think the Niso Q was close to 80 gigawatts on top of that, so more than the whole of the UK you know currently uses.
00:23:45.279 --> 00:23:46.400
So can they fix it?
00:23:46.640 --> 00:23:54.480
I think the the current solution is well, if you're in the queue, we're going to charge you money to be in the queue, and therefore there's a self-thing.
00:23:54.559 --> 00:24:00.160
Well, do we if I if I'm a farmer with a windmill, um, do I want to pay 20 grand to stay in the queue?
00:24:00.319 --> 00:24:01.119
Maybe not.
00:24:01.440 --> 00:24:13.680
But ultimately, um, you know, a bit of dedicated focus, a bit of reframing, maybe using a bit of AI to assess the applications properly, could be the right solution.
00:24:13.759 --> 00:24:16.799
But unfortunately, it was a solution created by themselves.
00:24:17.039 --> 00:24:17.839
Yeah, it's a problem.
00:24:18.000 --> 00:24:19.440
Um, what's the solution?
00:24:19.599 --> 00:24:20.720
Uh you touched on it.
00:24:20.799 --> 00:24:26.079
I think um I I saw and I saw some information recently that this is being looked at quite seriously, I think.
00:24:26.240 --> 00:24:39.200
Um and you talked about the the the fee, let's say, an application fee, but it's not just a fee,
I think there's also a qualification around it as well from the grid, which is great, we need more of that.
00:24:39.359 --> 00:24:47.519
So you can't just if you just have a uh a windmill, you know, and you put 70 megawatts in, they won't just say yes, you know, there's a big qualification around it.
00:24:47.680 --> 00:24:52.880
In terms of do you have a tenant or who's your tenant, who's your customer, when's it going to come in, what's the contract look like?
00:24:53.039 --> 00:24:55.680
I think this is what the grid has to do more of.
00:24:55.839 --> 00:25:13.200
Um, and they start to qualify out these um these applications, which really are, let's say, um very, very ambitious from them, highly, highly a bit too optimistic, and they're causing this cue and this this problem and this this view, incorrect view of demand, because we're nowhere near 80 gigawatts, of course.
00:25:13.359 --> 00:25:24.880
I think I also touch on it when when the announcements come out, and Rory knows it's kind of annoys me, like we're going to develop a one gigawatt site in pick a pick a place, uh, London, and it's never ever going to be one gigawatt.
00:25:24.960 --> 00:25:31.200
And and the and the community see that it's one gigawatt and they suddenly panic, where actually it's 20 megawatts being approved or whatever it might be.
00:25:31.279 --> 00:25:34.400
I think there's a there's a challenge there as well around the PR and the perception.
00:25:34.640 --> 00:25:35.839
I mean, it is an interesting one.
00:25:36.000 --> 00:25:39.359
The uh it's the off-gen demand connections reform that you're talking about, right?
00:25:39.440 --> 00:25:41.759
And and and getting rid of all of these zombie applications.
00:25:41.839 --> 00:25:44.079
I think it's it's divided opinion in industry.
00:25:44.240 --> 00:25:46.240
On the one hand, it's a good thing.
00:25:46.400 --> 00:25:59.920
It means that developers, operators who have a strong track record for actually delivering large-scale infrastructure can can continue to do that, and they should rightly, all of those applications should be re-evaluated using AI is a good idea, Carl, to do that.
00:26:00.160 --> 00:26:02.319
A smart idea, maybe the government will listen to that.
00:26:02.559 --> 00:26:10.559
On the other side of it, the actual costs, the upfront costs that they're asking for, I think have they're far too expensive, right?
00:26:10.799 --> 00:26:20.160
Having that additional capital investment on top of you know the development of a data centre now, which is you know millions of pounds a megawatt, is it just seems nuts.
00:26:20.400 --> 00:26:27.680
I think there's our their industry associations, Tech UK, who are taking that on and trying to kind of uh trying to address those things on behalf of the industry.
00:26:27.920 --> 00:26:34.559
All I will say is that this unravels a whole other topic that I think we can get into here and talk for absolutely hours and we're out of time.
00:26:34.640 --> 00:26:38.319
So, guys, thank you very much for joining us on the Inside Data Centre Podcast.
00:26:38.480 --> 00:26:41.039
Appreciate your time and uh enjoy the rest of the show.
00:26:41.359 --> 00:26:44.000
That's it for this episode of the Inside Data Centre Podcast.
00:26:44.079 --> 00:26:45.039
Thanks for tuning in.
00:26:45.200 --> 00:26:53.119
This show is brought to you by Data X and Data Exec, the recruitment partners powering global talent across the data centre and digital infrastructure world.
00:26:53.359 --> 00:26:57.200
If you enjoyed today's conversations, don't forget to follow, like, and subscribe.
00:26:57.359 --> 00:26:59.839
It really helps us grow the podcast community.
00:27:00.079 --> 00:27:05.279
Stay tuned for more stories, insights, and conversations with the people shaping the future of our industry.