WEBVTT - KKR on 'Enormous' AI-Driven Data Center Demand 0:00:02.600 --> 0:00:06.800 You're listening to Asia Centric from Bloomberg Intelligence, the podcast 0:00:06.840 --> 0:00:10.280 that explores the big ideas and trends moving money across 0:00:10.320 --> 0:00:12.760 the region. I'm John Lee in Hong Kong. 0:00:12.720 --> 0:00:15.320 And I'm Katie Dmitrieva, also in Hong Kong. 0:00:15.720 --> 0:00:19.760 Katya. When investors think of AI, they always think of Nvidia. 0:00:19.880 --> 0:00:21.880 That did you know there was a stock last year 0:00:22.040 --> 0:00:25.480 called Vertive Holdings that went up four times more than 0:00:25.520 --> 0:00:26.239 in Vidia. 0:00:26.360 --> 0:00:27.319 What does the company do? 0:00:27.840 --> 0:00:30.240 They make cooling systems for data centers. 0:00:30.560 --> 0:00:33.280 Cooling systems, so not the data centers themselves, but the 0:00:33.320 --> 0:00:36.440 infrastructure for the data centers. Imagine there's a lot of 0:00:36.479 --> 0:00:39.599 investment opportunity there. You know, I was reading the Moody's 0:00:39.600 --> 0:00:43.960 report recently and in this region in particular, data center 0:00:44.040 --> 0:00:47.760 capacity is set to double just the next four years. 0:00:47.760 --> 0:00:50.360 So you can think about how much opportunity is there. 0:00:50.880 --> 0:00:53.320 And that's actually what we're going to be discussing today. 0:00:53.520 --> 0:00:57.200 We have a guest today, Projesh Banerjee, who's the director 0:00:57.240 --> 0:00:59.680 of Infrastructure at KKR, joining us. 0:01:00.200 --> 0:01:03.440 Hello Projesh, thank you, thanks very much for having me. 0:01:04.520 --> 0:01:08.080 I wanted to ask you, is there a race to 0:01:08.440 --> 0:01:10.800 build data centers right now? 0:01:10.840 --> 0:01:12.560 Across Asia. What does that look like? 0:01:14.160 --> 0:01:19.920 Eris is probably a strong representation. There's certainly enormous demand 0:01:20.000 --> 0:01:24.280 for data centers. There are really two major trends that 0:01:24.319 --> 0:01:27.520 are driving these volumes. One that's been playing out for 0:01:27.560 --> 0:01:30.880 a long time, and that's just the continued migration of 0:01:31.040 --> 0:01:34.880 data to the cloud from on site servers, and this 0:01:35.080 --> 0:01:39.000 trend has been a key driver of DC capacity growth 0:01:39.000 --> 0:01:42.080 over the last few years. Globally. ISSHA is playing catch 0:01:42.120 --> 0:01:44.240 up on this front as well. And the new trend, 0:01:44.280 --> 0:01:46.640 of course is AI and I'd say we're probably seeing 0:01:46.640 --> 0:01:49.600 this a little more in the West, in North America 0:01:49.640 --> 0:01:53.200 and increasingly in Europe than we are in Asia at 0:01:53.240 --> 0:01:56.200 the moment, but it's almost certain to play out here 0:01:56.400 --> 0:01:58.520 over the next few heres as well. And this is 0:01:58.600 --> 0:02:01.960 driving another step change in DEMI. So yes, there is 0:02:02.000 --> 0:02:05.080 an enormous amount of demand and activity across is in 0:02:05.120 --> 0:02:06.160 the DC space at the moment. 0:02:07.000 --> 0:02:10.640 And PROTESSHA, can we take a step backwards? What exactly 0:02:10.960 --> 0:02:14.440 is a data center? I think people have some preconceptions, 0:02:14.480 --> 0:02:16.160 but I'd love to hear your explanation. 0:02:16.440 --> 0:02:17.360 Yeah, what do they look like? 0:02:20.000 --> 0:02:24.399 Well, thankfully they aren't particularly eye catching. Something's gone wrong 0:02:24.400 --> 0:02:27.400 if someone's invested a lot in the architecture and then 0:02:27.400 --> 0:02:28.960 you sort of drive by and you go, ooh, this 0:02:29.000 --> 0:02:31.760 is a tourist attraction. Hopefully these are buildings that you 0:02:31.880 --> 0:02:35.040 drive past and never really realize are there. They're just 0:02:35.320 --> 0:02:39.639 very large structures with a lot of different data holes. 0:02:39.680 --> 0:02:43.000 Just think of them as cold dark rooms, very purpose 0:02:43.080 --> 0:02:46.480 built for the most part to house a huge number 0:02:46.560 --> 0:02:52.240 of servers and digital infrastructure where data is stored, very 0:02:52.360 --> 0:02:57.560 very specifically built from a power water cooling perspective to 0:02:57.560 --> 0:03:00.600 be energy efficient or as an energy efficient dispossible and 0:03:01.760 --> 0:03:05.240 essentially maximize space and minimize power usage to the extent 0:03:05.320 --> 0:03:10.120 possible to keep the service cool. So a physical structure 0:03:10.440 --> 0:03:14.440 recently nondescript building which is basically full of goal of 0:03:14.520 --> 0:03:16.519 dark rooms which house service. 0:03:17.400 --> 0:03:21.520 Yeah, you mentioned the water and electricity component. Is there 0:03:21.600 --> 0:03:24.519 kind of a trend and where these things are located 0:03:24.560 --> 0:03:28.480 and also does that kind of impact your investment outlook 0:03:28.520 --> 0:03:31.160 and where these data centers are actually going to get built? 0:03:32.200 --> 0:03:36.120 Yes, there is, and it gets a little nuanced because 0:03:36.160 --> 0:03:38.200 it goes into the type of data centers and they're 0:03:38.240 --> 0:03:42.720 you know, historically say three types of data centers. So 0:03:42.760 --> 0:03:45.400 there are some that focused on interconnects, so these are 0:03:45.560 --> 0:03:48.640 more carrier hotels where there's just a lot of cross 0:03:48.680 --> 0:03:53.880 connectivity between people who house equipment in those data centers. 0:03:53.880 --> 0:03:56.680 The same there's enterprise data centers, which is a little 0:03:56.720 --> 0:04:00.560 more focused on enterprise clients. And then there's what is 0:04:00.680 --> 0:04:03.360 driving a whole lot of growth and capacity, and that 0:04:03.440 --> 0:04:07.000 is the hyper scale data centers, which is where you 0:04:07.040 --> 0:04:10.800 see this week step change in demand. That's more AI driven. 0:04:11.160 --> 0:04:14.000 Obviously from a cloud computing perspective, those are the types 0:04:14.040 --> 0:04:17.799 of clients you know we speak with as well, and 0:04:18.560 --> 0:04:24.520 from a location perspective, the hyper scale side gets quite 0:04:24.720 --> 0:04:29.080 technical around the client's network architecture as well. So if 0:04:29.120 --> 0:04:34.600 you think about how most companies architect their networks, to say, 0:04:34.680 --> 0:04:39.760 Amazon for example, has what it calls a cloud region, 0:04:39.960 --> 0:04:43.160 and when it opens cloud regions in certain countries, there 0:04:43.240 --> 0:04:46.440 are multiple availability zone so think of these is just 0:04:46.480 --> 0:04:49.800 basically zones that have multiple different data centers with their 0:04:49.800 --> 0:04:55.320 own power, cooling, physical security, connected through very low latency 0:04:55.800 --> 0:05:01.280 and redundant fiber networks as well, and cloned so that 0:05:01.360 --> 0:05:04.320 if you one data center goes down, hopefully the end 0:05:04.320 --> 0:05:09.080 customer doesn't experience an outage. So yes, absolutely. When we 0:05:09.320 --> 0:05:12.839 look at investing in data centers, we think very carefully 0:05:12.839 --> 0:05:15.880 through the type of data center and whether or not 0:05:16.200 --> 0:05:19.799 specific parcels of land from a development perspective are located 0:05:19.880 --> 0:05:23.520 within these zones and regions where the customers are building 0:05:23.520 --> 0:05:26.479 out their own network architecture. And if you do have 0:05:27.320 --> 0:05:31.240 you know the sites that are flood proof, connected to 0:05:31.279 --> 0:05:36.840 the utilities, with access to power transmission so substations and 0:05:36.880 --> 0:05:41.120 transmission capacity as well, then they can be frankly as 0:05:41.200 --> 0:05:44.760 valuable as called. They're very very attractive pieces of land 0:05:44.760 --> 0:05:45.239 at the moment. 0:05:46.960 --> 0:05:50.880 Asia Centric is produced by Bloomberg Intelligence, where more than 0:05:50.920 --> 0:05:55.000 five hundred experienced analysts and strategists work around the clock 0:05:55.080 --> 0:05:59.120 to bring you timely, world class research. Our coverage spans 0:05:59.160 --> 0:06:03.960 two hundred market indices, currencies, commodities and industries, as well 0:06:03.960 --> 0:06:07.240 as over two thousand equities and credits. If you like 0:06:07.279 --> 0:06:11.640 what you hear, don't forget to subscribe and chirm and progress. 0:06:11.680 --> 0:06:18.320 You mentioned the hyperscalers like Amazon, Google, Microsoft, Where are 0:06:18.360 --> 0:06:21.159 the hyper scalers and your cells? Which countries are seeing 0:06:21.200 --> 0:06:24.800 the biggest demand for creating these data centers? 0:06:25.760 --> 0:06:31.760 So honestly, there are different drivers across regions. In markets 0:06:31.920 --> 0:06:36.719 like frankly, in almost every market across Asia, there's an 0:06:36.720 --> 0:06:40.000 almost demand. India, for example, I think it's forecast to 0:06:40.080 --> 0:06:42.839 grow it over a twenty percent giga over the next 0:06:43.000 --> 0:06:48.720 five to seven years. That's true off Tokyo as well, 0:06:48.839 --> 0:06:52.280 that is true of the Philippines, true of Indonesia. These 0:06:52.320 --> 0:06:55.600 are very very high growth fits. And of course they're 0:06:55.640 --> 0:06:58.200 starting off of small bases as well. So one of 0:06:58.240 --> 0:07:00.360 the reasons we see the growth is because they're also 0:07:00.600 --> 0:07:05.520 historical under investment or there just wasn't a need for 0:07:05.880 --> 0:07:08.960 this type of capacity to be set up in these regions. 0:07:09.560 --> 0:07:12.320 But for a whole bunch of reasons, you know there 0:07:12.400 --> 0:07:14.920 is that need today. Data sovereignty is one of them. 0:07:15.040 --> 0:07:18.680 But also from a market perspective, these markets are maturing 0:07:19.040 --> 0:07:22.080 and are attractive enough for the hyperscalers themselves to be 0:07:22.080 --> 0:07:25.400 investing in a very meaningful way. And so for us 0:07:25.440 --> 0:07:29.120 as data center investors, we're investing alongside or ahead of 0:07:29.240 --> 0:07:32.600 our end clients, typically Big Deck on the hyperscale side, 0:07:32.880 --> 0:07:35.080 to ensure that we have this capacity up and running 0:07:35.520 --> 0:07:36.480 as and when they need it. 0:07:37.520 --> 0:07:40.440 So you guys sort of pave the way you come in. 0:07:41.160 --> 0:07:44.080 You kind of lay the groundwork and then you get 0:07:44.280 --> 0:07:46.200 the Googles and Amazons of the world on board. 0:07:47.040 --> 0:07:50.720 It's it's a close partnership, I would say. So you 0:07:51.120 --> 0:07:53.240 do need to invest ahead of the client, but it's 0:07:53.240 --> 0:07:56.680 not quite speculative, so there's a strong sense of whether 0:07:56.800 --> 0:08:00.400 or not there is demand. So it's very collaborative in 0:08:00.480 --> 0:08:03.920 terms of working with the Googles and Amazons and Microsoft's 0:08:04.760 --> 0:08:07.680 to follow them or and often in many cases to 0:08:07.720 --> 0:08:11.280 actually have them direct the management teams to where their 0:08:11.320 --> 0:08:13.920 needs are. So these companies are very large owners and 0:08:13.960 --> 0:08:16.440 operators of data centers themselves, so these are captive data 0:08:16.440 --> 0:08:20.160 centers that they themselves manage and build. So often if 0:08:20.200 --> 0:08:23.600 we see them moving into a certain market and executing 0:08:23.640 --> 0:08:26.160 on a self build plan, then we know that there 0:08:26.200 --> 0:08:28.440 is demand there, and we know that there's probably room 0:08:28.520 --> 0:08:30.600 for a third party provider as well. 0:08:32.559 --> 0:08:36.400 Building these data centers, especially those related to AI, must 0:08:36.440 --> 0:08:39.560 be really expensive. A lot of people are talking about 0:08:39.600 --> 0:08:44.160 how expensive it is to buy these Nvidia chips used 0:08:44.200 --> 0:08:47.840 for artificial intelligence. Now, can you give us some information 0:08:48.000 --> 0:08:51.720 who actually buys the invidio chips? Is it yourselves or 0:08:51.760 --> 0:08:53.160 is it the hyperscalers? 0:08:54.600 --> 0:08:58.480 So that part the active equipment, the stuff that goes 0:08:58.559 --> 0:09:02.040 into the servers themselves is typically the client's responsibility, so 0:09:02.120 --> 0:09:05.560 that would be the hyperscalers themselves. The data center owners 0:09:05.559 --> 0:09:08.920 and operators typically are just responsible for the physical to 0:09:09.000 --> 0:09:11.160 more the gore shell, the civil infrastructure of the M 0:09:11.240 --> 0:09:13.880 and E, just making sure that the data holes are available, 0:09:14.320 --> 0:09:17.760 which we then lease out to the hyperscalers, you know, 0:09:17.760 --> 0:09:21.320 for them to put that equipment in. And to your point, 0:09:22.120 --> 0:09:25.400 the active equipment, so the servers and the chips going 0:09:25.440 --> 0:09:29.000 into the data centers are often a lot more expensive 0:09:29.040 --> 0:09:33.080 than the data centers themselves. So the rule of thumb 0:09:33.120 --> 0:09:35.520 is that it's called it sort of ten dollars of 0:09:35.640 --> 0:09:40.000 what for a data center itself. The equipment going in 0:09:40.040 --> 0:09:40.960 will be many times that. 0:09:42.400 --> 0:09:46.000 There's a lot of discussion of how much computing power 0:09:46.080 --> 0:09:49.840 AI requires, and this is obviously sounds like it's driving 0:09:49.840 --> 0:09:52.360 the demand for these data centers. Can you give us 0:09:52.400 --> 0:09:56.200 some numbers if AI really takes off, is there enough 0:09:56.240 --> 0:09:59.240 capacity right now? And how much needs to be built 0:09:59.559 --> 0:10:00.920 to satis fi all that demand. 0:10:03.000 --> 0:10:06.640 Look, there are many numbers that get thrown out, so 0:10:06.679 --> 0:10:10.400 I'll refrain from giving you numbers because they're quite country 0:10:10.440 --> 0:10:15.079 specific and client specific, just in terms of the power needs. 0:10:15.080 --> 0:10:20.040 But yes, it's not just the absolute megawatts. It's also 0:10:20.200 --> 0:10:24.319 the architecture within the data centers themselves. So the intensity 0:10:24.400 --> 0:10:27.600 of the power use is quite different for the latest 0:10:28.160 --> 0:10:31.520 generations of chips versus historical ones. And this is changing. 0:10:31.960 --> 0:10:34.199 You know, in two or three years, we're seeing enormous 0:10:34.200 --> 0:10:37.040 differences in terms of the types of power from a 0:10:37.120 --> 0:10:40.280 racked density perspective. You know, the needs of the latest 0:10:40.280 --> 0:10:44.640 generation of say the in video chips. So you know, 0:10:44.679 --> 0:10:47.720 some of the examples are just from a power demand perspective. 0:10:48.720 --> 0:10:52.880 I think globally, data centers, if historically accounted for maybe 0:10:53.200 --> 0:10:57.640 two to three percent of overall power demand, that is 0:10:57.679 --> 0:10:59.680 expected to go up there maybe about ten percent in 0:10:59.720 --> 0:11:03.679 some mats. In some markets, like say Singapore, DC is 0:11:03.720 --> 0:11:06.200 already account for ten percent of power demand, and so 0:11:06.200 --> 0:11:09.200 there's actually a moratory in some countries in terms of 0:11:09.720 --> 0:11:13.800 letting data centers continue to run unleashed, so power is 0:11:14.160 --> 0:11:18.960 being rationed out. But this is an enormous step change 0:11:19.000 --> 0:11:22.440 from a power demand perspective and what that means for 0:11:22.640 --> 0:11:26.360 associated infrastructure as well, both on the power generation side 0:11:26.360 --> 0:11:28.920 but also the transmission lines that you need the utilities 0:11:28.920 --> 0:11:33.960 that manage these electric networks is frankly an upgrading of 0:11:34.080 --> 0:11:38.240 a lot of old architecture because it is moving nations 0:11:38.240 --> 0:11:40.680 as power demands in a meaningful way. 0:11:42.080 --> 0:11:45.800 If data center demand is rising, if it's becoming a 0:11:45.840 --> 0:11:51.760 lot more energy intensive. Are all countries prepared for that 0:11:51.880 --> 0:11:55.200 right now? I know you mentioned India, the Philippines, US 0:11:55.200 --> 0:11:57.600 with DC there. I mean, are all of these places 0:11:58.240 --> 0:12:00.360 ready or do you see the potential for the need 0:12:00.440 --> 0:12:02.600 to do a lot of upgrades. 0:12:03.880 --> 0:12:07.000 I think you probably need to see upgrades across the 0:12:07.040 --> 0:12:09.920 board because it's also just the way some of these 0:12:10.000 --> 0:12:13.760 electrical networks, of the utilities, transmission and distribution networks were 0:12:13.760 --> 0:12:19.040 built didn't necessarily require that volume of power to certain 0:12:19.080 --> 0:12:21.800 regions and sights. So there's often bottlenecks in these systems, 0:12:21.840 --> 0:12:25.559 and it's ensuring that there is that substution capacity and redundancy, 0:12:25.600 --> 0:12:27.600 and you know there are no bottlenecks in terms of 0:12:27.640 --> 0:12:30.360 just how that transmission system was built out. So there's 0:12:30.400 --> 0:12:33.920 different degrees of upgrading that are required. That's on the 0:12:34.000 --> 0:12:37.200 utility side, but in a lot of different places. Countries 0:12:37.200 --> 0:12:40.559 that are power shot need more primary electrons in the 0:12:40.600 --> 0:12:44.360 grid anyway, So lots of markets were playing catch up. 0:12:44.360 --> 0:12:47.040 From a power perspective, there is now a whole new 0:12:47.080 --> 0:12:50.400 stack of demand that has come in, further increasing the 0:12:50.400 --> 0:12:54.040 need to have more electrons in the grid. So, yes, 0:12:54.120 --> 0:12:55.520 it's a lot to think through. 0:12:56.440 --> 0:12:59.640 And is this extra power demand coming from renewable energy 0:12:59.720 --> 0:13:03.600 or is it just traditional sources or both both. 0:13:03.640 --> 0:13:05.720 I think it depends on the market. It can be 0:13:05.880 --> 0:13:09.120 important powers. So Singapore is talking about multiple import projects 0:13:09.120 --> 0:13:12.360 around the region. Obviously most of Singapore's powered ninety plus 0:13:12.320 --> 0:13:16.040 percent today is gas. A lot of other countries will 0:13:16.080 --> 0:13:17.320 be in a similar position. 0:13:18.480 --> 0:13:21.160 Yeah, you had mentioned that. You know, there's this picture 0:13:21.240 --> 0:13:23.200 of demand growing. 0:13:23.600 --> 0:13:25.800 This is a trend that's not going away, but are 0:13:25.840 --> 0:13:28.760 there also risks to that outlook? You know, we were 0:13:28.840 --> 0:13:32.800 speaking with Deron Amoglu a few weeks ago in an episode. 0:13:32.840 --> 0:13:35.840 He's one of these AI critics. He sees this sort 0:13:35.880 --> 0:13:38.840 of as AI not having as big of an economic 0:13:38.920 --> 0:13:42.880 impact or perhaps as big impact on industry as we 0:13:42.960 --> 0:13:45.959 previously thought or as people assume. So if a lot 0:13:45.960 --> 0:13:48.960 of these data center growth is being led by AI, 0:13:49.320 --> 0:13:51.280 you know, is that one of the risks going forward 0:13:51.320 --> 0:13:55.319 that we could see maybe a pullback in that trend 0:13:55.600 --> 0:13:57.800 or do you see that as as staying. 0:13:58.960 --> 0:14:01.520 Another great question. Look, so there are definitely different points 0:14:01.559 --> 0:14:04.720 of view here right around whether investments in AI makes 0:14:04.800 --> 0:14:07.640 sense for the people who are really buying the chips 0:14:07.720 --> 0:14:11.200 right investing behind the architecture and all the costs, and 0:14:11.640 --> 0:14:13.520 people will say, oh, this is a bubble. There have 0:14:13.559 --> 0:14:15.720 been deck bubbles in the past as well, but this 0:14:15.800 --> 0:14:18.240 is very different. So you have some of the largest, 0:14:18.360 --> 0:14:22.000 most sophisticated and well capitalized technology companies in the world 0:14:22.160 --> 0:14:26.520 investing behind this wave, as opposed to say, you know, 0:14:26.560 --> 0:14:28.360 the late nineties, early two thousand and one, it was 0:14:28.400 --> 0:14:32.920 a lot more startup. Is there some degree of hype? Frankly, 0:14:33.080 --> 0:14:35.520 you know, it's not my place to comment on what 0:14:35.680 --> 0:14:38.080 the fair price for some of these companies should be. 0:14:38.160 --> 0:14:39.800 I think a lot of people can can form their 0:14:39.960 --> 0:14:42.920 own opinions on that front. But on the data center side, 0:14:42.960 --> 0:14:46.400 these investments are typically underpinned by long term contracts to 0:14:46.600 --> 0:14:52.440 these very high quality counterparties, and so there's enormous comfort 0:14:52.600 --> 0:14:56.360 in knowing that the counterparty will honor the contract because 0:14:56.400 --> 0:14:58.440 these tend to be pretty rock solid contracts as well, 0:14:59.000 --> 0:15:02.400 and provided you're in resting in the right regions. There 0:15:02.440 --> 0:15:04.760 could be other use cases. Will it be a large 0:15:04.840 --> 0:15:09.600 language model that is ultimately trained there, or inference occurs. 0:15:09.680 --> 0:15:11.880 Could there be a cloud zone. So long as you're 0:15:11.880 --> 0:15:13.840 in the right location, there are lots of different uses 0:15:13.880 --> 0:15:17.080 that you can put power and that capacity to. So 0:15:17.800 --> 0:15:20.400 maybe the numbers won't play out to the extent that 0:15:20.440 --> 0:15:23.000 they are projected today. But one thing about the data 0:15:23.000 --> 0:15:26.600 center space is that anyone who's historically forecast growth here 0:15:26.920 --> 0:15:31.480 has typically been wrong in terms of being conservative. Demand 0:15:31.560 --> 0:15:36.280 is actually outstripped historically what almost anyone has forecast as 0:15:36.320 --> 0:15:40.480 the market demand to be. So you wager a beer 0:15:40.560 --> 0:15:44.040 at your own perle. But I my estimate is that 0:15:44.280 --> 0:15:46.600 demand's actually going to be creative in what most people 0:15:46.640 --> 0:15:47.280 expect today. 0:15:48.960 --> 0:15:51.240 It sounds like a great business, like you make these 0:15:51.320 --> 0:15:56.080 long term contracts. Your counterparties are these massive hyperscals with 0:15:56.120 --> 0:15:58.000 a lot of cash. Everyone knows how strong they are 0:15:58.040 --> 0:16:00.520 in terms of the balance sheet. There must be a 0:16:00.520 --> 0:16:02.840 lot of competitors in this space. Is like tell us 0:16:02.840 --> 0:16:05.240 about the competitive landscape that you guys are doing. 0:16:06.760 --> 0:16:11.080 You're right, they are great businesses if you can access 0:16:11.120 --> 0:16:13.400 the customers, if you have the right team and the 0:16:13.480 --> 0:16:16.400 right sites, and it takes a long time to build 0:16:16.520 --> 0:16:21.040 that trust and that connectivity and to have frankly invested 0:16:21.040 --> 0:16:23.240 in the space for years to be able to now 0:16:23.320 --> 0:16:27.800 reap the rewards of those early investments and those locations 0:16:27.880 --> 0:16:31.320 and that power capacity and so on. So there are 0:16:31.360 --> 0:16:34.400 lots of different folks who are trying to enter this space, 0:16:34.920 --> 0:16:36.840 and it is i would say, getting tougher and tougher 0:16:36.880 --> 0:16:40.120 to do so because access to capital is a differentiata. 0:16:40.680 --> 0:16:44.440 These are enormous gazzlers of capital these businesses, and obviously 0:16:44.560 --> 0:16:48.640 without the fundamental assets, without the land and the power 0:16:48.720 --> 0:16:52.240 and the mutility connectivity, the fib or the water, ultimately 0:16:52.240 --> 0:16:55.880 you can't build. So the people who are best positioned 0:16:55.880 --> 0:16:57.960 to capitalize on the wave are the ones who've been 0:16:58.000 --> 0:17:01.120 investing in their business for many years now and in 0:17:01.160 --> 0:17:04.879 the relationships with these hyperskillers. And the markets are fragmented. 0:17:05.200 --> 0:17:10.040 There's no player that is such a larger market here 0:17:10.359 --> 0:17:12.720 that we would say, you know, is the only one 0:17:12.720 --> 0:17:15.160 that wins. This isn't a winner take all market. This 0:17:15.240 --> 0:17:18.359 is one where frankly, many firms should be able to 0:17:18.400 --> 0:17:22.200 do well provided they execute and deliver for their clients. 0:17:22.440 --> 0:17:25.560 Is it mostly large firms in the space or you 0:17:25.600 --> 0:17:26.600 just said it was fragmented. 0:17:26.640 --> 0:17:28.720 Is there a lot of smaller players? 0:17:29.720 --> 0:17:33.600 It depends on the market. So every nation will have 0:17:34.119 --> 0:17:35.879 some firms that are very large. The guys who are 0:17:35.960 --> 0:17:40.440 very large in Australia aren't necessarily very large in Singapore 0:17:40.600 --> 0:17:42.160 or India or Japan. 0:17:41.920 --> 0:17:43.840 For example, So it's really domestic. 0:17:44.520 --> 0:17:47.080 It's a combination. So there are some players who are 0:17:47.160 --> 0:17:49.960 multi country platforms, and there are others who are very 0:17:50.040 --> 0:17:52.560 large in one country and are now looking at building 0:17:52.560 --> 0:17:56.000 out across the board. Ultimately, our viue is that good 0:17:56.040 --> 0:17:59.760 platforms will move from country to country as their customers 0:18:00.080 --> 0:18:02.200 expanding those countries as well, So you sort of follow 0:18:02.240 --> 0:18:03.240 your customer in. 0:18:03.240 --> 0:18:07.760 This business and protecs, who are your main investors? Are 0:18:07.760 --> 0:18:11.000 they sort of pension funds, sovereign wealth funds? Like are 0:18:11.040 --> 0:18:13.520 they rich individuals? Tell us like, who are the people 0:18:13.560 --> 0:18:16.159 that's backing you to invest in these data centers. 0:18:17.400 --> 0:18:20.560 So our LPs are a combination of exactly those types 0:18:20.600 --> 0:18:23.200 of folks, typically a blue chip roster of the most 0:18:23.320 --> 0:18:27.000 sophisticated capital alligators in the world, and they can be 0:18:27.040 --> 0:18:31.720 pension funds, sovereign wealth funds, large insurance companies, institutional investors. 0:18:32.119 --> 0:18:36.000 It's a broad mix of both types of investors and 0:18:36.400 --> 0:18:38.639 cross geographies as well in terms of where these investors 0:18:38.640 --> 0:18:40.600 are domiciled and pratrescia. 0:18:40.680 --> 0:18:44.600 You mentioned quite a few regions, in particular in Southeast Asia, 0:18:44.640 --> 0:18:46.840 and it seems like Southeast Asia is getting a lot 0:18:46.840 --> 0:18:50.640 of the growth, but you did not mention China. Now 0:18:50.680 --> 0:18:54.000 there's obviously a lot of geopolitical tensions between the US 0:18:54.040 --> 0:18:59.160 and China. Is this in any way impacting where hyperscalas 0:18:59.400 --> 0:19:01.200 build their data centers? 0:19:02.600 --> 0:19:06.040 I think what I can probably say is that data 0:19:06.680 --> 0:19:14.800 is regulated, whether that's requirements to store data domestically, you 0:19:14.800 --> 0:19:18.159 know how that data is used. It gets quite technical 0:19:18.359 --> 0:19:22.680 and illegal quite quickly. Companies, I think are taking a 0:19:22.800 --> 0:19:25.479 view on where their growth is going to be, and 0:19:25.560 --> 0:19:31.080 Western hyperscalas are probably expanding across regions where they see 0:19:31.680 --> 0:19:34.480 more prolonged growth and less regulatory challenges. 0:19:35.320 --> 0:19:37.200 Yeah, so lots of things to consider when you're trying 0:19:37.240 --> 0:19:41.640 to open a data center. Yeah, what is something that 0:19:41.760 --> 0:19:45.040 people might be missing when it comes to data centers. 0:19:45.880 --> 0:19:48.920 I think there's a tendency to assume that all data 0:19:48.920 --> 0:19:51.359 centers are the same, or just to paint the entire 0:19:51.400 --> 0:19:54.720 sector with one brush, and that is actually not the case. 0:19:55.000 --> 0:19:57.919 So with all the changes that are happening now on 0:19:58.119 --> 0:20:03.119 the chip side and then there architecture side, the older 0:20:03.200 --> 0:20:07.000 data centers are in a lot of cases increasingly more challenged, 0:20:07.600 --> 0:20:12.800 so deployment sizes and that's just effectively megawatts are getting larger. 0:20:13.240 --> 0:20:15.320 So it's possible to have you know, a ten year 0:20:15.359 --> 0:20:17.879 old or fifteen year old site that is not quite 0:20:17.880 --> 0:20:21.840 after scratch in terms of where technology or design and 0:20:21.960 --> 0:20:25.200 architecture have moved on from a power usage effectiveness and 0:20:25.200 --> 0:20:28.320 from a scale perspective. So while there's a lot of 0:20:28.400 --> 0:20:32.000 excitement around the sector as a whole, there is a 0:20:32.160 --> 0:20:36.920 buyer beware sort of you know disclaimer as well around 0:20:37.040 --> 0:20:40.720 what type of data center you're buying and how relevant 0:20:40.760 --> 0:20:45.080 that is in terms of lease up and location for 0:20:45.160 --> 0:20:48.200 future growth as well. There's so much need for new 0:20:48.359 --> 0:20:52.720 assets to be created, there's real risk of some stuff 0:20:52.760 --> 0:20:54.159 getting outdated in the space. 0:20:54.520 --> 0:20:55.880 What are some examples of that? 0:20:56.640 --> 0:20:59.040 I know you can't talk about specific companies probably, but 0:20:59.400 --> 0:21:02.120 you know general, what's an example of that situation? Where 0:21:02.440 --> 0:21:04.600 is it just a data center that becomes very old 0:21:05.080 --> 0:21:06.520 and maybe even so old that. 0:21:06.480 --> 0:21:07.480 It can be updated. 0:21:07.600 --> 0:21:10.120 Or is it more that you know, technology is advancing 0:21:10.160 --> 0:21:12.520 so quickly that even if you start building one now, 0:21:12.520 --> 0:21:14.280 it might be outdated by the time it's done. 0:21:14.760 --> 0:21:17.280 Now, it's more the former, I would say, because remember here, 0:21:17.320 --> 0:21:19.960 the long term contracts protect you for the duration of 0:21:19.960 --> 0:21:22.479 that contract, so at least for a while. For so 0:21:22.520 --> 0:21:24.920 long as you have that contract, you're in a pretty 0:21:24.960 --> 0:21:28.520 good place. One example I can give you is actually 0:21:28.560 --> 0:21:31.800 in our portfolio. One of the underlying PCs for our 0:21:31.840 --> 0:21:35.280 investment in Cintel's data centers, a company called Etceera for 0:21:35.320 --> 0:21:39.440 finance last year, is actually the reallocation of power to 0:21:39.600 --> 0:21:42.840 more efficient and new data centers as opposed to some 0:21:42.840 --> 0:21:45.840 of the older stuff which is the first generation. So 0:21:46.000 --> 0:21:49.119 third and fourth generation data centers today, especially in a 0:21:49.160 --> 0:21:51.960 market like Singapore, will be a lot more efficient and 0:21:52.000 --> 0:21:55.280 you can get a lot more by aggregating the power 0:21:55.320 --> 0:21:57.840 supply that goes into some of these older sites and 0:21:57.880 --> 0:22:01.280 reallocating that to a new facility that is just more 0:22:01.280 --> 0:22:04.160 purpose built. It's a great story, you know, to give 0:22:04.240 --> 0:22:06.800 the centrality team credit. They thought this through and real 0:22:06.880 --> 0:22:09.320 at least to us pioneers in the space around trying 0:22:09.359 --> 0:22:14.080 to rebuild or repower these facilities in Singapore and effectively 0:22:14.200 --> 0:22:18.800 squeeze more it megawatts out of the same power allocation, 0:22:18.960 --> 0:22:23.320 so really avoiding emissions. There's a real environmental story here 0:22:23.359 --> 0:22:26.760 around avoided emissions because the data center itself takes the 0:22:26.760 --> 0:22:29.119 same power down from the grid, but is able to 0:22:29.200 --> 0:22:32.679 sell on more of that power to its customers because 0:22:32.680 --> 0:22:34.520 it's just more efficient. So the building needs a lot 0:22:34.560 --> 0:22:37.840 less power to cool itself, among other things. So we're 0:22:37.840 --> 0:22:41.560 seeing that as a very interesting angle in certain markets 0:22:41.600 --> 0:22:44.160 that are more power constrained, and we're frankly investing behind 0:22:44.160 --> 0:22:47.760 that team as well, just just building, reallocating from the 0:22:47.800 --> 0:22:50.600 old and building new, more purpose built facilities. 0:22:51.440 --> 0:22:53.600 Sort of what's happening with the office space now, I 0:22:53.600 --> 0:22:55.919 guess as well? A lot of empty space and then 0:22:55.960 --> 0:22:58.960