WEBVTT - Movie Marathons and Grid Flexibility 0:00:00.160 --> 0:00:02.840 Hi, I'm Dana Perkins and you're listening to Switched on 0:00:03.000 --> 0:00:05.840 the B n F podcast. So whatever device you're listening 0:00:05.920 --> 0:00:09.160 to this on right now, there is a data center involved. 0:00:09.440 --> 0:00:13.520 This podcast is out there in cyberspace, which actually isn't 0:00:13.560 --> 0:00:17.880 an intangible space at all. Cyberspace is comprised of many, 0:00:17.920 --> 0:00:23.080 many data centers in warehouses with significant energy demand in 0:00:23.120 --> 0:00:28.440 Europe across five key geographies the UK, Germany, Ireland, the 0:00:28.480 --> 0:00:32.800 Netherlands and Norway. This demand could grow by eight by 0:00:32.840 --> 0:00:35.960 the end of the decade. In this same period between 0:00:36.000 --> 0:00:40.280 now and wind and solar power are projected to approach 0:00:40.440 --> 0:00:44.640 six of total power generation. So what is a grid 0:00:44.640 --> 0:00:48.040 with a lot of renewable energy need well flexibility to 0:00:48.120 --> 0:00:50.519 handle the times when the sun isn't shining and the 0:00:50.560 --> 0:00:54.000 wind isn't blowing so hard. The question we're exploring today 0:00:54.160 --> 0:00:57.160 is could data centers and all of the backup energy 0:00:57.200 --> 0:01:01.240 that they have become a distributed energy resource and help 0:01:01.360 --> 0:01:05.880 provide grid flexibility. Michael ken Effect and Sarah Raza from 0:01:05.880 --> 0:01:08.520 B and E F worked on a report titled Data 0:01:08.560 --> 0:01:14.200 Centers and Decarbonization Unlocking Flexibility in Europe's Data Centers. This 0:01:14.280 --> 0:01:17.600 was done in partnership with Eaton and stack Craft. As 0:01:17.600 --> 0:01:20.039 a reminder, B and E F does not provide investment 0:01:20.120 --> 0:01:22.520 or strategy advice, and our full disclaimer is going to 0:01:22.560 --> 0:01:24.920 be at the end of the show. And now let's 0:01:24.920 --> 0:01:33.720 speak with Michael and Sarah about data centers and GRIT flexibility. 0:01:35.120 --> 0:01:37.800 I am joined today with Michael and Sarah to talk 0:01:37.880 --> 0:01:40.200 about data centers, but not in the way that you 0:01:40.240 --> 0:01:42.320 may think about them. I think oftentimes when we're thinking 0:01:42.319 --> 0:01:46.360 about data centers, were actually thinking of the intense energy demand. 0:01:46.520 --> 0:01:49.520 Immediately what comes to mind actually is bitcoin mining. But 0:01:49.600 --> 0:01:52.480 that is a conversation maybe for another show. But we're 0:01:52.520 --> 0:01:54.600 here today to talk about data centers as a force 0:01:54.760 --> 0:01:59.640 and source of flexible energy demand. So Sarah and Michael, 0:02:00.120 --> 0:02:04.040 you very much for joining today. Thanks data So let's 0:02:04.080 --> 0:02:07.520 start at the very beginning with what specifically we're talking about. 0:02:07.720 --> 0:02:10.160 What is a data center? Paint a picture in my 0:02:10.280 --> 0:02:13.800 mind of what this looks like. So a data center 0:02:14.000 --> 0:02:18.640 is a physical facility that organizations would use to house 0:02:18.720 --> 0:02:22.840 their applications and data. A data center's design is basically 0:02:23.040 --> 0:02:26.200 sort of based on a network of computing and storage 0:02:26.240 --> 0:02:31.400 resources that enable that delivery of the shared applications and data. 0:02:31.480 --> 0:02:33.920 And if you're interested, the kind of key components of 0:02:33.960 --> 0:02:39.560 a data center include routers, switches, firewalls, storage systems, servers, 0:02:39.680 --> 0:02:42.600 and all sorts of other computer resources. So I mean 0:02:42.680 --> 0:02:45.760 essentially everything that we interact with in the electronic world. 0:02:45.800 --> 0:02:48.920 The recording platform we're on right now, my Netflix, and 0:02:49.080 --> 0:02:52.079 the bitcoin that I don't own. All of those things 0:02:52.080 --> 0:02:54.760 are stored in a data center somewhere exactly. And the 0:02:54.840 --> 0:02:58.200 type of data center varies, right. So the main kind 0:02:58.200 --> 0:03:00.960 of two categories that we have and that we looked 0:03:00.960 --> 0:03:04.280 at in this report our co location and hyper scale 0:03:04.360 --> 0:03:07.120 data centers. And you may have heard both of those, 0:03:07.160 --> 0:03:08.960 and in case you're not entirely sure what they are. 0:03:09.000 --> 0:03:12.000 So a co location data center those built for the 0:03:12.000 --> 0:03:15.960 purpose of renting space for servers and other computing hardware 0:03:16.000 --> 0:03:18.480 to clients. So you know, you might get a company 0:03:18.520 --> 0:03:21.600 who says, right, we need one room housing x amount 0:03:21.639 --> 0:03:23.640 of servers for us, and then and next door to 0:03:23.680 --> 0:03:27.160 that a different company. A hyper scale data center often 0:03:27.200 --> 0:03:30.040 called a self build. These data centers are basically owned 0:03:30.040 --> 0:03:34.480 and operated by cloud service providers like AWS, Microsoft or 0:03:34.520 --> 0:03:38.040 Facebook or Apple. To put this in context, how energy 0:03:38.120 --> 0:03:41.480 intensive are these and how big is the scale of 0:03:41.920 --> 0:03:44.720 the energy demand here, So dass it is are very 0:03:45.120 --> 0:03:49.120 large buildings that consume huge amounts of electrical power, not 0:03:49.280 --> 0:03:51.560 only for the servers themselves actually I guess that we 0:03:51.680 --> 0:03:55.240 said all your our bitcoin and podcasts live upon, but 0:03:55.280 --> 0:03:59.120 then also the supporting so the exporting service like the 0:03:59.160 --> 0:04:01.760 cooling for the building as well, which is quite a 0:04:01.800 --> 0:04:05.520 significant demand because they can overheat, right because their energy intensive, 0:04:05.560 --> 0:04:08.080 they get warm. Yeah, exactly, there's lots lots of really 0:04:08.120 --> 0:04:11.320 cool things looking at the efficiency of cooling for these 0:04:11.360 --> 0:04:14.320 buildings to really though that's only part of it for 0:04:14.480 --> 0:04:17.400 most of the demand, Like if the cooling partment makes 0:04:17.440 --> 0:04:21.360 about of overall electrical demand and the data center, most 0:04:21.360 --> 0:04:25.120 of that computing power and energy goes to doing the 0:04:25.160 --> 0:04:29.680 applications and hosting the data itself. So these are actually 0:04:29.680 --> 0:04:33.039 really sophisticated energy resources. They have their own on site 0:04:33.160 --> 0:04:38.320 backup power generators, they have their own batteries on site 0:04:38.360 --> 0:04:41.279 as well, so they're really interesting. Legs actually can be 0:04:41.880 --> 0:04:44.599 not just connecting to the grids and causing more and 0:04:44.680 --> 0:04:48.679 more issues with supply or with congestion, But I should 0:04:48.880 --> 0:04:51.960 be more of this solution, you know, like how they 0:04:51.960 --> 0:04:56.480 can interact and be more responsive to what the grid needs. Okay, 0:04:56.520 --> 0:04:58.960 so we've established that they're definitely a source of energy 0:04:59.000 --> 0:05:01.919 demand and will to the flexibility in a second, but 0:05:02.040 --> 0:05:04.000 just before we get too deep into that, and I 0:05:04.040 --> 0:05:06.640 think it matters for us to point out which countries 0:05:06.839 --> 0:05:08.479 in kind of which part of the world you guys 0:05:08.480 --> 0:05:10.520 specifically looked at, because this is a bit of a 0:05:10.560 --> 0:05:13.000 case study. It seems like everything is a bit unique here. 0:05:13.120 --> 0:05:15.640 So we've done a case study on some specific countries 0:05:15.640 --> 0:05:19.039 for some specific reasons which will hopefully have applicability to 0:05:19.120 --> 0:05:21.479 other parts of the world. And maybe you could shed 0:05:21.520 --> 0:05:25.279 some light on that. What did we look at? Yeah, so, Danna, 0:05:25.360 --> 0:05:28.359 you might be familiar with the flat D markets. I 0:05:28.360 --> 0:05:30.560 don't know if you've come across that. No, I have 0:05:30.680 --> 0:05:33.680 not come across. Said so, most data center capacity in 0:05:33.720 --> 0:05:39.240 Europe is concentrated in five countries Germany, UK, France, Ireland 0:05:39.360 --> 0:05:42.600 and the Netherlands, and that kind of capacity is almost 0:05:42.640 --> 0:05:45.000 in five of those cities, and that's referred to as 0:05:45.080 --> 0:05:47.880 flat D. So these markets are the largest for a 0:05:47.920 --> 0:05:50.919 bunch of reasons, Like Frankfurt, London and Amsdam obviously have 0:05:51.040 --> 0:05:54.520 the largest Internet exchanges Dublin is a data center hubs 0:05:54.520 --> 0:05:57.880 specifically for you know the history of large technology companies 0:05:58.000 --> 0:06:02.599 headquartering in Ireland. But we actually decided to yes, focus 0:06:02.640 --> 0:06:06.360 on the UK, Ireland, Germany and the Netherlands, which, as 0:06:06.480 --> 0:06:08.279 you guys just heard, you know, they're four of the 0:06:08.360 --> 0:06:12.240 largest markets. But we also chose to focus on Norway, 0:06:12.320 --> 0:06:15.760 which is a very sort of emerging market could potentially, 0:06:16.200 --> 0:06:18.359 you know, in the future accommodate a lot of green 0:06:18.440 --> 0:06:21.520 and flexible data centers. Does every country have their own 0:06:21.560 --> 0:06:25.560 market or do they sometimes by this data center space 0:06:25.680 --> 0:06:29.880 from another country When we talk about European data centers, 0:06:29.920 --> 0:06:32.880 A lot of that is really in these five countries. 0:06:33.279 --> 0:06:35.520 I mean we we are seeing more sort of data 0:06:35.520 --> 0:06:38.359 centers cropping up in Italy and Croatia and places that 0:06:38.400 --> 0:06:42.240 traditionally weren't massive hubs for data centers. And I think 0:06:42.279 --> 0:06:44.600 also a lot of the newer places that are seeing 0:06:44.760 --> 0:06:46.800 a lot more build our due to you know, more 0:06:46.920 --> 0:06:50.480 land being available and milder climate, perhaps near more solar 0:06:50.520 --> 0:06:53.640 and wind resources. So what about the future of this 0:06:53.880 --> 0:06:56.520 space in terms of demand? Because I don't see myself 0:06:56.560 --> 0:06:59.760 deleting any of my children's baby pictures anytime soon they're 0:06:59.800 --> 0:07:02.640 going to stay on that cloud server forever. Is this 0:07:02.760 --> 0:07:05.080 growing rapidly? Is this one of those kind of up 0:07:05.080 --> 0:07:08.080 into the right or almost straight up energy demand stories. 0:07:08.720 --> 0:07:12.000 So in this report, our actual goal here in one 0:07:12.080 --> 0:07:13.760 of the you know, one section of the report was 0:07:13.840 --> 0:07:17.880 to actually extrapolate and understand what demand would look like 0:07:18.000 --> 0:07:20.560 in twenty thirty. So we kind of looked at current 0:07:20.640 --> 0:07:24.680 data center power demand in one data center actricity demand 0:07:24.800 --> 0:07:27.960 and you know, project that out until twenty thirty, and 0:07:28.040 --> 0:07:31.000 kind of some of our conclusions here were that this 0:07:31.520 --> 0:07:34.720 level of growth, you know, is not slowing. You know 0:07:34.760 --> 0:07:37.600 that there is obviously uncertainty around how much data centers 0:07:37.600 --> 0:07:40.800 in Europe will grow. So for the twenty thirty forecast, 0:07:40.880 --> 0:07:44.240 we actually built three different scenarios. So we built a low, 0:07:44.600 --> 0:07:47.880 a medium, and a high growth and these scenarios assume 0:07:47.960 --> 0:07:51.080 sort of different compound downual growth rates, taking into account 0:07:51.080 --> 0:07:53.000 a lot of other things within the data centers to 0:07:53.080 --> 0:07:58.480 reflect those uncertainties regarding planning, permission, constrained land options, available 0:07:58.480 --> 0:08:01.800 power connections. So it's definitely a difficult task in all 0:08:01.840 --> 0:08:05.160 about scenarios. You know, data centers do grow across all 0:08:05.240 --> 0:08:07.760 five of those regions. Since I was kind of under 0:08:07.880 --> 0:08:10.120 sun nerves less slightly, there should a huge amount of 0:08:10.120 --> 0:08:15.320 work of building from a database of how much capacity 0:08:15.480 --> 0:08:18.640 of data centers exist in these five countries, and we 0:08:18.840 --> 0:08:21.320 use that they're actually too while amount of capacity in 0:08:21.360 --> 0:08:24.360 each Markish could be the same, are are less even, 0:08:24.360 --> 0:08:26.520 but actually the impact they have on the grid is 0:08:26.800 --> 0:08:29.240 really relevant to the size of the grid that they're 0:08:29.520 --> 0:08:31.920 connected to. There's a big range there. So m I say, 0:08:31.960 --> 0:08:35.200 like like Ireland and UK might have similar amounts of 0:08:35.360 --> 0:08:39.040 data center capacity connected while in the UK doesn't just 0:08:39.120 --> 0:08:41.320 make up about two and a half percent of the 0:08:41.320 --> 0:08:45.400 amount of electricity the UK consumes. In Ireland, data centers 0:08:45.480 --> 0:08:49.800 today consumed by fourteen or fift of all of Ireland's 0:08:49.880 --> 0:08:53.800 electrical generations. So the big gap between the impacts that 0:08:53.840 --> 0:08:56.360 the data centers can have on grids. Okay, so they 0:08:56.400 --> 0:08:58.120 have an impact, and I think we've set this up. 0:08:58.160 --> 0:09:01.360 Let's go to the flexible part of the demand. What 0:09:01.559 --> 0:09:05.160 is the opportunity here because I'm thinking as we talk 0:09:05.200 --> 0:09:07.400 about flexible demand, we're not going to be able to 0:09:07.400 --> 0:09:10.400 take them offline. This is there's a security of supply 0:09:10.520 --> 0:09:14.480 issue here. So what are the choices available to data 0:09:14.480 --> 0:09:17.880 centers to build that flexibility. I mean, this is kind 0:09:17.880 --> 0:09:19.280 of the crux of it, and I guess what we 0:09:19.320 --> 0:09:22.200 really wanted to dive into with the reports. When you're 0:09:22.200 --> 0:09:27.120 looking at a data center, there's several sources of flexibility available, 0:09:27.200 --> 0:09:28.719 so you can have the load. So actually if that 0:09:28.920 --> 0:09:31.520 is the applications and the data's going on, and you 0:09:31.559 --> 0:09:33.439 could shift that by time, you know, so if you've 0:09:33.440 --> 0:09:35.760 got something that doesn't have to happen at two o'clock 0:09:35.760 --> 0:09:37.720 in the day, and you can actually do that too 0:09:37.880 --> 0:09:41.439 a m at night, let's shift that also if something 0:09:41.480 --> 0:09:45.599 can happen in Germany, especially the German grids really congested 0:09:45.640 --> 0:09:49.280 at the moment, how we can shift that computational task 0:09:49.760 --> 0:09:52.520 to France where actually there's a it's a far more 0:09:52.600 --> 0:09:55.280 relaxed grid. What is the load you can shift us? 0:09:55.360 --> 0:09:57.520 As I was saying earlier, is a lot of electrical 0:09:57.559 --> 0:10:00.679 equipment on site, so you have the up Yes, so 0:10:00.760 --> 0:10:04.480 that's the uninterruptible power supply. So this is a battery 0:10:04.520 --> 0:10:08.040 that pretty much every data center will have that if 0:10:08.080 --> 0:10:11.440 the grid goes down, which is a huge issue for 0:10:11.880 --> 0:10:14.160 data stays, as you said, das do not want to 0:10:14.160 --> 0:10:16.880 go down they get paid to be online when we 0:10:16.920 --> 0:10:19.120 want to have our podcasts, and we all get very 0:10:19.160 --> 0:10:22.920 upset when our podcasts aren't there. I certainly do. So 0:10:23.480 --> 0:10:26.160 they have what they call the UPS and the power 0:10:26.240 --> 0:10:29.920 supply that will come on instantly. The power supply goes 0:10:29.960 --> 0:10:32.439 down and we'll run for about five ten minutes until 0:10:32.559 --> 0:10:36.200 the backup generators come online. Because it's like starting up 0:10:36.240 --> 0:10:38.160 your card, you have to turn the engine on, it 0:10:38.200 --> 0:10:40.520 takes the while for the engine to warm up. So 0:10:40.760 --> 0:10:43.800 UPS is a battery that can jump in. Every dozener 0:10:43.840 --> 0:10:46.720 has a backup generator. The problem is most of those 0:10:46.760 --> 0:10:50.480 are diesel. We don't want to use diesels for flexibility. Ideally, 0:10:50.760 --> 0:10:54.240 these backup generors will never be run. There are alternative options. 0:10:54.320 --> 0:10:57.400 You could use hydrogen, new options that if those were installed, 0:10:57.800 --> 0:11:00.679 maybe we can actually start to use that backup generation 0:11:01.000 --> 0:11:04.200 to do more flexible things with it. Well, so first 0:11:04.200 --> 0:11:08.400 of all, how much flexibility are you talking about? Minutes, hours, days? 0:11:08.960 --> 0:11:11.959 What's the duration of time that's being proposed here. It 0:11:11.960 --> 0:11:14.280 will all be quite short. I think it does vary 0:11:14.440 --> 0:11:17.400 on those different aspects of the load shifting. The UPS 0:11:17.440 --> 0:11:20.839 and the backup generation. The ups to be very short 0:11:21.240 --> 0:11:25.040 we're like, we're talking it's seconds to two minutes. Also, 0:11:25.120 --> 0:11:29.120 does centerpposers wary of using the UPS to do these 0:11:29.120 --> 0:11:31.720 interesting things for the grid because it's a backup plan, 0:11:31.840 --> 0:11:33.440 then you don't have a backup plan for the backup 0:11:33.480 --> 0:11:36.600 plan exactly. So if there's a lot of hesitation, and 0:11:36.640 --> 0:11:39.200 we can definitely talk about that in the future. But 0:11:39.440 --> 0:11:43.080 were those actually those ups can work really well? Is 0:11:43.120 --> 0:11:46.199 to provide frequency response to the grid. This is a 0:11:46.320 --> 0:11:48.400 very big topic get too. But you can think about 0:11:48.440 --> 0:11:52.480 frequency response as like it's the as frequency as the 0:11:52.520 --> 0:11:55.520 heartbeat of the grid. So the grid needs to need 0:11:55.600 --> 0:11:59.400 suppose a fifty hurts, so that's fifty times second. And 0:11:59.640 --> 0:12:02.280 how the UPS could work with like a pace maker. 0:12:03.200 --> 0:12:05.240 When the fifty hurts goes a little bit too higher, 0:12:05.240 --> 0:12:09.240 a little bit too low, the UPS the pacemaker kind 0:12:09.240 --> 0:12:11.520 of gives it a shock to keep the hurts, to 0:12:11.600 --> 0:12:14.319 keep the heartbeat going at that healthy point all of 0:12:14.440 --> 0:12:16.320 fifty hurts. So that's kind of one of the main 0:12:16.480 --> 0:12:19.440 uses we saw. So let's say the backup generation was 0:12:19.480 --> 0:12:21.760 built into the system with the intention to use it, 0:12:21.800 --> 0:12:24.319 and you mentioned a number of solutions. Okay, so bio 0:12:24.360 --> 0:12:27.800 fuels one of them. I definitely am seeing this coming 0:12:27.920 --> 0:12:31.240 very much back into the lexicon after a decade of 0:12:31.280 --> 0:12:33.320 not being talked about quite as much. But then you 0:12:33.360 --> 0:12:36.640 mentioned to other things batteries and hydrogen, which as of 0:12:36.760 --> 0:12:42.839 right now are very expensive sources of energy. Is there 0:12:42.880 --> 0:12:48.880 a point at which they become economically viable because the 0:12:48.880 --> 0:12:51.120 flexible demand is kicking in at a time when the 0:12:51.160 --> 0:12:53.840 grid really needs it, so the pricing signal is there 0:12:54.200 --> 0:12:57.800 or will it always be a commitment to let's say, 0:12:57.800 --> 0:13:01.840 certain emissions profiles as opposed is too Actually I don't know. 0:13:02.000 --> 0:13:05.199 I mean, what's what's the what's the incentive? So you're 0:13:05.240 --> 0:13:07.959 hitting on a great point of the incentives and I 0:13:08.040 --> 0:13:09.719 definitely want to get to that in terms of these 0:13:09.760 --> 0:13:14.000 two technologies, so like batteries and hydrogen, I think to 0:13:14.160 --> 0:13:16.920 focus on those. There's only one data center in the 0:13:16.960 --> 0:13:20.720 world that we know that is even at the trial 0:13:20.840 --> 0:13:25.000 stage of using a large battery as backup Google as 0:13:25.040 --> 0:13:28.240 a data center in Belgium. So that's really interesting. See 0:13:28.280 --> 0:13:31.240 they're already thinking about the next step. Microsoft is pretty 0:13:31.320 --> 0:13:34.520 keen on hydrogen. It's running a couple of projects at 0:13:34.559 --> 0:13:38.480 the moment, and Microsoft actually set a goal to be 0:13:38.679 --> 0:13:42.800 diesel free by twenty thirty. So there's an end goal there, 0:13:42.840 --> 0:13:44.960 and it's gonna be interesting, Actually, what are the options 0:13:44.960 --> 0:13:47.440 they're going to use to try and hit that goal? Because, 0:13:47.880 --> 0:13:51.199 as we're saying a number, but it's the vast majority 0:13:51.480 --> 0:13:55.680 of backup generation is I think that green incentive might 0:13:55.760 --> 0:13:58.840 be one of the best pushes we have to tapping 0:13:58.920 --> 0:14:03.080 into that data center flexibility. And you're seeing these these 0:14:03.080 --> 0:14:06.160 two leaders of Google and Microsoft. You know, they're starting 0:14:06.160 --> 0:14:09.520 to talk about seven remove of energy matching. You know, 0:14:09.559 --> 0:14:11.200 shifting your load is going to make that a whole 0:14:11.200 --> 0:14:13.319 lot easier if you can do that. And are they 0:14:13.520 --> 0:14:16.319 sitting in the benchmark how corporations are going to think 0:14:16.360 --> 0:14:19.000 about the energy procure image? Eliot at that benchmark ten 0:14:19.080 --> 0:14:22.440 years ago with p P A S. They setting that 0:14:22.480 --> 0:14:26.800 benchmark again today with this twenty seven renewable energy matching 0:14:27.240 --> 0:14:29.880 that other corporations will have to keep up with and 0:14:29.960 --> 0:14:32.560 years to come. Now for a very short break, stay 0:14:32.600 --> 0:14:40.120 with us out of curiosity the companies that are operating 0:14:40.120 --> 0:14:43.080 and running these data centers. I mean, you mentioned Microsoft 0:14:43.440 --> 0:14:47.120 and Google and companies that I know have net zero targets, 0:14:47.240 --> 0:14:50.120 in some cases very ambitious ones. We maybe even negative 0:14:50.440 --> 0:14:53.640 missions targets. But what about the rest of the industry. 0:14:53.720 --> 0:14:57.240 Is there a high number of companies with net zero 0:14:57.280 --> 0:14:59.200 targets in this I don't don't have been called too 0:14:59.240 --> 0:15:01.960 many edge It's on fair because obviously those those companies 0:15:02.000 --> 0:15:05.440 have been so public with it. You, i think, said Sarah, 0:15:05.480 --> 0:15:08.560 spoke through a number of these co location providers as well. 0:15:08.600 --> 0:15:11.280 But this is actually gets pretty difficult between let's say, 0:15:11.280 --> 0:15:14.160 the hyper scalers of those cloud operators we know well 0:15:14.560 --> 0:15:17.400 and these co location companies because on the one hand, 0:15:17.440 --> 0:15:21.200 the hypersco operators, they are very well resourced, you know, 0:15:21.240 --> 0:15:23.800 some of the wealthiest companies in the world, and the 0:15:23.800 --> 0:15:26.840 others had you had these co location operators who operate 0:15:26.920 --> 0:15:30.840 the majority of data centers we see, but they are 0:15:30.880 --> 0:15:33.840 not as well resourced. I think Equinics were the largest 0:15:33.960 --> 0:15:36.640 data center collocator operators has a market cap of by 0:15:36.720 --> 0:15:40.600 thirty billion, as opposed to Google I think, which is 0:15:40.600 --> 0:15:44.520 at one point five trillion. Amazons around that two trillion 0:15:44.560 --> 0:15:47.480 mark as well, So it's kind of a it's difficult 0:15:47.560 --> 0:15:51.440 going to compare them. Also, the big companies have far 0:15:51.480 --> 0:15:54.520 more control over what they can do, like they own 0:15:54.880 --> 0:15:58.280 all their own data centers, they decide what servers that 0:15:58.400 --> 0:16:00.640 go into them. They know the ass that are going 0:16:00.680 --> 0:16:03.640 through them, what their users are using it. For collocation 0:16:03.680 --> 0:16:06.760 oppertors don't have that visibility. You know, companies come in 0:16:06.760 --> 0:16:09.400 to build their servers in there. They have to serve 0:16:09.440 --> 0:16:11.760 their customers where it's called an s l A, a 0:16:11.880 --> 0:16:15.400 service level agreement, which can be quite restrictive to what 0:16:15.560 --> 0:16:18.760 these co location companies can do. Yeah, and I think 0:16:18.760 --> 0:16:21.200 just to add to that, exactly what Michael said, it 0:16:21.320 --> 0:16:23.840 is a lot easier I think for hyper scale as 0:16:23.960 --> 0:16:26.920 compared to co location. That's another reason why we looked 0:16:26.920 --> 0:16:29.640 at Norway as an emerging market because of a lot 0:16:29.680 --> 0:16:31.520 of it's you know, a lot of the integration and 0:16:31.560 --> 0:16:34.960 renewables in Norway, and there's a provider I think Green Mountain. 0:16:35.360 --> 0:16:39.040 Green Mountain uses add percent renewable power because obviously one 0:16:39.080 --> 0:16:41.080 of the main challenges in the data center industry is 0:16:41.200 --> 0:16:44.200 ensuring continual cooling throughout the day, and one of the 0:16:44.240 --> 0:16:47.840 benefits of being somewhere like Norway is you know, you 0:16:47.880 --> 0:16:50.760 can they can utilize some of the cold waters from 0:16:50.800 --> 0:16:54.080 the deep Norwegian areas located adjacent to the facilities of 0:16:54.120 --> 0:16:57.000 the data centers, which again have so much land. So 0:16:57.040 --> 0:17:00.840 I think that we are seeing more and more colocation 0:17:01.360 --> 0:17:06.159 providers and operators, you know, prioritize net zero simultaneously with 0:17:06.240 --> 0:17:09.640 the hyper scalers. What's the policy environment for these data 0:17:09.680 --> 0:17:14.000 centers and our countries welcoming them in or they something 0:17:14.040 --> 0:17:17.000 where they're saying, you know, literally my grid has so 0:17:17.119 --> 0:17:19.320 much demand, You're going to have to find somewhere else. 0:17:19.600 --> 0:17:21.439 So I mean I can just speak to sort of 0:17:21.480 --> 0:17:26.360 Ireland Breedflely. Obviously, growth in Ireland, you know, you would expect, 0:17:26.359 --> 0:17:29.040 typically as someone who knows nothing about data centers, you know, 0:17:29.080 --> 0:17:31.520 out of those five regions, you'd immediately be like, oh, 0:17:31.600 --> 0:17:34.119 you know, we expect Ireland to be have the biggest growth, 0:17:34.400 --> 0:17:36.840 just because we know that you know all the aws 0:17:36.920 --> 0:17:40.080 and you know Google's data centers. But actually it might 0:17:40.080 --> 0:17:41.639 not be as rapid as it has been in the 0:17:41.680 --> 0:17:45.080 last few years. And that's that's really due to Ireland's 0:17:45.160 --> 0:17:48.440 recent sort of policy proposals, and some of those include 0:17:48.920 --> 0:17:53.520 limiting new data center build unless certain flexibility requirements are met. 0:17:54.119 --> 0:17:55.919 You know, there are a lot of pushbacks from Irish 0:17:55.920 --> 0:17:59.040 citizens who are concerned about power at climate issues, and 0:17:59.040 --> 0:18:01.920 I think unlike somewhere like Germany, most data centers in 0:18:01.960 --> 0:18:05.760 Ireland are really concentrated in one major city, Dublin, and 0:18:05.840 --> 0:18:09.960 Dublin's electricity grid was not built to cater for such 0:18:10.040 --> 0:18:13.040 high demand from data centers, which I think has resulted 0:18:13.080 --> 0:18:16.520 in network supply constraints. Michael, and sure you can speach 0:18:16.560 --> 0:18:18.760 this little bit more, but you know, in June last year, 0:18:19.200 --> 0:18:22.280 the Commission for Regulation of Utility is the CRU proposed 0:18:22.280 --> 0:18:27.000 that the Irish network operators a grid and ESP networks 0:18:27.119 --> 0:18:29.679 and which are kind of the largest, proposed that they 0:18:29.720 --> 0:18:34.200 prioritize data center connection requests outside of network constrained areas, 0:18:34.720 --> 0:18:37.919 which means there was a lot of uncertainty around you know, 0:18:38.119 --> 0:18:42.040 estimating Ireland growth, especially for the future of thirty or 0:18:42.080 --> 0:18:45.600 more proposed data centers in and around Dublin Islands. Obviously 0:18:46.160 --> 0:18:49.119 is we can tell by accident is quite close to 0:18:49.600 --> 0:18:53.880 my heart and Ireland is dozens in Ireland's consume four 0:18:54.760 --> 0:18:58.959 of the car contre's entire electricity consumption. Our estimates are 0:18:59.040 --> 0:19:02.560 twenty thirty, which actually aligned with what are grid the 0:19:02.800 --> 0:19:07.439 grid operator is looking at, is about like a quarter 0:19:07.800 --> 0:19:12.840 of all Irish electricity consumption will be data centers. That's 0:19:12.840 --> 0:19:16.120 a big number, just to give a scale. In Germany, 0:19:16.200 --> 0:19:19.720 which is the largest data center capacity with a much 0:19:19.800 --> 0:19:23.119 charger grid, that numbers about one point five percent, So 0:19:23.600 --> 0:19:27.520 doesn't Actually they are scaring the Irish grid operator and 0:19:27.680 --> 0:19:30.800 there have been quite drastic proposals that Sarah just mentioned, 0:19:30.840 --> 0:19:33.520 but that's slightly different to actually what we're seeing in 0:19:33.560 --> 0:19:37.120 other regions. So in Ireland it's actually a concern of 0:19:37.160 --> 0:19:39.919 security of supply. The Irish grid will not be able 0:19:40.080 --> 0:19:43.720 to meet you know, will not be able to meet 0:19:43.840 --> 0:19:47.080 all of the demand as well as meeting gridade energy targets, 0:19:47.080 --> 0:19:51.480 which is no electricity by twenty thirty. What we see 0:19:51.480 --> 0:19:55.240 in other regions, the most notably Abstam, which introduced a 0:19:55.400 --> 0:19:57.560 ban for an entire year. They said you could not 0:19:57.720 --> 0:20:01.119 build any more new data center is They have never 0:20:01.240 --> 0:20:04.040 lifted that, but said you can only build a certain 0:20:04.080 --> 0:20:08.400 amount by thirty and they have to be in these 0:20:08.440 --> 0:20:11.199 particular regions. Fand first is talking to a bunch of 0:20:11.200 --> 0:20:14.639 proposals as well to like forced data centers to build 0:20:14.640 --> 0:20:17.920 in particular areas, and I think that's maybe some of 0:20:17.960 --> 0:20:21.280 the lessons to learn for like emerging regions that are 0:20:21.320 --> 0:20:24.080 looking to attract data centers is to think a bit 0:20:24.119 --> 0:20:26.960 more about where these data centers are going to sit 0:20:27.280 --> 0:20:29.880 on the grid. And it's actually the issues are not 0:20:30.000 --> 0:20:32.840 just purely the grid and the capacity of the grid 0:20:33.000 --> 0:20:34.960 locally to take it, but also the amount of land. 0:20:35.080 --> 0:20:38.159 You know, the dozens are very really big things, and 0:20:38.200 --> 0:20:40.000 they take up a lot of land that could be 0:20:40.080 --> 0:20:44.639 used for you know, warehouses or offices. Reneable energy are 0:20:44.760 --> 0:20:47.320 exactly the actually it weren't the rules proposed. And find 0:20:47.320 --> 0:20:49.640 first is that that sid just can't be located near 0:20:50.280 --> 0:20:53.520 public transport hubs because they don't actually have many people. 0:20:54.119 --> 0:20:56.440 So they actually would rather push, you know, a warehouse 0:20:56.480 --> 0:20:58.879 that has a lot of staff, put that near the 0:20:58.960 --> 0:21:02.600 public transport hub. And what is the hourly demand profile? 0:21:02.640 --> 0:21:04.560 Look for them? Is it constant or does it have 0:21:04.600 --> 0:21:08.000 a peak? It's flash, it's a flat so so there 0:21:08.040 --> 0:21:10.760 you it's pancake, but the rest of the grid is not. 0:21:10.960 --> 0:21:14.199 And you know, I'm guessing many of the people listening, 0:21:14.200 --> 0:21:17.359 although not all, know that when people come home in 0:21:17.400 --> 0:21:20.359 the evening on a regular day, now that many people 0:21:20.359 --> 0:21:22.440 are going back to work and going about their days 0:21:22.680 --> 0:21:25.000 a bit more than they maybe had been in the 0:21:25.080 --> 0:21:29.960 last two years, you see this very peak and trough 0:21:30.160 --> 0:21:33.199 throughout the day in terms of demand profile. So data centers, 0:21:33.240 --> 0:21:36.680 if they're able to take some of their demand offline, 0:21:36.760 --> 0:21:40.359 presumably in the evening when everybody gets home, they're sort 0:21:40.400 --> 0:21:43.640 of solving part of the demand that they take. They're 0:21:43.680 --> 0:21:48.120 at least solving part of the demand that they are 0:21:48.359 --> 0:21:51.200 adding to the grid. They're reducing the aggregate demand added 0:21:51.240 --> 0:21:53.240 to the grid if they're reducing that peak at certain 0:21:53.280 --> 0:21:56.040 points in the day, because the alternative, I suppose it's 0:21:56.080 --> 0:21:59.480 just to build more energy even with the diesel generators. 0:22:00.240 --> 0:22:03.639