WEBVTT
00:00:00.160 --> 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.
00:00:06.160 --> 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 --> 00:00:19.839
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 --> 00:00:28.079
Welcome to the Inside Data Centre Podcast.
00:00:28.160 --> 00:00:33.520
We're back with another live edition, this time from the Accelerated Compute Show in London.
00:00:33.840 --> 00:00:36.799
Delighted to be here, good friend Rory Flashman Wells once again.
00:00:36.960 --> 00:00:45.280
I'm going to introduce you to the guests shortly, but in these episodes we're going to be talking largely about AI infrastructure with a specific focus on the UK.
00:00:45.439 --> 00:00:47.200
I'll let my guests introduce themselves.
00:00:47.439 --> 00:00:48.880
Yeah, Manash Patel.
00:00:48.960 --> 00:01:00.479
I'm the uh head of data centre development at Radiant, and uh we are a new entrant to the market this year, uh owned by the Brookfield AI Infrastructure Fund and uh focused on delivering compute at a global level.
00:01:00.799 --> 00:01:04.799
I am Michael Began, I'm the managing director of TES Power.
00:01:04.879 --> 00:01:11.200
We specialise in switch gear and power distribution systems to feed the AI data centers.
00:01:11.519 --> 00:01:12.239
Good to be here again.
00:01:12.400 --> 00:01:13.840
And Ian, good to be here with you both.
00:01:13.920 --> 00:01:17.040
I'm Rory Flashman, Managing Director of Spark Communications.
00:01:17.120 --> 00:01:21.439
We're Europe's leading PR and social media agency for data centres and digital infrastructure.
00:01:21.599 --> 00:01:24.560
And yeah, really excited to be your co-host again, mate.
00:01:24.640 --> 00:01:25.760
So let's kick it off.
00:01:25.920 --> 00:01:30.400
We've only got a short amount of time to get through uh lots and lots of discussion.
00:01:30.560 --> 00:01:34.959
The main focus of the accelerated compute show is AI, AI infrastructure and neo clouds.
00:01:35.120 --> 00:01:40.560
Where does Europe have its biggest advantages in terms of AI and AI infrastructure, Mickey?
00:01:40.959 --> 00:01:49.760
From my point of view, I think the the key to all of that is the actual skills and the skill level that is actually inherent within the European countries.
00:01:49.840 --> 00:01:52.319
Uh we're very strong and engineering.
00:01:52.799 --> 00:01:59.040
Uh we've got the skills, the manufacturing capabilities to be able to deliver at speed and at pace.
00:01:59.200 --> 00:02:02.959
So I think that is probably the key advantage that I can see for Europe.
00:02:03.280 --> 00:02:16.159
I I think being in Europe, delivering AI, there's a lot of opportunity in in um proximity to sort of regions that are, you know, you talk about in the data center space, Flap D.
00:02:16.719 --> 00:02:21.599
Um, you know, in the AI space, there's there's not as much of a heavy requirement on that latency.
00:02:21.680 --> 00:02:27.039
So proximity to cities, you know, you don't have to be in the middle of London, you can afford to be in sort of adjacent cities.
00:02:27.120 --> 00:02:34.960
Whereas in you know the US, for example, if you're out of city, you're talking hundreds of miles, um, which is very expensive for utility distribution.
00:02:35.039 --> 00:02:37.199
So I think in Europe it's a lot more condensed.
00:02:37.280 --> 00:02:50.719
I think in the traditional data center play um that has its challenges, but when you bring AI into it and the the the levers are slightly different in terms of what we're looking for and how we're looking to deploy, um, and that presents advantages.
00:02:51.039 --> 00:03:05.919
The other interesting point about Europe is that there's a lot of heavy historical industry in in Europe, and um so that sort of brownfield um deployment is is uh something that you can capitalise on and you can use that to your advantage.
00:03:06.159 --> 00:03:20.240
So I'm thinking sort of old industrial facilities, factories, warehouses that have power connections, there's a lot more of that in in Europe, in in those areas, and I think that lends itself quite nicely to that AI model, which again it's it's quite agile, it's quite adaptable.
00:03:20.319 --> 00:03:30.240
You know, you can retrofit just into just about anywhere, and that's something that we're doing at Radiant, and um yeah, um that that means that Europe is still a really attractive prospect for AI.
00:03:30.560 --> 00:03:31.919
Interesting point on the sites.
00:03:32.000 --> 00:03:39.840
Can we just listen to the keynote and they were talking about brownfield sites and within the UK specifically, and how we should be utilising these areas?
00:03:39.919 --> 00:03:44.080
And my sort of initial thought process was can we do it fast enough?
00:03:44.240 --> 00:03:45.520
Can we scale these sites?
00:03:45.599 --> 00:03:52.400
We're seeing a lot of conflict in the uh media at the moment around greenfield sites, you know, and quite rightly in a number of areas.
00:03:52.639 --> 00:04:01.439
However, I question whether brownfield sites, manufacturing facilities, etc., can be scaled fast enough to meet the demand of these customers.
00:04:01.680 --> 00:04:06.479
At the end of the day, the customers putting the pressure on the developers like yourself, the manufacturers like yourself.
00:04:06.639 --> 00:04:19.279
So, in on that point, do you have any views on whether we can if we look at the UK specifically and Northern Ireland, can we scale fast enough on those sites to be competitive in the race?
00:04:19.600 --> 00:04:26.079
Really interesting sort of subject area, and I I um I deal with it on a day-to-day basis, right?
00:04:26.160 --> 00:04:35.600
So a lot of what we're we're doing now compared to sort of three years ago, if you were looking at uh due diligence for like a land parcel, your three ticks are is is is is it buildable?
00:04:35.680 --> 00:04:41.519
I is there a reason why I cannot build on this land, like structurally or or or issues with the land?
00:04:41.680 --> 00:04:43.360
Can I get power and can I get fiber?
00:04:43.519 --> 00:04:45.120
Those are your three green ticks.
00:04:45.279 --> 00:04:50.720
Um now a lot of those sort of um perfect land parcels are gone.
00:04:50.879 --> 00:04:55.279
And what you're left with is one of those ticks is not gonna be green.
00:04:55.360 --> 00:04:58.879
It's gonna be, you know, it's gonna be red, it's gonna be amber, there's gonna be a challenge there.
00:04:59.040 --> 00:05:11.120
And and what that means is that in any sort of new development, whether it's brownfield or greenfield, there's gonna have to be a sacrifice or a trade-off, which you didn't have two or three years ago when there was an abundance of really good quality land.
00:05:11.279 --> 00:05:27.600
And so in that situation, you know, brownfield land is is right back in the mix because you will have power, there will be some aspect of infrastructure that's already there, um, but you will have to sacrifice some things like maybe the fiber's not quite there, and you might have to spend a little bit more on fiber to get it to the actual site.
00:05:27.759 --> 00:05:41.199
But it in in terms of the actual physical infrastructure itself, you know, this um term box within a box, um, it there's a really good example in in France that I worked on of literally you have a warehouse, you have a building, it's structurally sound.
00:05:41.360 --> 00:05:50.079
There is absolutely no reason that you need to change anything with that building, therefore, from a planning perspective, that's a that's a leg up, that's a that's a um an advantage that you don't have to go through.
00:05:50.240 --> 00:05:55.199
And if you're retrofitting a box into that box, um that's a that's a timeline upside.
00:05:55.279 --> 00:05:57.920
So there is opportunity, there's absolutely opportunity there.
00:05:58.079 --> 00:06:02.160
Um the the challenge is what does it look like from a functional building perspective?
00:06:02.240 --> 00:06:12.000
You know, customers, hyperscalers, they want to use the facility for more than just deploying compute in some cases, they want nice office spaces, they want all the space to be able to have meetings and that type of thing.
00:06:12.160 --> 00:06:14.319
That's where you sort of get those challenges in Greenfield.
00:06:14.399 --> 00:06:17.600
You don't have that, um, you don't have that flexibility with Brownfield so much.
00:06:17.839 --> 00:06:31.120
Kind of from my perspective, the whole concept really of constructing a data center is changing now to manufacturing of a data center, so a lot of off-site manufacturing, so everything can come together at the right time.
00:06:31.279 --> 00:06:37.519
Now that obviously means that consecutive tasks can all happen now concurrently.
00:06:37.759 --> 00:06:46.240
So that that actually will definitely help speed up the whole uh deployment side of things, and it is definitely becoming more and more and more.
00:06:46.560 --> 00:06:52.079
The first question we get now isn't how much does all this package cost, it's can you deliver it on time?
00:06:52.319 --> 00:06:54.560
Then the next question is how much does it cost?
00:06:54.720 --> 00:07:12.879
So it's just showing you how essential it is now that we as manufacturers as well need to start looking at better ways to streamline processes and concepts and be able to deliver at scale faster to keep up with the demand and to make sure that we stay ahead of the curve on that.
00:07:13.120 --> 00:07:14.480
I mean, it's an interesting point.
00:07:14.560 --> 00:07:30.000
Everybody's saying access to speed, uh access to land are the kind of two of the bit, sorry, access to energy, access to land, or access to sites are two of the biggest challenges the industry is facing, but surely to build out faster, one of the big challenges is skills, right?
00:07:30.079 --> 00:07:38.879
You know, it's not just the UK all over the world, we need more engineers, more plumbers, more electricians, and so on and so forth to be able to build infrastructure quickly.
00:07:39.040 --> 00:07:44.720
So, from your perspective, Mickey, let's do it, let's focus on Northern Ireland for a second and then come on to wider Europe.
00:07:44.800 --> 00:07:53.199
You know, like why has suddenly, from your point of view, Northern Ireland become a major export of AI skills and more importantly, you know, power technology?
00:07:53.519 --> 00:07:55.519
Northern Ireland is uniquely positioned.
00:07:55.680 --> 00:08:04.639
Uh, we have the Windsor uh framework agreement in place, which actually allows us to actually feed both the European single market and the UK.
00:08:04.800 --> 00:08:07.199
So that has been a massive advantage.
00:08:07.279 --> 00:08:12.399
But traditionally and historically, Northern Ireland has a heavy industrial base.
00:08:12.720 --> 00:08:18.160
There's lots of skills there, and there's lots of companies that are capitalizing on those skills.
00:08:18.399 --> 00:08:22.959
Now, the other question there that you mentioned about you know, we definitely need more and more people.
00:08:23.199 --> 00:08:28.000
This is true, so we don't expect those people to actually come from anywhere else.
00:08:28.079 --> 00:08:30.000
We're going to have to create those people.
00:08:30.079 --> 00:08:36.399
So we've got different levels of talent, uh, working with regional colleges, apprenticeships in place.
00:08:36.559 --> 00:08:56.960
But equally, the other side of that is that you know, the practicalities is you are not going to get all the people that you may want, so that then relies more heavily at looking at the processes, how can we streamline how we're doing to create the same output volume with less people that we need traditionally.
00:08:57.039 --> 00:09:00.399
So it's about becoming way more efficient at how we do things as well.
00:09:00.559 --> 00:09:02.320
And look at that, that I think is a key.
00:09:02.639 --> 00:09:07.440
Yeah, and I I think just building on that, the the key point is um transferable skills, right?
00:09:07.600 --> 00:09:20.960
So um there's so many examples across Europe, and and Northern Ireland is one of those where you you look at adjacent um sort of markets and the the data center industry, like fundamentally, it's not one of those industries that's hard to pick up or hard to learn.
00:09:21.120 --> 00:09:38.639
It's just the sheer volume of of um of requirements is there, and so with the right um incentives and and with the right, I think, regulatory or sort of government backing, which I think is is quite prevalent in Northern Ireland, you have that opportunity from sort of you know, in football they say grassroots, right?
00:09:38.720 --> 00:09:42.799
So in in data center grassroots, I suppose it's it's in schools, right?
00:09:42.879 --> 00:09:54.159
It's apprenticeships, it's encouraging the the next generation that this is the market to go in and study in, and you have that sort of um that sort of fresh injection of of new talent that's there, that's ready.
00:09:54.320 --> 00:10:00.960
Um, and I think the UK also um is is really well positioned with that with the AI growth zone initiative.
00:10:01.200 --> 00:10:13.120
I know it's not um it's not uh perfect, it doesn't cover everything, but it certainly has step forward to helping unblock that regulatory um challenge that we have when we're trying to deploy these facilities.
00:10:13.519 --> 00:10:46.799
You know, you go through any any build project, any retrofit, there is an aspect of stakeholder management and and and you have to you know local community engagement, and a lot of it is centered around jobs and what we can bring to the local area, the local community, and um just building on the point that we mentioned earlier around industrials and sort of pivoting, you know, there's power stations who have um people who have been sort of employed for a particular function, and if you can re-energize that land, you can do something with that land, you're saving a lot of jobs, you're saving a lot of um the local economy there as well.
00:10:46.960 --> 00:10:57.440
And I I think in order for us in the AI space to be able to bring about real change and to build on those capabilities, it doesn't come unless you have that regulatory backing, and I think that's really important.
00:10:57.679 --> 00:11:04.240
Yeah, I could talk all day about skills and and recruitment, but I think um we released our salary survey this week, yesterday.
00:11:04.399 --> 00:11:05.200
I don't know what day it is.
00:11:05.279 --> 00:11:06.240
Monday we released it.
00:11:06.320 --> 00:11:13.120
Um and it look and one of the things I wanted to look at this year, which is really important, is what skills, what industries are skills coming from to the data centre sector?
00:11:13.200 --> 00:11:16.000
Because we've always looked at it as how many years' experience do you have in the industry?
00:11:16.159 --> 00:11:17.440
I was like, but where are they coming from?
00:11:17.519 --> 00:11:31.360
And and obviously within Europe we found a lot of it from energy infrastructure, you know, those kind of high-tech industries that are they have the skills, they have the mission critical element of it, um, engineering background that are now coming into the industry.
00:11:31.519 --> 00:11:33.039
The industry has opened its doors a little.
00:11:33.200 --> 00:11:39.600
It was very, I need five years experience for everything, doesn't matter what the job is, but out of necessity, it's had to open the door.
00:11:39.679 --> 00:11:41.279
So that's that's a positive, I think.
00:11:41.360 --> 00:11:43.039
Um, but we definitely need to do more of it.
00:11:43.200 --> 00:11:50.240
And I also think there's that conflict now between community of is AI good, is AI bad, that does impact people's career choices.
00:11:50.399 --> 00:12:05.120
So I think uh the conversation around the industry and how we present ourselves, it's important that we present that for a skills perspective as well, because people will be put off the industry now because of what they're reading in the media about the data centre industry and and AI as such.
00:12:05.200 --> 00:12:17.840
And I also think it's really important that we retain those skills, we don't train these people and they leave because there's a hell of a lot of people at the moment working internationally, not in the UK, or on data centre projects or for data centre platforms.
00:12:17.919 --> 00:12:19.120
So it's interesting.
00:12:19.279 --> 00:12:31.200
I think when we look forward as well, and that covers a wide spectrum of pretty much the whole industry, including skills, but and you touched on things like a box within a box, and and you touched on modulo, like how we construct and how we develop.
00:12:31.279 --> 00:12:34.399
But what do you see as a future for AI infrastructure?
00:12:34.639 --> 00:12:39.440
What's coming next, or what do we have to do next to ensure we're competitive in this market?
00:12:39.759 --> 00:12:51.600
Um, another really good question, and I think without sort of talking about hot topics that everybody discusses, um, I think the key thing is in the tech space, you're going to drive more power into the data holes.
00:12:51.679 --> 00:12:53.759
So densification is is really critical.
00:12:53.919 --> 00:13:11.919
And I think one thing that I'm noticing when you you explore NVIDIA and they're sort of the pipeline with their chips, you're you're progressing beyond the sort of the normal limits of a day-to-day electrician or a trained, a skilled worker to be able to work in that space just by the throughput of the power through the racks.
00:13:12.080 --> 00:13:33.120
And I think um there's got to be a significant amount of work done like across the entire industry around how to make working at that level safe for people, number one, um, but also all the challenges and tricks that you have around how expensive that equipment is and liabilities and insurance and all that type of thing.
00:13:33.200 --> 00:13:40.320
You know, the more power you put into a rack, the more expensive it is, the harder it becomes to have someone to work on it without voiding warranties and things like that.
00:13:40.480 --> 00:14:07.120
So we have to be really, really careful that we don't go sort of too far with what we're trying to do with these racks, and we we still allow there to be this sort of sensible balance between um what is sort of peak performance and you know cutting edge and really pivotal, and what makes economic sense in terms of can we actually deliver and operate this equipment in a way that is um is sensible, is is cost efficient, and also you need people to be able to run this equipment, right?
00:14:07.279 --> 00:14:09.200
So um, yeah, really topical thing.
00:14:09.279 --> 00:14:18.799
I think from my perspective, AI is is going to continue to push the boundaries of of what is what is possible in terms of power, power delivery, power distribution, which I'm sure you can you can talk to more than I can.
00:14:18.960 --> 00:14:26.480
But um what we have to do as an industry is we have to stay on top of those those challenges and we have to make sure that we're putting things in place from the beginning.
00:14:26.559 --> 00:14:38.960
You know, when you're designing data centers, when you're thinking about how it's going to be operated, you have to make sure that all those plans, procedures, the people that are going to run it are involved from day one, and it's not just a concept that gets built and you can't do anything with it.
00:14:39.200 --> 00:14:42.879
Yeah, I think the key takeaway to this is density, isn't it?
00:14:43.039 --> 00:14:44.080
Power density.
00:14:44.240 --> 00:14:51.679
Um switchgear and power distribution inherently is the same, it's just of a larger scale now.
00:14:52.000 --> 00:14:57.519
But it has limitations for low voltage uh switchgear designs.
00:14:57.600 --> 00:15:07.120
There's a there's an upper limit to the amount of power that you can really uh perform before you get into real thermal and magnetic effects through all of that.
00:15:07.279 --> 00:15:18.080
So there is a pivot to look at in other technologies like low voltage DC uh moving making its way into the the data holes and seeing how that is going to affect.
00:15:18.159 --> 00:15:22.720
Now that technology will happen, I'm convinced of it, further down the line.
00:15:22.960 --> 00:15:34.080
But right now, what we need to do to keep up with that is just ensure that, first of all, it's fine that we can produce all of that, but the supply chain behind all of this needs to be there.
00:15:34.320 --> 00:15:45.600
So, you know, it's making sure that all the manufacturers are capable of actually producing everything to come together for us to actually be able to be able to build and deliver on time.
00:15:45.840 --> 00:16:04.879
So there's a lot of focus on our side on just trying to the the the upscale, the process, the people, uh the the backstream for the supply chain, ensuring that we no longer it's just a supplier-client kind of agreement.
00:16:05.039 --> 00:16:13.759
We're there, we're in the middle of their operations, we're looking at what they're doing to ensure that we have the confidence that they can supply us for what we need moving forward.
00:16:14.080 --> 00:16:22.559
So I think the whole industry in a whole is actually becoming more intertwined, and we're all connecting on a deeper level than we did before.
00:16:22.720 --> 00:16:26.000
So that is going to help us moving forward, I believe, too.
00:16:26.320 --> 00:16:27.120
Thanks, Mickey.
00:16:27.200 --> 00:16:31.440
So speed, densification, power, people, skills.
00:16:31.519 --> 00:16:33.519
They're the big, big takeaways for the future.
00:16:33.759 --> 00:16:37.360
Guys, that literally brings us perfectly on time to the end of the session.
00:16:37.440 --> 00:16:42.240
So, look, really appreciate you taking the time to talk with us both today and uh enjoy the rest of the show.
00:16:42.960 --> 00:16:45.759
That's it for this episode of the Inside Data Centre Podcast.
00:16:45.840 --> 00:16:46.799
Thanks for tuning in.
00:16:46.960 --> 00:16:54.879
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.
00:16:55.120 --> 00:16:58.960
If you enjoyed today's conversations, don't forget to follow, like, and subscribe.
00:16:59.120 --> 00:17:01.600
It really helps us grow the podcast community.
00:17:01.759 --> 00:17:07.119
Stay tuned for more stories, insights, and conversations with the people shaping the future of our industry.
00:00:00.160 --> 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.
00:00:06.160 --> 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 --> 00:00:19.839
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 --> 00:00:28.079
Welcome to the Inside Data Centre Podcast.
00:00:28.160 --> 00:00:33.520
We're back with another live edition, this time from the Accelerated Compute Show in London.
00:00:33.840 --> 00:00:36.799
Delighted to be here, good friend Rory Flashman Wells once again.
00:00:36.960 --> 00:00:45.280
I'm going to introduce you to the guests shortly, but in these episodes we're going to be talking largely about AI infrastructure with a specific focus on the UK.
00:00:45.439 --> 00:00:47.200
I'll let my guests introduce themselves.
00:00:47.439 --> 00:00:48.880
Yeah, Manash Patel.
00:00:48.960 --> 00:01:00.479
I'm the uh head of data centre development at Radiant, and uh we are a new entrant to the market this year, uh owned by the Brookfield AI Infrastructure Fund and uh focused on delivering compute at a global level.
00:01:00.799 --> 00:01:04.799
I am Michael Began, I'm the managing director of TES Power.
00:01:04.879 --> 00:01:11.200
We specialise in switch gear and power distribution systems to feed the AI data centers.
00:01:11.519 --> 00:01:12.239
Good to be here again.
00:01:12.400 --> 00:01:13.840
And Ian, good to be here with you both.
00:01:13.920 --> 00:01:17.040
I'm Rory Flashman, Managing Director of Spark Communications.
00:01:17.120 --> 00:01:21.439
We're Europe's leading PR and social media agency for data centres and digital infrastructure.
00:01:21.599 --> 00:01:24.560
And yeah, really excited to be your co-host again, mate.
00:01:24.640 --> 00:01:25.760
So let's kick it off.
00:01:25.920 --> 00:01:30.400
We've only got a short amount of time to get through uh lots and lots of discussion.
00:01:30.560 --> 00:01:34.959
The main focus of the accelerated compute show is AI, AI infrastructure and neo clouds.
00:01:35.120 --> 00:01:40.560
Where does Europe have its biggest advantages in terms of AI and AI infrastructure, Mickey?
00:01:40.959 --> 00:01:49.760
From my point of view, I think the the key to all of that is the actual skills and the skill level that is actually inherent within the European countries.
00:01:49.840 --> 00:01:52.319
Uh we're very strong and engineering.
00:01:52.799 --> 00:01:59.040
Uh we've got the skills, the manufacturing capabilities to be able to deliver at speed and at pace.
00:01:59.200 --> 00:02:02.959
So I think that is probably the key advantage that I can see for Europe.
00:02:03.280 --> 00:02:16.159
I I think being in Europe, delivering AI, there's a lot of opportunity in in um proximity to sort of regions that are, you know, you talk about in the data center space, Flap D.
00:02:16.719 --> 00:02:21.599
Um, you know, in the AI space, there's there's not as much of a heavy requirement on that latency.
00:02:21.680 --> 00:02:27.039
So proximity to cities, you know, you don't have to be in the middle of London, you can afford to be in sort of adjacent cities.
00:02:27.120 --> 00:02:34.960
Whereas in you know the US, for example, if you're out of city, you're talking hundreds of miles, um, which is very expensive for utility distribution.
00:02:35.039 --> 00:02:37.199
So I think in Europe it's a lot more condensed.
00:02:37.280 --> 00:02:50.719
I think in the traditional data center play um that has its challenges, but when you bring AI into it and the the the levers are slightly different in terms of what we're looking for and how we're looking to deploy, um, and that presents advantages.
00:02:51.039 --> 00:03:05.919
The other interesting point about Europe is that there's a lot of heavy historical industry in in Europe, and um so that sort of brownfield um deployment is is uh something that you can capitalise on and you can use that to your advantage.
00:03:06.159 --> 00:03:20.240
So I'm thinking sort of old industrial facilities, factories, warehouses that have power connections, there's a lot more of that in in Europe, in in those areas, and I think that lends itself quite nicely to that AI model, which again it's it's quite agile, it's quite adaptable.
00:03:20.319 --> 00:03:30.240
You know, you can retrofit just into just about anywhere, and that's something that we're doing at Radiant, and um yeah, um that that means that Europe is still a really attractive prospect for AI.
00:03:30.560 --> 00:03:31.919
Interesting point on the sites.
00:03:32.000 --> 00:03:39.840
Can we just listen to the keynote and they were talking about brownfield sites and within the UK specifically, and how we should be utilising these areas?
00:03:39.919 --> 00:03:44.080
And my sort of initial thought process was can we do it fast enough?
00:03:44.240 --> 00:03:45.520
Can we scale these sites?
00:03:45.599 --> 00:03:52.400
We're seeing a lot of conflict in the uh media at the moment around greenfield sites, you know, and quite rightly in a number of areas.
00:03:52.639 --> 00:04:01.439
However, I question whether brownfield sites, manufacturing facilities, etc., can be scaled fast enough to meet the demand of these customers.
00:04:01.680 --> 00:04:06.479
At the end of the day, the customers putting the pressure on the developers like yourself, the manufacturers like yourself.
00:04:06.639 --> 00:04:19.279
So, in on that point, do you have any views on whether we can if we look at the UK specifically and Northern Ireland, can we scale fast enough on those sites to be competitive in the race?
00:04:19.600 --> 00:04:26.079
Really interesting sort of subject area, and I I um I deal with it on a day-to-day basis, right?
00:04:26.160 --> 00:04:35.600
So a lot of what we're we're doing now compared to sort of three years ago, if you were looking at uh due diligence for like a land parcel, your three ticks are is is is is it buildable?
00:04:35.680 --> 00:04:41.519
I is there a reason why I cannot build on this land, like structurally or or or issues with the land?
00:04:41.680 --> 00:04:43.360
Can I get power and can I get fiber?
00:04:43.519 --> 00:04:45.120
Those are your three green ticks.
00:04:45.279 --> 00:04:50.720
Um now a lot of those sort of um perfect land parcels are gone.
00:04:50.879 --> 00:04:55.279
And what you're left with is one of those ticks is not gonna be green.
00:04:55.360 --> 00:04:58.879
It's gonna be, you know, it's gonna be red, it's gonna be amber, there's gonna be a challenge there.
00:04:59.040 --> 00:05:11.120
And and what that means is that in any sort of new development, whether it's brownfield or greenfield, there's gonna have to be a sacrifice or a trade-off, which you didn't have two or three years ago when there was an abundance of really good quality land.
00:05:11.279 --> 00:05:27.600
And so in that situation, you know, brownfield land is is right back in the mix because you will have power, there will be some aspect of infrastructure that's already there, um, but you will have to sacrifice some things like maybe the fiber's not quite there, and you might have to spend a little bit more on fiber to get it to the actual site.
00:05:27.759 --> 00:05:41.199
But it in in terms of the actual physical infrastructure itself, you know, this um term box within a box, um, it there's a really good example in in France that I worked on of literally you have a warehouse, you have a building, it's structurally sound.
00:05:41.360 --> 00:05:50.079
There is absolutely no reason that you need to change anything with that building, therefore, from a planning perspective, that's a that's a leg up, that's a that's a um an advantage that you don't have to go through.
00:05:50.240 --> 00:05:55.199
And if you're retrofitting a box into that box, um that's a that's a timeline upside.
00:05:55.279 --> 00:05:57.920
So there is opportunity, there's absolutely opportunity there.
00:05:58.079 --> 00:06:02.160
Um the the challenge is what does it look like from a functional building perspective?
00:06:02.240 --> 00:06:12.000
You know, customers, hyperscalers, they want to use the facility for more than just deploying compute in some cases, they want nice office spaces, they want all the space to be able to have meetings and that type of thing.
00:06:12.160 --> 00:06:14.319
That's where you sort of get those challenges in Greenfield.
00:06:14.399 --> 00:06:17.600
You don't have that, um, you don't have that flexibility with Brownfield so much.
00:06:17.839 --> 00:06:31.120
Kind of from my perspective, the whole concept really of constructing a data center is changing now to manufacturing of a data center, so a lot of off-site manufacturing, so everything can come together at the right time.
00:06:31.279 --> 00:06:37.519
Now that obviously means that consecutive tasks can all happen now concurrently.
00:06:37.759 --> 00:06:46.240
So that that actually will definitely help speed up the whole uh deployment side of things, and it is definitely becoming more and more and more.
00:06:46.560 --> 00:06:52.079
The first question we get now isn't how much does all this package cost, it's can you deliver it on time?
00:06:52.319 --> 00:06:54.560
Then the next question is how much does it cost?
00:06:54.720 --> 00:07:12.879
So it's just showing you how essential it is now that we as manufacturers as well need to start looking at better ways to streamline processes and concepts and be able to deliver at scale faster to keep up with the demand and to make sure that we stay ahead of the curve on that.
00:07:13.120 --> 00:07:14.480
I mean, it's an interesting point.
00:07:14.560 --> 00:07:30.000
Everybody's saying access to speed, uh access to land are the kind of two of the bit, sorry, access to energy, access to land, or access to sites are two of the biggest challenges the industry is facing, but surely to build out faster, one of the big challenges is skills, right?
00:07:30.079 --> 00:07:38.879
You know, it's not just the UK all over the world, we need more engineers, more plumbers, more electricians, and so on and so forth to be able to build infrastructure quickly.
00:07:39.040 --> 00:07:44.720
So, from your perspective, Mickey, let's do it, let's focus on Northern Ireland for a second and then come on to wider Europe.
00:07:44.800 --> 00:07:53.199
You know, like why has suddenly, from your point of view, Northern Ireland become a major export of AI skills and more importantly, you know, power technology?
00:07:53.519 --> 00:07:55.519
Northern Ireland is uniquely positioned.
00:07:55.680 --> 00:08:04.639
Uh, we have the Windsor uh framework agreement in place, which actually allows us to actually feed both the European single market and the UK.
00:08:04.800 --> 00:08:07.199
So that has been a massive advantage.
00:08:07.279 --> 00:08:12.399
But traditionally and historically, Northern Ireland has a heavy industrial base.
00:08:12.720 --> 00:08:18.160
There's lots of skills there, and there's lots of companies that are capitalizing on those skills.
00:08:18.399 --> 00:08:22.959
Now, the other question there that you mentioned about you know, we definitely need more and more people.
00:08:23.199 --> 00:08:28.000
This is true, so we don't expect those people to actually come from anywhere else.
00:08:28.079 --> 00:08:30.000
We're going to have to create those people.
00:08:30.079 --> 00:08:36.399
So we've got different levels of talent, uh, working with regional colleges, apprenticeships in place.
00:08:36.559 --> 00:08:56.960
But equally, the other side of that is that you know, the practicalities is you are not going to get all the people that you may want, so that then relies more heavily at looking at the processes, how can we streamline how we're doing to create the same output volume with less people that we need traditionally.
00:08:57.039 --> 00:09:00.399
So it's about becoming way more efficient at how we do things as well.
00:09:00.559 --> 00:09:02.320
And look at that, that I think is a key.
00:09:02.639 --> 00:09:07.440
Yeah, and I I think just building on that, the the key point is um transferable skills, right?
00:09:07.600 --> 00:09:20.960
So um there's so many examples across Europe, and and Northern Ireland is one of those where you you look at adjacent um sort of markets and the the data center industry, like fundamentally, it's not one of those industries that's hard to pick up or hard to learn.
00:09:21.120 --> 00:09:38.639
It's just the sheer volume of of um of requirements is there, and so with the right um incentives and and with the right, I think, regulatory or sort of government backing, which I think is is quite prevalent in Northern Ireland, you have that opportunity from sort of you know, in football they say grassroots, right?
00:09:38.720 --> 00:09:42.799
So in in data center grassroots, I suppose it's it's in schools, right?
00:09:42.879 --> 00:09:54.159
It's apprenticeships, it's encouraging the the next generation that this is the market to go in and study in, and you have that sort of um that sort of fresh injection of of new talent that's there, that's ready.
00:09:54.320 --> 00:10:00.960
Um, and I think the UK also um is is really well positioned with that with the AI growth zone initiative.
00:10:01.200 --> 00:10:13.120
I know it's not um it's not uh perfect, it doesn't cover everything, but it certainly has step forward to helping unblock that regulatory um challenge that we have when we're trying to deploy these facilities.
00:10:13.519 --> 00:10:46.799
You know, you go through any any build project, any retrofit, there is an aspect of stakeholder management and and and you have to you know local community engagement, and a lot of it is centered around jobs and what we can bring to the local area, the local community, and um just building on the point that we mentioned earlier around industrials and sort of pivoting, you know, there's power stations who have um people who have been sort of employed for a particular function, and if you can re-energize that land, you can do something with that land, you're saving a lot of jobs, you're saving a lot of um the local economy there as well.
00:10:46.960 --> 00:10:57.440
And I I think in order for us in the AI space to be able to bring about real change and to build on those capabilities, it doesn't come unless you have that regulatory backing, and I think that's really important.
00:10:57.679 --> 00:11:04.240
Yeah, I could talk all day about skills and and recruitment, but I think um we released our salary survey this week, yesterday.
00:11:04.399 --> 00:11:05.200
I don't know what day it is.
00:11:05.279 --> 00:11:06.240
Monday we released it.
00:11:06.320 --> 00:11:13.120
Um and it look and one of the things I wanted to look at this year, which is really important, is what skills, what industries are skills coming from to the data centre sector?
00:11:13.200 --> 00:11:16.000
Because we've always looked at it as how many years' experience do you have in the industry?
00:11:16.159 --> 00:11:17.440
I was like, but where are they coming from?
00:11:17.519 --> 00:11:31.360
And and obviously within Europe we found a lot of it from energy infrastructure, you know, those kind of high-tech industries that are they have the skills, they have the mission critical element of it, um, engineering background that are now coming into the industry.
00:11:31.519 --> 00:11:33.039
The industry has opened its doors a little.
00:11:33.200 --> 00:11:39.600
It was very, I need five years experience for everything, doesn't matter what the job is, but out of necessity, it's had to open the door.
00:11:39.679 --> 00:11:41.279
So that's that's a positive, I think.
00:11:41.360 --> 00:11:43.039
Um, but we definitely need to do more of it.
00:11:43.200 --> 00:11:50.240
And I also think there's that conflict now between community of is AI good, is AI bad, that does impact people's career choices.
00:11:50.399 --> 00:12:05.120
So I think uh the conversation around the industry and how we present ourselves, it's important that we present that for a skills perspective as well, because people will be put off the industry now because of what they're reading in the media about the data centre industry and and AI as such.
00:12:05.200 --> 00:12:17.840
And I also think it's really important that we retain those skills, we don't train these people and they leave because there's a hell of a lot of people at the moment working internationally, not in the UK, or on data centre projects or for data centre platforms.
00:12:17.919 --> 00:12:19.120
So it's interesting.
00:12:19.279 --> 00:12:31.200
I think when we look forward as well, and that covers a wide spectrum of pretty much the whole industry, including skills, but and you touched on things like a box within a box, and and you touched on modulo, like how we construct and how we develop.
00:12:31.279 --> 00:12:34.399
But what do you see as a future for AI infrastructure?
00:12:34.639 --> 00:12:39.440
What's coming next, or what do we have to do next to ensure we're competitive in this market?
00:12:39.759 --> 00:12:51.600
Um, another really good question, and I think without sort of talking about hot topics that everybody discusses, um, I think the key thing is in the tech space, you're going to drive more power into the data holes.
00:12:51.679 --> 00:12:53.759
So densification is is really critical.
00:12:53.919 --> 00:13:11.919
And I think one thing that I'm noticing when you you explore NVIDIA and they're sort of the pipeline with their chips, you're you're progressing beyond the sort of the normal limits of a day-to-day electrician or a trained, a skilled worker to be able to work in that space just by the throughput of the power through the racks.
00:13:12.080 --> 00:13:33.120
And I think um there's got to be a significant amount of work done like across the entire industry around how to make working at that level safe for people, number one, um, but also all the challenges and tricks that you have around how expensive that equipment is and liabilities and insurance and all that type of thing.
00:13:33.200 --> 00:13:40.320
You know, the more power you put into a rack, the more expensive it is, the harder it becomes to have someone to work on it without voiding warranties and things like that.
00:13:40.480 --> 00:14:07.120
So we have to be really, really careful that we don't go sort of too far with what we're trying to do with these racks, and we we still allow there to be this sort of sensible balance between um what is sort of peak performance and you know cutting edge and really pivotal, and what makes economic sense in terms of can we actually deliver and operate this equipment in a way that is um is sensible, is is cost efficient, and also you need people to be able to run this equipment, right?
00:14:07.279 --> 00:14:09.200
So um, yeah, really topical thing.
00:14:09.279 --> 00:14:18.799
I think from my perspective, AI is is going to continue to push the boundaries of of what is what is possible in terms of power, power delivery, power distribution, which I'm sure you can you can talk to more than I can.
00:14:18.960 --> 00:14:26.480
But um what we have to do as an industry is we have to stay on top of those those challenges and we have to make sure that we're putting things in place from the beginning.
00:14:26.559 --> 00:14:38.960
You know, when you're designing data centers, when you're thinking about how it's going to be operated, you have to make sure that all those plans, procedures, the people that are going to run it are involved from day one, and it's not just a concept that gets built and you can't do anything with it.
00:14:39.200 --> 00:14:42.879
Yeah, I think the key takeaway to this is density, isn't it?
00:14:43.039 --> 00:14:44.080
Power density.
00:14:44.240 --> 00:14:51.679
Um switchgear and power distribution inherently is the same, it's just of a larger scale now.
00:14:52.000 --> 00:14:57.519
But it has limitations for low voltage uh switchgear designs.
00:14:57.600 --> 00:15:07.120
There's a there's an upper limit to the amount of power that you can really uh perform before you get into real thermal and magnetic effects through all of that.
00:15:07.279 --> 00:15:18.080
So there is a pivot to look at in other technologies like low voltage DC uh moving making its way into the the data holes and seeing how that is going to affect.
00:15:18.159 --> 00:15:22.720
Now that technology will happen, I'm convinced of it, further down the line.
00:15:22.960 --> 00:15:34.080
But right now, what we need to do to keep up with that is just ensure that, first of all, it's fine that we can produce all of that, but the supply chain behind all of this needs to be there.
00:15:34.320 --> 00:15:45.600
So, you know, it's making sure that all the manufacturers are capable of actually producing everything to come together for us to actually be able to be able to build and deliver on time.
00:15:45.840 --> 00:16:04.879
So there's a lot of focus on our side on just trying to the the the upscale, the process, the people, uh the the backstream for the supply chain, ensuring that we no longer it's just a supplier-client kind of agreement.
00:16:05.039 --> 00:16:13.759
We're there, we're in the middle of their operations, we're looking at what they're doing to ensure that we have the confidence that they can supply us for what we need moving forward.
00:16:14.080 --> 00:16:22.559
So I think the whole industry in a whole is actually becoming more intertwined, and we're all connecting on a deeper level than we did before.
00:16:22.720 --> 00:16:26.000
So that is going to help us moving forward, I believe, too.
00:16:26.320 --> 00:16:27.120
Thanks, Mickey.
00:16:27.200 --> 00:16:31.440
So speed, densification, power, people, skills.
00:16:31.519 --> 00:16:33.519
They're the big, big takeaways for the future.
00:16:33.759 --> 00:16:37.360
Guys, that literally brings us perfectly on time to the end of the session.
00:16:37.440 --> 00:16:42.240
So, look, really appreciate you taking the time to talk with us both today and uh enjoy the rest of the show.
00:16:42.960 --> 00:16:45.759
That's it for this episode of the Inside Data Centre Podcast.
00:16:45.840 --> 00:16:46.799
Thanks for tuning in.
00:16:46.960 --> 00:16:54.879
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.
00:16:55.120 --> 00:16:58.960
If you enjoyed today's conversations, don't forget to follow, like, and subscribe.
00:16:59.120 --> 00:17:01.600
It really helps us grow the podcast community.
00:17:01.759 --> 00:17:07.119
Stay tuned for more stories, insights, and conversations with the people shaping the future of our industry.
