1 00:00:02,960 --> 00:00:06,119 Speaker 1: Rakoto. Welcome to the Shared Lunch brought to you by Sharesy's. 2 00:00:06,200 --> 00:00:09,920 Speaker 1: I'm Garth Bray. Look, we've recently had some staggering results 3 00:00:09,920 --> 00:00:13,520 Speaker 1: from the world's biggest company in video to which people 4 00:00:13,680 --> 00:00:18,000 Speaker 1: investors said, Eh, what is going on in tech? We 5 00:00:18,120 --> 00:00:21,040 Speaker 1: are getting right down into the weeds with tech sector 6 00:00:21,079 --> 00:00:25,639 Speaker 1: specialist Andrew Kurtain from Milford Asset Management. Before we do that, though, 7 00:00:25,680 --> 00:00:28,840 Speaker 1: some important information you should always consider when investing. 8 00:00:29,000 --> 00:00:31,560 Speaker 2: Investing involves a risk you might lose the money you 9 00:00:31,600 --> 00:00:34,879 Speaker 2: start with. We recommend talking to a licensed financial advisor. 10 00:00:35,600 --> 00:00:39,440 Speaker 2: We also recommend breading product disclosure documents before deciding to invest. 11 00:00:39,680 --> 00:00:42,080 Speaker 2: Everything you're about to see and here is current at 12 00:00:42,080 --> 00:00:42,960 Speaker 2: the time of recording. Ok. 13 00:00:43,080 --> 00:00:48,760 Speaker 1: Andrew Curtain from Milford Asset Management. A tech sector specialist does. 14 00:00:48,600 --> 00:00:52,400 Speaker 3: What all day Well, in my job, it's it's tech 15 00:00:52,440 --> 00:00:54,880 Speaker 3: sector specialist looking at equities, so we look at the 16 00:00:54,880 --> 00:00:58,520 Speaker 3: stock market. I focus on the tech sector. So the 17 00:00:58,520 --> 00:01:02,440 Speaker 3: tech sector would include anything from software companies like Microsoft 18 00:01:02,520 --> 00:01:07,119 Speaker 3: or Salesforce, semiconductor companies like Navidio, broad Conn, right through 19 00:01:07,120 --> 00:01:10,560 Speaker 3: to say communication services which includes companies like Meta or Google, 20 00:01:10,760 --> 00:01:14,200 Speaker 3: Amazon clear to So my job and the tech is 21 00:01:14,240 --> 00:01:16,480 Speaker 3: to sort of focus on researching these companies, get to 22 00:01:16,560 --> 00:01:19,320 Speaker 3: understand them, understand their industries, and then feed that into 23 00:01:19,319 --> 00:01:22,000 Speaker 3: stock picks which go into our funds at milf AC Management. 24 00:01:22,520 --> 00:01:25,480 Speaker 1: I guess we've got two results that really frame things 25 00:01:25,560 --> 00:01:29,319 Speaker 1: quite remarkably. Very recently. We've gotten VideA reporting I think 26 00:01:29,360 --> 00:01:33,800 Speaker 1: it was its quarterlies and salesforce reporting quarterlies. And if 27 00:01:33,800 --> 00:01:36,720 Speaker 1: you want to pick an analogy, whether it's railroads or 28 00:01:36,760 --> 00:01:39,320 Speaker 1: whether it's gold mining or whatever, you've got a company 29 00:01:39,319 --> 00:01:44,200 Speaker 1: there that is building the infrastructure, building the way that 30 00:01:44,240 --> 00:01:46,000 Speaker 1: this will happen in the video, and you've had a 31 00:01:46,000 --> 00:01:48,640 Speaker 1: company in salesforce that's building the thing that's going to 32 00:01:48,640 --> 00:01:52,160 Speaker 1: clip the ticket. Right, So let's start maybe with those. 33 00:01:52,320 --> 00:01:54,160 Speaker 1: What's been going on with these companies lately? 34 00:01:54,840 --> 00:01:57,560 Speaker 3: Yeah, I mean the two technology companies. Both of them 35 00:01:57,560 --> 00:02:00,720 Speaker 3: are in sectors which have very biggin markets own rights, 36 00:02:00,760 --> 00:02:04,000 Speaker 3: but the way they participate in this in this technology 37 00:02:04,000 --> 00:02:07,160 Speaker 3: stack is very very different. So so as you sort 38 00:02:07,200 --> 00:02:10,360 Speaker 3: of gathered example of a railway, Nividia is sort of 39 00:02:10,400 --> 00:02:12,440 Speaker 3: like the railway company. It is the company that is 40 00:02:12,520 --> 00:02:17,239 Speaker 3: laying the physical infrastructure to allow artificial intelligence to grow 41 00:02:17,320 --> 00:02:19,920 Speaker 3: and become a part of our daily lives. So you know, 42 00:02:19,919 --> 00:02:22,560 Speaker 3: in the sense like like you might build a railway, 43 00:02:22,639 --> 00:02:25,000 Speaker 3: like laying down the railway tracks and that that's sort 44 00:02:25,040 --> 00:02:27,680 Speaker 3: of what Nividia is doing. It's building the physical chips 45 00:02:28,280 --> 00:02:31,799 Speaker 3: that power AI data centers. So that's the brains, that's 46 00:02:31,880 --> 00:02:34,600 Speaker 3: that's what makes these data centers tech. So and that's a. 47 00:02:34,560 --> 00:02:36,399 Speaker 1: Point of scarcity, which we'll come back to you later. 48 00:02:36,400 --> 00:02:39,680 Speaker 1: Because anywhere you can find scarcity and there's a company 49 00:02:39,680 --> 00:02:41,600 Speaker 1: capturing that, they're going to be capturing a lot of value, 50 00:02:41,600 --> 00:02:43,400 Speaker 1: aren't they. It turns out there's a lot of scarcity 51 00:02:43,440 --> 00:02:45,920 Speaker 1: and they at the moment because it's grown so fast, 52 00:02:46,000 --> 00:02:48,760 Speaker 1: so it makes the supply of everything going into a 53 00:02:48,800 --> 00:02:49,600 Speaker 1: little bit challenge. 54 00:02:49,639 --> 00:02:52,000 Speaker 3: So that's yes. So that's what Nividia does. It designs 55 00:02:52,040 --> 00:02:53,920 Speaker 3: these chips and they're very important chips. And that's why 56 00:02:53,960 --> 00:02:55,800 Speaker 3: Nvidia is they are the largest company in the world. 57 00:02:56,800 --> 00:03:00,240 Speaker 3: Salesforces has comes from it as a software company. Them 58 00:03:00,280 --> 00:03:03,600 Speaker 3: if you look back over the last fifteen years, one 59 00:03:03,639 --> 00:03:06,120 Speaker 3: of the best sectors to be invested in has been 60 00:03:06,200 --> 00:03:07,160 Speaker 3: software and. 61 00:03:07,120 --> 00:03:10,320 Speaker 1: They specifically do customer relationship management. 62 00:03:11,040 --> 00:03:15,800 Speaker 3: Customer relationship management software and so they basically provide software 63 00:03:15,800 --> 00:03:18,480 Speaker 3: which sits on your computer, and it helps make people 64 00:03:18,480 --> 00:03:21,639 Speaker 3: that work in sales people, marketing people, and customer services 65 00:03:21,680 --> 00:03:24,799 Speaker 3: makes their lives easier. It will track the relationship, will 66 00:03:24,840 --> 00:03:26,799 Speaker 3: help you find leads for customers you want to call 67 00:03:26,960 --> 00:03:29,040 Speaker 3: you cool them. It will help record all that information. 68 00:03:29,120 --> 00:03:32,320 Speaker 3: It might then pass it information onto customer servicing people 69 00:03:32,360 --> 00:03:34,040 Speaker 3: who then have to deal with their complaints when they 70 00:03:34,080 --> 00:03:37,200 Speaker 3: come in. And so it's been sitting in sort of 71 00:03:37,240 --> 00:03:40,240 Speaker 3: the tops at are five to ten software companies in 72 00:03:40,280 --> 00:03:42,120 Speaker 3: the world by market cap, So it's been a very 73 00:03:42,120 --> 00:03:44,120 Speaker 3: good business. And it's also a business which is called 74 00:03:44,960 --> 00:03:47,520 Speaker 3: which is a cloud based business, which what that basically 75 00:03:47,560 --> 00:03:50,400 Speaker 3: means now is you don't actually have to physically buy 76 00:03:50,400 --> 00:03:53,920 Speaker 3: any hardware to run their program. It's all stored on 77 00:03:53,920 --> 00:03:56,320 Speaker 3: the cloud, and you get a subscription service and you 78 00:03:56,360 --> 00:03:58,640 Speaker 3: sort of pay like this sort of yearly subscription to 79 00:03:58,640 --> 00:04:01,040 Speaker 3: get access to their software. They'll upgrade it for you 80 00:04:01,120 --> 00:04:03,000 Speaker 3: and we'll get better. Then. At the moment, there's big 81 00:04:03,000 --> 00:04:05,680 Speaker 3: debate about which which is the better business to invest 82 00:04:05,680 --> 00:04:07,720 Speaker 3: it is the software companies, is at the hardware or 83 00:04:07,720 --> 00:04:08,520 Speaker 3: the chip companies. 84 00:04:08,600 --> 00:04:10,360 Speaker 1: If we look most recently, I think I think so 85 00:04:10,400 --> 00:04:15,000 Speaker 1: in Vidia. It really surprised people, whether it's most recent 86 00:04:15,160 --> 00:04:17,560 Speaker 1: quarterly update. I think what we're looking at revenue of 87 00:04:17,600 --> 00:04:22,080 Speaker 1: like sixty eight billion US and it's sort of targeting 88 00:04:22,120 --> 00:04:24,080 Speaker 1: a whole lot more than that seventy eight billion in 89 00:04:24,120 --> 00:04:27,760 Speaker 1: the next quarter. These are staggering numbers. How come the 90 00:04:27,800 --> 00:04:31,400 Speaker 1: stock price is just kind of like dropping or rising 91 00:04:31,520 --> 00:04:33,760 Speaker 1: or doing whatever it's doing. What's going on there? 92 00:04:34,080 --> 00:04:37,080 Speaker 3: Yeah, the share price actually fell five percent after, which 93 00:04:37,120 --> 00:04:38,760 Speaker 3: is one of the one of the biggest beats the 94 00:04:38,800 --> 00:04:42,599 Speaker 3: companies had in sort of recent result quarters. And I 95 00:04:42,640 --> 00:04:44,560 Speaker 3: think you just sort of need to step back here 96 00:04:44,600 --> 00:04:48,599 Speaker 3: and realize stop market lists of stop market investing, particularly 97 00:04:48,600 --> 00:04:50,720 Speaker 3: in the technology space and particularly in these sort of 98 00:04:50,760 --> 00:04:54,599 Speaker 3: environments with uncertainty, the short term share price action doesn't 99 00:04:54,640 --> 00:04:57,720 Speaker 3: always sort of reflect exactly what the fundamentals are from 100 00:04:57,760 --> 00:05:00,440 Speaker 3: the piece of news. It often does, but times that 101 00:05:00,480 --> 00:05:03,279 Speaker 3: can stray away from that. So I think every single 102 00:05:03,320 --> 00:05:06,719 Speaker 3: person look at the Navidia Navidia result yesterday, I would agree 103 00:05:06,720 --> 00:05:08,760 Speaker 3: that was a really strong result. As you sort of said, 104 00:05:08,800 --> 00:05:12,120 Speaker 3: it guided to a ten billion quarter and quarter growth 105 00:05:12,120 --> 00:05:14,279 Speaker 3: in revenue from sixty eight to seventy eight billion. The 106 00:05:14,320 --> 00:05:16,719 Speaker 3: market thought the revenues should have been guided to about 107 00:05:16,720 --> 00:05:19,560 Speaker 3: seventy four So four billion dollar beats five or six percent. 108 00:05:19,640 --> 00:05:22,320 Speaker 3: It's a huge beat. So what that is telling you 109 00:05:22,400 --> 00:05:24,799 Speaker 3: is that demand for their products are really really strong. 110 00:05:25,160 --> 00:05:27,480 Speaker 3: They're able to sell their products, are good, strong prices. 111 00:05:27,560 --> 00:05:30,320 Speaker 3: They even confirm that the profit margins there making on 112 00:05:30,360 --> 00:05:32,280 Speaker 3: these products were as good as everyone was sort of 113 00:05:32,320 --> 00:05:33,880 Speaker 3: hoping to see the numbers come out. So there's no 114 00:05:33,880 --> 00:05:36,400 Speaker 3: real negative and the result whatsoever. So while the share 115 00:05:36,400 --> 00:05:40,000 Speaker 3: price fall of five percent, well, the positioning what we 116 00:05:40,000 --> 00:05:42,200 Speaker 3: could position in it is the mountain investors that own 117 00:05:42,400 --> 00:05:45,640 Speaker 3: certain companies or certain stocks is really long semiconductors. At 118 00:05:45,640 --> 00:05:48,800 Speaker 3: the moment. Semiconductors has been an area which has sort 119 00:05:48,800 --> 00:05:51,279 Speaker 3: of been you know, you'll probably say the headline of 120 00:05:51,279 --> 00:05:54,280 Speaker 3: the AI rally so far on the stock market over 121 00:05:54,320 --> 00:05:56,400 Speaker 3: the last couple of years. So a lot of positions own, 122 00:05:56,600 --> 00:05:58,720 Speaker 3: a lot of investors own navidio. They're really long in it, 123 00:05:59,040 --> 00:06:01,640 Speaker 3: and it gets harder and for the incremental positive news 124 00:06:01,680 --> 00:06:04,200 Speaker 3: to keep pushing the share price up. So sometimes you 125 00:06:04,240 --> 00:06:06,720 Speaker 3: get people sitting there and the vestors and particularly hedge 126 00:06:06,720 --> 00:06:09,160 Speaker 3: funds and really short term traders are sitting there going 127 00:06:09,480 --> 00:06:12,200 Speaker 3: right on the Nvidia result we're going to sell our winners, 128 00:06:12,240 --> 00:06:14,040 Speaker 3: which is the semiconduct to companies, and when we go 129 00:06:14,080 --> 00:06:15,920 Speaker 3: back and buy some of these losers, like the software 130 00:06:15,960 --> 00:06:18,040 Speaker 3: companies which have been selling off, and it's almost like 131 00:06:18,080 --> 00:06:19,599 Speaker 3: it doesn't matter what the result is, it's going to 132 00:06:19,600 --> 00:06:21,760 Speaker 3: happen that way. And so, yeah, we see it as 133 00:06:21,800 --> 00:06:23,800 Speaker 3: a bit of a short term maybe pull back in 134 00:06:23,880 --> 00:06:25,760 Speaker 3: Navideo's but we're not sort of like translating this and 135 00:06:25,800 --> 00:06:27,600 Speaker 3: so this is actually a negative result. 136 00:06:27,600 --> 00:06:29,920 Speaker 1: You're seeing and you're talking to your clients and saying, 137 00:06:30,000 --> 00:06:31,960 Speaker 1: we see a future for this company in line with 138 00:06:32,040 --> 00:06:33,000 Speaker 1: its results kind of thing. 139 00:06:33,160 --> 00:06:36,240 Speaker 3: Yeah, I mean, we're an active manager. So we move 140 00:06:36,279 --> 00:06:40,240 Speaker 3: our positions and move our views quite regularly, and we 141 00:06:40,320 --> 00:06:42,440 Speaker 3: move that based on the information that's in the market. 142 00:06:42,440 --> 00:06:44,000 Speaker 3: It could be on the results of the company, it 143 00:06:44,040 --> 00:06:46,840 Speaker 3: can be leading information. And in the case in Navidia 144 00:06:46,920 --> 00:06:51,640 Speaker 3: right now, what really drives the value of Navidia is 145 00:06:51,880 --> 00:06:54,760 Speaker 3: for how many years are they able to grow their 146 00:06:54,800 --> 00:06:57,840 Speaker 3: revenues from the sale of these chips? Now, what makes 147 00:06:57,880 --> 00:06:59,880 Speaker 3: you grow your revenue from seller chips? Where you need 148 00:06:59,920 --> 00:07:02,360 Speaker 3: the demand for the chips. Demand from the chips as 149 00:07:02,400 --> 00:07:05,040 Speaker 3: people using AI, so you need more people using AI. 150 00:07:05,600 --> 00:07:08,240 Speaker 3: You need more people willing to pay up to use 151 00:07:08,279 --> 00:07:11,000 Speaker 3: the AI, and then you need to make sure that 152 00:07:11,000 --> 00:07:12,840 Speaker 3: the video is producing the best chips and is going 153 00:07:12,880 --> 00:07:16,800 Speaker 3: to capture that market share. At the moment, there's undeniably 154 00:07:16,840 --> 00:07:19,720 Speaker 3: strong demand for AI. We'll probably touch on this further 155 00:07:20,360 --> 00:07:23,480 Speaker 3: in this interview, but Claude has really some fantastic products 156 00:07:23,480 --> 00:07:25,640 Speaker 3: and people use it more and more every day. So 157 00:07:25,680 --> 00:07:28,440 Speaker 3: the demand for AI is sort of actually going exponential. 158 00:07:28,520 --> 00:07:31,160 Speaker 3: It's taking off. NA Video is positioned to capture a 159 00:07:31,240 --> 00:07:33,440 Speaker 3: huge part of that share, and so we think it's 160 00:07:33,520 --> 00:07:35,480 Speaker 3: quite well positioned based on what we can see sort 161 00:07:35,520 --> 00:07:37,800 Speaker 3: of twelve to eighteen months out. So you were to 162 00:07:37,840 --> 00:07:39,840 Speaker 3: pick a and you know, sort of nippick a little 163 00:07:39,840 --> 00:07:41,480 Speaker 3: bit at what's the negative of nivideo at the moment. 164 00:07:41,480 --> 00:07:43,560 Speaker 3: It's the fact that this competition is getting a bit stronger. 165 00:07:43,640 --> 00:07:47,080 Speaker 3: There's a company called Broadcom which makes some very similar 166 00:07:47,440 --> 00:07:50,680 Speaker 3: chips that called TPUs. They make them primarily for Google. 167 00:07:51,160 --> 00:07:53,200 Speaker 3: They're producing some really good chips and they're starting to 168 00:07:53,240 --> 00:07:55,920 Speaker 3: encroach on Nividya's market share a little bit and take 169 00:07:55,960 --> 00:07:56,880 Speaker 3: some of those revenues. 170 00:07:57,080 --> 00:08:00,800 Speaker 1: Looking at salesforce then, which is this company that's making 171 00:08:00,840 --> 00:08:04,160 Speaker 1: a buck out of software and also bringing more and 172 00:08:04,200 --> 00:08:07,640 Speaker 1: more AI applications into the situation. And I guess you know, 173 00:08:07,800 --> 00:08:10,320 Speaker 1: in your railway analogy, what they're trying to they're trying 174 00:08:10,320 --> 00:08:12,320 Speaker 1: to build trains, or they're trying to sell tickets to 175 00:08:12,360 --> 00:08:14,240 Speaker 1: the trains, or they're trying to you know, they're trying 176 00:08:14,280 --> 00:08:15,760 Speaker 1: to get that kind of revenue through. 177 00:08:16,160 --> 00:08:18,640 Speaker 3: Well, yeah, she depends on your perspective here. It could 178 00:08:18,720 --> 00:08:22,880 Speaker 3: actually be that salesforces is the horse and cart that's 179 00:08:22,920 --> 00:08:27,680 Speaker 3: actually getting replaced by the train here. So yes, if 180 00:08:27,680 --> 00:08:31,200 Speaker 3: you sort of look at software, what AI is going 181 00:08:31,200 --> 00:08:33,680 Speaker 3: to do is it's going to make it easier and 182 00:08:33,800 --> 00:08:36,520 Speaker 3: aldy is it's making it easier and cheaper to build software. 183 00:08:37,200 --> 00:08:41,960 Speaker 3: What used to take specialized engineers called software engineers, and 184 00:08:42,000 --> 00:08:44,840 Speaker 3: they and they build to write code and computers and 185 00:08:44,880 --> 00:08:48,240 Speaker 3: basically have their own language that me and you wouldn't 186 00:08:48,240 --> 00:08:51,160 Speaker 3: be able to do. We can do that now. We 187 00:08:51,200 --> 00:08:53,400 Speaker 3: can type in normal English into a place in that 188 00:08:53,559 --> 00:08:56,280 Speaker 3: like called code or Chatchipitis codex, and we can ask 189 00:08:56,320 --> 00:08:59,160 Speaker 3: it to create a piece of software that will do something, 190 00:08:59,160 --> 00:09:01,240 Speaker 3: whether it's you know, simple things that we're playing with. 191 00:09:01,280 --> 00:09:04,080 Speaker 3: We're creating dashboards at work, we're trying to you know, 192 00:09:04,080 --> 00:09:05,760 Speaker 3: we have a lot of information we have to process, 193 00:09:05,840 --> 00:09:08,200 Speaker 3: and we and we actually were creating little software applications 194 00:09:08,200 --> 00:09:10,559 Speaker 3: that actually do what we want it to do human 195 00:09:10,720 --> 00:09:13,959 Speaker 3: just writing natural English language. And so in one sense 196 00:09:14,040 --> 00:09:15,600 Speaker 3: you kind of go, look, there's going to be no 197 00:09:16,000 --> 00:09:18,560 Speaker 3: definitely no shortage of using software products. We're going to 198 00:09:18,640 --> 00:09:21,559 Speaker 3: keep using them. So Salesforce, we're going to keep using Salesforce. 199 00:09:22,160 --> 00:09:25,560 Speaker 3: But other people might be able to create products which 200 00:09:25,559 --> 00:09:29,000 Speaker 3: compete with Salesforce a lot cheaper and easier. And so 201 00:09:29,080 --> 00:09:32,160 Speaker 3: what does that mean Salesforce business disappears. I don't think so. 202 00:09:33,520 --> 00:09:36,160 Speaker 3: I think it just means it's competition to keep its place. 203 00:09:36,240 --> 00:09:38,920 Speaker 3: So at the moment, Salesforce has really large market share 204 00:09:38,960 --> 00:09:42,640 Speaker 3: in this in this customer relationship management seement dealing with 205 00:09:42,679 --> 00:09:45,679 Speaker 3: the salesforce of companies. Well, maybe companies might decide or 206 00:09:45,720 --> 00:09:47,559 Speaker 3: we might try build some of the cell sales internally. 207 00:09:47,720 --> 00:09:50,320 Speaker 3: We might try to use this other this new AI. 208 00:09:50,360 --> 00:09:52,160 Speaker 3: Companies come up with a core product, we might we 209 00:09:52,240 --> 00:09:55,320 Speaker 3: might use their product, or well we might use you Salesforce. 210 00:09:55,360 --> 00:09:56,960 Speaker 3: But actually, you know what we might put the price, 211 00:09:57,000 --> 00:09:59,160 Speaker 3: we might not take that five percent price increase you're 212 00:09:59,160 --> 00:10:00,960 Speaker 3: trying to push through because if you do, we might 213 00:10:01,000 --> 00:10:03,719 Speaker 3: go try this. So it's an interesting dynamic. For what 214 00:10:03,800 --> 00:10:06,679 Speaker 3: it's worth, we think, you know, the business isn't going 215 00:10:06,679 --> 00:10:08,960 Speaker 3: to disappear anytime soon. I think we're just a little 216 00:10:08,960 --> 00:10:11,960 Speaker 3: bit more concerned generally with software as do they continue 217 00:10:11,960 --> 00:10:14,079 Speaker 3: these historical high growth rates. They used to grow at 218 00:10:14,080 --> 00:10:17,120 Speaker 3: fifteen twenty percent of sales growth per annum? Is that 219 00:10:17,200 --> 00:10:19,439 Speaker 3: now five percent? Is that ten percent? And that's what 220 00:10:19,480 --> 00:10:21,000 Speaker 3: we and the rest of the market are trying to 221 00:10:21,040 --> 00:10:21,520 Speaker 3: figure out. 222 00:10:22,600 --> 00:10:25,760 Speaker 1: I guess what if we look at the inn at 223 00:10:25,760 --> 00:10:27,480 Speaker 1: the results that they had where you kind of had 224 00:10:27,520 --> 00:10:31,400 Speaker 1: and Video Amazing doesn't really get rewarded, actually drops, you know, 225 00:10:31,520 --> 00:10:36,000 Speaker 1: Salesforce pretty good, pretty close, actually took a dip. I 226 00:10:36,040 --> 00:10:37,959 Speaker 1: think it's maybe then it climbed back a little bit. 227 00:10:38,280 --> 00:10:40,640 Speaker 1: What does that tell you about what investors are seeing 228 00:10:41,120 --> 00:10:42,760 Speaker 1: where we're out in this sort of cycle. 229 00:10:43,520 --> 00:10:46,480 Speaker 3: Yeah, I think that the one day reaction is not 230 00:10:46,880 --> 00:10:49,880 Speaker 3: really reflective of the results of these companies themselves. I mean, 231 00:10:49,920 --> 00:10:51,840 Speaker 3: if you actually step back and the Video had fantastic 232 00:10:51,880 --> 00:10:54,200 Speaker 3: results and the shares went down and Salesforce actually had 233 00:10:54,280 --> 00:10:57,040 Speaker 3: quite weak results, they sort of hit their numbers on 234 00:10:57,080 --> 00:10:59,400 Speaker 3: their quarter. Okay, but they guided to about a percent 235 00:10:59,440 --> 00:11:02,240 Speaker 3: lower revene growth for the next for the calendar year 236 00:11:02,280 --> 00:11:05,040 Speaker 3: twenty six in the market expecting, the shares went up. 237 00:11:05,320 --> 00:11:07,240 Speaker 3: So what it tells me actually is that as the 238 00:11:07,280 --> 00:11:11,559 Speaker 3: market has probably oversold software names over the last six weeks. 239 00:11:11,600 --> 00:11:13,920 Speaker 3: I mean, most software names are down around about thirty 240 00:11:13,920 --> 00:11:16,280 Speaker 3: to forty percent year to date. And we're not talking 241 00:11:16,280 --> 00:11:20,760 Speaker 3: about small, small cat software companies. We're talking about the 242 00:11:20,760 --> 00:11:22,880 Speaker 3: bigger software companies in the world are down thirty to 243 00:11:22,920 --> 00:11:25,640 Speaker 3: forty percent. When the market's flat, you know, someday. If 244 00:11:25,640 --> 00:11:27,800 Speaker 3: you ever usually see stocks down thirty forty it's because 245 00:11:27,800 --> 00:11:30,320 Speaker 3: the market's fall and twenty on COVID or a big 246 00:11:30,360 --> 00:11:32,320 Speaker 3: seal event. I mean, seeing stocks down this much and 247 00:11:32,360 --> 00:11:37,160 Speaker 3: the flat market is actually very very unusual, And so 248 00:11:37,480 --> 00:11:40,559 Speaker 3: I think the market's oversold software companies. It's probably overbought 249 00:11:40,760 --> 00:11:43,079 Speaker 3: semiconductor sub companies, and you've just seen a bit of 250 00:11:43,120 --> 00:11:46,280 Speaker 3: a position reversal short term. I think the other thing 251 00:11:46,280 --> 00:11:48,560 Speaker 3: it probably does sort of lead to you is, I mean, 252 00:11:48,679 --> 00:11:51,560 Speaker 3: the headlines have been so negative for software companies this 253 00:11:51,760 --> 00:11:54,200 Speaker 3: entire year. It's been one thing after another. It's been 254 00:11:54,720 --> 00:11:57,120 Speaker 3: clawed codes come out and we can build own software. 255 00:11:57,160 --> 00:11:59,600 Speaker 3: Claud Cowork has come out and all of a sudden 256 00:11:59,600 --> 00:12:01,920 Speaker 3: we can manage our whole desktops. And we're not going 257 00:12:01,960 --> 00:12:03,000 Speaker 3: to be able to get a need to go and 258 00:12:03,000 --> 00:12:06,040 Speaker 3: excel spridsheets, our sales or anymore because core coed is 259 00:12:06,080 --> 00:12:08,360 Speaker 3: going to do it for us. And there's been sort 260 00:12:08,360 --> 00:12:11,160 Speaker 3: of announcement after announcement of new AA company coming out 261 00:12:11,200 --> 00:12:14,199 Speaker 3: with software business. I think the market's sort of going right. 262 00:12:14,240 --> 00:12:16,160 Speaker 3: This is all negative, all negative, But has this year 263 00:12:16,240 --> 00:12:17,880 Speaker 3: is sold off enough now? And I think the market's 264 00:12:17,920 --> 00:12:20,360 Speaker 3: finally got to the point let's pause, Let's have a lot. 265 00:12:20,880 --> 00:12:24,120 Speaker 3: What's this company where Salesforce is trading around about fourteen 266 00:12:24,160 --> 00:12:28,120 Speaker 3: times it's earnings. I used to trade on thirty forty 267 00:12:28,200 --> 00:12:31,360 Speaker 3: fifty times earning. So it's sort of as evaluations sort 268 00:12:31,360 --> 00:12:33,959 Speaker 3: are overharved, and I think the market's now going right. 269 00:12:34,160 --> 00:12:35,520 Speaker 3: We're at a point here now I think we can 270 00:12:35,559 --> 00:12:37,680 Speaker 3: start having the debate as this overdone. 271 00:12:37,920 --> 00:12:39,920 Speaker 1: Yeah. At the same time, I think I saw somewhere 272 00:12:39,960 --> 00:12:42,120 Speaker 1: there was a point recently where Walmart was trading at 273 00:12:42,160 --> 00:12:46,760 Speaker 1: sort of a double the multiple that Microsoft was trading at, 274 00:12:46,760 --> 00:12:48,439 Speaker 1: which is sort of crazy is a tech company, it's 275 00:12:48,440 --> 00:12:52,160 Speaker 1: in the toilet basically, and here's a plain old business 276 00:12:52,160 --> 00:12:56,400 Speaker 1: that just helps roll stuff out the door like Briscoes 277 00:12:56,520 --> 00:12:58,000 Speaker 1: or or cam Art would do. 278 00:12:58,040 --> 00:13:00,320 Speaker 3: Here. Part of the reason you're seeing companies like Mart 279 00:13:01,200 --> 00:13:03,960 Speaker 3: and what we sort of call consumers staples companies, companies 280 00:13:03,960 --> 00:13:06,440 Speaker 3: that provide non discretionary services that you kind of have 281 00:13:06,520 --> 00:13:10,560 Speaker 3: to buy food and goods or healthcare. They've been trading 282 00:13:10,640 --> 00:13:13,760 Speaker 3: up quite high, particularly over the last few years, they've 283 00:13:13,760 --> 00:13:15,400 Speaker 3: been getting more expensive. Will Much's been next to you 284 00:13:15,520 --> 00:13:18,040 Speaker 3: in particularly well. But over the last sort of six 285 00:13:18,120 --> 00:13:20,920 Speaker 3: to eight weeks, as you started seeing AI and certainty 286 00:13:20,960 --> 00:13:24,000 Speaker 3: impact software stock selling off, investors have been sort of 287 00:13:24,040 --> 00:13:27,240 Speaker 3: craving or rotating to companies where they've got more earning certainty. 288 00:13:27,400 --> 00:13:28,920 Speaker 3: The companies where you just sort of think, we kind 289 00:13:28,920 --> 00:13:30,480 Speaker 3: of know what this company is going to earn next year. 290 00:13:30,520 --> 00:13:32,840 Speaker 3: They're going to pay me a dividend, may and exactly 291 00:13:32,920 --> 00:13:35,520 Speaker 3: pay you a dividend, And so you sort of touched 292 00:13:36,520 --> 00:13:38,840 Speaker 3: a premium to the valuation of companies where you've got 293 00:13:38,840 --> 00:13:41,000 Speaker 3: a little bit more confidence of what they're going to earn, 294 00:13:41,000 --> 00:13:42,760 Speaker 3: and you've got a bit of confidence that there isn't 295 00:13:42,800 --> 00:13:44,560 Speaker 3: going to disappear on you in the next couple of years. 296 00:13:44,960 --> 00:13:47,600 Speaker 3: And so the same reason people are selling software businesses 297 00:13:47,640 --> 00:13:49,320 Speaker 3: because the've sort of lost confidence in what they're going 298 00:13:49,360 --> 00:13:52,640 Speaker 3: to earn in five years time. People rotating to companies 299 00:13:52,679 --> 00:13:56,880 Speaker 3: like Walmarts. They're actually rotating into Telco's utility infrastructure businesses, 300 00:13:56,880 --> 00:13:58,679 Speaker 3: companies where you think AI is not really going to 301 00:13:58,760 --> 00:13:59,960 Speaker 3: replaced us anytime soon. 302 00:14:00,080 --> 00:14:01,960 Speaker 1: I want to come back to those two that we're 303 00:14:02,000 --> 00:14:04,319 Speaker 1: talking about there, and maybe even more specifically the sort 304 00:14:04,320 --> 00:14:07,280 Speaker 1: of in video. There's a company that, as you say, 305 00:14:07,760 --> 00:14:10,000 Speaker 1: it's got a huge future ahead of it, but a 306 00:14:10,000 --> 00:14:13,320 Speaker 1: lot of that is based on not money that's coming 307 00:14:13,360 --> 00:14:17,920 Speaker 1: from you or I using an application and paying anthropic 308 00:14:18,160 --> 00:14:20,560 Speaker 1: claud or whatever it is to use it. It's coming from 309 00:14:20,640 --> 00:14:24,120 Speaker 1: these companies, these hyperscaler companies, Meta and so on, that 310 00:14:24,200 --> 00:14:27,400 Speaker 1: are putting huge bets, huge capital purchases and so on 311 00:14:27,480 --> 00:14:30,800 Speaker 1: in there and saying, oh, yes, we'll buy this many chips, 312 00:14:30,800 --> 00:14:32,440 Speaker 1: We're going to invest this much and so on, and 313 00:14:32,440 --> 00:14:36,040 Speaker 1: that's where that's coming from. So I just want to 314 00:14:36,040 --> 00:14:38,920 Speaker 1: try and get a bit of a sense whether the 315 00:14:39,040 --> 00:14:42,640 Speaker 1: buildout is going to be sustained because the adoption part 316 00:14:42,640 --> 00:14:45,360 Speaker 1: doesn't seem to quite be turning into money at the moment. 317 00:14:45,640 --> 00:14:48,040 Speaker 3: As you say, the revenues in a video are linked 318 00:14:48,080 --> 00:14:53,480 Speaker 3: to continued capital expenditures from these hyperscalers Amazon, Microsoft, Meta 319 00:14:54,600 --> 00:14:57,400 Speaker 3: and the one, the willingness of them to continue to 320 00:14:57,440 --> 00:15:00,400 Speaker 3: spend that money, and then actually the ability for them 321 00:15:00,400 --> 00:15:02,640 Speaker 3: to continue to spend it by raising financing to keep 322 00:15:02,680 --> 00:15:05,640 Speaker 3: spending this. And you know, to give you some context, 323 00:15:06,160 --> 00:15:08,920 Speaker 3: these companies, you know, i'd sort of use round numbers 324 00:15:08,920 --> 00:15:11,600 Speaker 3: across them. All might have been spending around about thirty 325 00:15:11,600 --> 00:15:14,400 Speaker 3: billion dollars of capecks per year a couple of years ago. 326 00:15:14,640 --> 00:15:16,880 Speaker 3: We've got some of them spending two hundred billion dollars 327 00:15:16,920 --> 00:15:19,960 Speaker 3: per year. So Amazon was aspected to I think it's 328 00:15:20,000 --> 00:15:21,760 Speaker 3: coming into this year is expected to spend about one 329 00:15:21,840 --> 00:15:24,080 Speaker 3: hundred and thirty billion, one hundred and forty billion in CAPEX. 330 00:15:24,080 --> 00:15:26,320 Speaker 3: That just told everyone it's going to spend two hundred billion. 331 00:15:27,240 --> 00:15:29,600 Speaker 3: One company's spinning two hundred million on and capolic expendence 332 00:15:29,680 --> 00:15:31,920 Speaker 3: in a year. And so you can kind of you 333 00:15:31,920 --> 00:15:35,240 Speaker 3: can kind of track written in Nvidia's revenues. So Nvidia's 334 00:15:36,040 --> 00:15:37,920 Speaker 3: expected to generate revenues. So it was sort of about 335 00:15:37,920 --> 00:15:40,960 Speaker 3: three hundred and so we high three hundred billion dollars 336 00:15:40,960 --> 00:15:44,920 Speaker 3: next year might be three fifty to three seventy. That 337 00:15:45,000 --> 00:15:48,000 Speaker 3: number needs to be directly tied to these Hyperscaler's capecks, 338 00:15:48,000 --> 00:15:50,200 Speaker 3: and if you go through the hype scale capecks, they 339 00:15:51,000 --> 00:15:53,000 Speaker 3: they's been around about. You know, this is taking the 340 00:15:53,000 --> 00:15:59,360 Speaker 3: big five companies out there, Amazon, Meta, Google, Microsoft, and 341 00:15:59,400 --> 00:16:01,960 Speaker 3: you can sort of some unlisted ones like Xai in there. 342 00:16:02,680 --> 00:16:05,080 Speaker 3: They've spent about four hundred billion last year in capex. 343 00:16:05,120 --> 00:16:08,120 Speaker 3: That's suspected to probably be around about seven fifty billion 344 00:16:08,400 --> 00:16:11,560 Speaker 3: or even higher, might even go high seven hundred billions 345 00:16:11,600 --> 00:16:14,600 Speaker 3: in twenty twenty six. That's a massive jump. N Vidia 346 00:16:14,680 --> 00:16:17,440 Speaker 3: is going to catchure a big share of that, but 347 00:16:17,520 --> 00:16:19,880 Speaker 3: then market's probably comfortable. They probably go, yeah, we know 348 00:16:19,920 --> 00:16:21,400 Speaker 3: they're going to spend the capex this year, but they 349 00:16:21,440 --> 00:16:23,280 Speaker 3: need to do it again. We need to keep raising 350 00:16:23,320 --> 00:16:25,760 Speaker 3: that number and raising it. And these are companies that 351 00:16:25,800 --> 00:16:27,720 Speaker 3: haven't really had to go into debt very much at 352 00:16:27,720 --> 00:16:30,280 Speaker 3: all in the last ten years. You're now seeing them 353 00:16:30,280 --> 00:16:33,440 Speaker 3: be free casually negative. Meta is going to be free 354 00:16:33,440 --> 00:16:36,920 Speaker 3: casually negative this year. Amazon's going to be free cashing negative. 355 00:16:36,960 --> 00:16:40,440 Speaker 3: Google's barely making any for basically tipping in all of 356 00:16:40,440 --> 00:16:42,320 Speaker 3: the cash their generator and the most profitable They've got, 357 00:16:42,360 --> 00:16:44,360 Speaker 3: the most profitable business in the world. Right, these guys 358 00:16:44,400 --> 00:16:47,520 Speaker 3: generiferate hundreds of billions of cash, slow tipping all of 359 00:16:47,520 --> 00:16:51,600 Speaker 3: that in capex to build data centers, a big chunker 360 00:16:51,600 --> 00:16:54,680 Speaker 3: that goes to na Video. And then they still don't 361 00:16:54,680 --> 00:16:56,480 Speaker 3: actually have enough cash, so they're going to be raising debts. 362 00:16:56,480 --> 00:16:58,640 Speaker 3: So they're tapping the debt markets now to help fuel 363 00:16:58,680 --> 00:17:01,800 Speaker 3: this ride. And and so you know, way everything's going 364 00:17:01,800 --> 00:17:04,040 Speaker 3: well when people seeing ai've been used more on the 365 00:17:04,119 --> 00:17:06,600 Speaker 3: utilities working, and I think you can get confidence that 366 00:17:06,640 --> 00:17:08,720 Speaker 3: this keeps going. But what happens if you get a 367 00:17:08,720 --> 00:17:11,960 Speaker 3: period where all of sudden expectations and aosoftens. Then all 368 00:17:11,960 --> 00:17:13,639 Speaker 3: of a sudden, that's when the pressure comes on and 369 00:17:14,480 --> 00:17:16,560 Speaker 3: everyone wants to see there's a return. I mean, you're 370 00:17:16,600 --> 00:17:19,160 Speaker 3: going to spend two hundred millions dollars a capex. Investors 371 00:17:19,160 --> 00:17:21,320 Speaker 3: want to see a return generated on that capex, and 372 00:17:21,359 --> 00:17:23,520 Speaker 3: at the moment, there's pretty high uncertainty on what their 373 00:17:23,520 --> 00:17:24,359 Speaker 3: return is going to be. 374 00:17:24,760 --> 00:17:27,080 Speaker 1: Yeah, right, I bet maybe we should talk a bit 375 00:17:27,119 --> 00:17:29,320 Speaker 1: about what they're actually building, so we can sort of 376 00:17:29,320 --> 00:17:31,919 Speaker 1: see that these things are doing millions and millions of 377 00:17:31,960 --> 00:17:34,160 Speaker 1: processes the whole time. Is there any kind of data 378 00:17:34,200 --> 00:17:36,640 Speaker 1: point that we can do try and understand just what 379 00:17:36,680 --> 00:17:38,320 Speaker 1: these things are doing and how well they're doing it. 380 00:17:38,960 --> 00:17:42,840 Speaker 3: Yeah, it's good question. So let's break it down. What 381 00:17:43,560 --> 00:17:49,160 Speaker 3: artificial intelligence intelligence is is exactly that is providing intelligence 382 00:17:49,200 --> 00:17:51,679 Speaker 3: and doing things that humans would like to do or 383 00:17:51,760 --> 00:17:54,040 Speaker 3: do do, but doing it faster and better than we 384 00:17:54,080 --> 00:17:56,680 Speaker 3: can possibly do that. And to do that, they need 385 00:17:56,720 --> 00:18:01,240 Speaker 3: to analyze information and come up with good answers and 386 00:18:01,280 --> 00:18:03,600 Speaker 3: then provide it back to us. So basically, a request 387 00:18:03,640 --> 00:18:07,160 Speaker 3: gets put in. We type an English written language request 388 00:18:07,200 --> 00:18:12,199 Speaker 3: into a large language model interface like CHET, Chiput or Claude. 389 00:18:12,960 --> 00:18:16,080 Speaker 3: The LM LM sends a large language model so this 390 00:18:16,160 --> 00:18:19,400 Speaker 3: as we're talking, chet, Chiput or Claude or Gemini will 391 00:18:19,440 --> 00:18:22,440 Speaker 3: take that information. It will go scan the internet, scan 392 00:18:22,560 --> 00:18:27,480 Speaker 3: its database, analyze a segregator, maybe branch it off to 393 00:18:27,520 --> 00:18:30,360 Speaker 3: its own specialized model to get the best answer. Come 394 00:18:30,400 --> 00:18:33,080 Speaker 3: back with us and look at that answer, determine if 395 00:18:33,080 --> 00:18:34,720 Speaker 3: that looks right. No, I'm going to go find a 396 00:18:34,720 --> 00:18:37,399 Speaker 3: bit more information, make it better. Then I'm going to 397 00:18:37,760 --> 00:18:41,000 Speaker 3: send it back to you. All of that takes computing power, 398 00:18:41,440 --> 00:18:46,880 Speaker 3: where we're using computer chips to process all that information. Now, 399 00:18:46,920 --> 00:18:50,840 Speaker 3: the way last the models has sort of been designed 400 00:18:50,840 --> 00:18:53,080 Speaker 3: to work as they break down complex problems into what's 401 00:18:53,119 --> 00:18:57,160 Speaker 3: called tokens. And I think it's like for every one 402 00:18:57,280 --> 00:18:59,520 Speaker 3: word in the English language on average, it's something like 403 00:18:59,560 --> 00:19:03,280 Speaker 3: one point five tokens. But then also text images get 404 00:19:03,280 --> 00:19:05,680 Speaker 3: breaking down in tokens, Videos get breaking down in tokens, 405 00:19:06,280 --> 00:19:08,280 Speaker 3: and so tokens has become it's almost like the picture 406 00:19:08,320 --> 00:19:10,560 Speaker 3: all the fuel that large language models run on. You know, 407 00:19:10,680 --> 00:19:13,840 Speaker 3: I might do a request and say, look, I've got 408 00:19:13,840 --> 00:19:17,159 Speaker 3: an I've got an interview coming up with with shears 409 00:19:17,200 --> 00:19:19,280 Speaker 3: these can you these are the rough questions? Can you 410 00:19:19,320 --> 00:19:21,440 Speaker 3: provide me answers for them? Well, that's that's that might 411 00:19:21,480 --> 00:19:22,960 Speaker 3: say yep, thank you, I'll do that. But that's going 412 00:19:23,040 --> 00:19:25,600 Speaker 3: to cost you a hundred thousand tokens. Someone needs to 413 00:19:25,600 --> 00:19:27,920 Speaker 3: pay for that, right, and so so that's what these 414 00:19:27,920 --> 00:19:31,120 Speaker 3: guys are building. But to run those models, they need tokens. 415 00:19:31,119 --> 00:19:34,120 Speaker 3: To get tokens, they need to buy computer chips called 416 00:19:34,200 --> 00:19:37,160 Speaker 3: GPUs General process and units is what NA Video sells, 417 00:19:37,680 --> 00:19:41,280 Speaker 3: and they spit out these tokens. To run those GPUs, 418 00:19:41,560 --> 00:19:43,679 Speaker 3: you need to put them, attach them to power, attach 419 00:19:43,720 --> 00:19:46,399 Speaker 3: them to other networking equipment, attach them to other types 420 00:19:46,400 --> 00:19:48,560 Speaker 3: of computer chips and your how's all this? And what's 421 00:19:48,600 --> 00:19:51,399 Speaker 3: called a data center And a lot of people like 422 00:19:51,440 --> 00:19:54,199 Speaker 3: to call them intelligence factories or AI factories, and they 423 00:19:54,200 --> 00:19:56,200 Speaker 3: are the new factories of the world. But but he 424 00:19:56,320 --> 00:19:58,600 Speaker 3: hels them and all that. So yeah, just just to 425 00:19:58,600 --> 00:20:01,359 Speaker 3: give you some context, everyone talk and giga what's a power? 426 00:20:01,400 --> 00:20:03,960 Speaker 3: Now when you talk about building data centers, So one 427 00:20:03,960 --> 00:20:06,679 Speaker 3: giga what data center has become kind of like the 428 00:20:06,720 --> 00:20:08,359 Speaker 3: standard I'm building and you want it's a gigga what 429 00:20:08,440 --> 00:20:11,520 Speaker 3: data center here? That costs you about fifty billion dollars 430 00:20:11,680 --> 00:20:15,919 Speaker 3: to produce one. So you know Amazon's capex of if 431 00:20:15,960 --> 00:20:18,399 Speaker 3: we're talking about two hundred billion next year, that's about 432 00:20:18,440 --> 00:20:22,679 Speaker 3: four of these one giga what data centers. Out of 433 00:20:22,720 --> 00:20:26,200 Speaker 3: that fifty billion dollars, twenty five to thirty billion goes 434 00:20:26,200 --> 00:20:28,800 Speaker 3: to Navidia. So Navidio takes the line of share. So 435 00:20:29,200 --> 00:20:30,919 Speaker 3: even though you've got to go buy this land, you 436 00:20:31,000 --> 00:20:35,160 Speaker 3: set up there. You know, there's a many football football 437 00:20:35,200 --> 00:20:38,960 Speaker 3: field sizes and size, build all equipment, calling equipment computer 438 00:20:39,080 --> 00:20:41,480 Speaker 3: racks shells that go around it. Thirty billion of it 439 00:20:41,520 --> 00:20:43,880 Speaker 3: is still with actually the chips that's never there, and 440 00:20:44,280 --> 00:20:47,600 Speaker 3: that's the infrastructure put in place to produce these tokens 441 00:20:47,640 --> 00:20:50,800 Speaker 3: to run these large language models. The lms are largely 442 00:20:50,840 --> 00:20:54,199 Speaker 3: getting the hyperscalers such as Microsoft, Amazon, Google to build 443 00:20:54,200 --> 00:20:57,320 Speaker 3: those factories for them so they can use them and 444 00:20:57,400 --> 00:20:58,840 Speaker 3: sell their services to us. 445 00:20:59,520 --> 00:21:02,120 Speaker 1: This springs to an interesting piece that are just real Recently, 446 00:21:02,160 --> 00:21:04,080 Speaker 1: Moody's apparently has been taking a look at this sort 447 00:21:04,080 --> 00:21:06,760 Speaker 1: of stuff and saying, there's a problem with the balance 448 00:21:06,760 --> 00:21:09,479 Speaker 1: sheets here because some of those data centers that are 449 00:21:09,520 --> 00:21:12,400 Speaker 1: being purchased or committed to they're not going to need 450 00:21:12,440 --> 00:21:16,359 Speaker 1: them for a certain period of time, and supposedly, if 451 00:21:16,520 --> 00:21:19,199 Speaker 1: it's far enough in the future, you can kind of 452 00:21:19,240 --> 00:21:22,159 Speaker 1: treat it as like this is an accounting thing. Perhaps 453 00:21:22,200 --> 00:21:25,120 Speaker 1: you know, it's not an immediate liability running to put 454 00:21:25,119 --> 00:21:26,840 Speaker 1: it on an books. And they seem to be saying 455 00:21:26,880 --> 00:21:30,199 Speaker 1: there's like six hundred billion dollars worth of future liabilities 456 00:21:30,240 --> 00:21:32,879 Speaker 1: that are even showing up on the balance sheets of 457 00:21:32,920 --> 00:21:35,760 Speaker 1: some of these companies, which is equivalent to like more 458 00:21:35,840 --> 00:21:38,480 Speaker 1: than they're borrowed in the last year. Is that the 459 00:21:38,560 --> 00:21:41,919 Speaker 1: kind of scary talk that we should be paying attention 460 00:21:42,000 --> 00:21:42,720 Speaker 1: to or is it all. 461 00:21:42,560 --> 00:21:44,440 Speaker 3: Just going to be fine, it'll wash out. I probably 462 00:21:44,440 --> 00:21:48,000 Speaker 3: wouldn't get too paranoid about it right now, but I 463 00:21:48,000 --> 00:21:50,040 Speaker 3: think you're correct to point it out as something to 464 00:21:50,119 --> 00:21:52,880 Speaker 3: keep an eye on. So the debt levels of financing 465 00:21:52,920 --> 00:21:54,639 Speaker 3: to keep fund this out is getting bigger. Because we 466 00:21:54,720 --> 00:21:56,879 Speaker 3: started off just funding all this development out of cash 467 00:21:56,880 --> 00:21:59,520 Speaker 3: lows and generators all got no problem. We're now get 468 00:21:59,560 --> 00:22:01,840 Speaker 3: into such size of investment that we need to start 469 00:22:01,840 --> 00:22:04,120 Speaker 3: tapping into debt because you've got to remember, this investment 470 00:22:04,200 --> 00:22:08,000 Speaker 3: is leading what hopefully will be returns, but it doesn't 471 00:22:08,000 --> 00:22:09,920 Speaker 3: you don't monetize it straight away. At the moment, there's 472 00:22:09,960 --> 00:22:12,359 Speaker 3: a huge amount of this. I estimate it is probably 473 00:22:12,359 --> 00:22:14,800 Speaker 3: going to be a trillion dollars spent in artificial intelligence 474 00:22:14,840 --> 00:22:18,360 Speaker 3: CAPEX in twenty twenty six. The revenues were probably generating 475 00:22:18,359 --> 00:22:20,359 Speaker 3: about AI at the moment is probably somewhere between fifty 476 00:22:20,359 --> 00:22:22,760 Speaker 3: tow one hundred billion wow, So in one year was 477 00:22:22,800 --> 00:22:26,119 Speaker 3: probably spinning ten times what the total revenues across all 478 00:22:26,160 --> 00:22:28,480 Speaker 3: the AI has been doing out at the moment. And 479 00:22:28,640 --> 00:22:30,760 Speaker 3: we're now starting to finance that with debt. So I 480 00:22:30,800 --> 00:22:32,800 Speaker 3: financed that with debt of stuff we currently are not 481 00:22:32,840 --> 00:22:35,680 Speaker 3: seeing the returns. And then, as you sort of point out, 482 00:22:35,720 --> 00:22:37,960 Speaker 3: some of it is debt that's set straight out on 483 00:22:37,960 --> 00:22:41,040 Speaker 3: the balance sheet, but a lot of it is commitments 484 00:22:41,080 --> 00:22:45,439 Speaker 3: for future leases to operate these data centers. So you 485 00:22:45,480 --> 00:22:47,000 Speaker 3: pointed out that said, I think you might have said 486 00:22:47,000 --> 00:22:49,680 Speaker 3: six hundred billion and the off balance sheet commitments. One 487 00:22:49,680 --> 00:22:52,800 Speaker 3: company that's sort of at the forefront of sort of 488 00:22:53,960 --> 00:22:57,040 Speaker 3: using its balance sheet and leveraging up to build these 489 00:22:57,119 --> 00:23:00,359 Speaker 3: data centers or roll them out, as Oracle so Ora 490 00:23:00,400 --> 00:23:03,240 Speaker 3: call very very large business, probably is going to spend 491 00:23:03,280 --> 00:23:05,439 Speaker 3: fifty billion on KPX this year. We'll spin up to 492 00:23:05,440 --> 00:23:07,560 Speaker 3: one hundred billion a year in the next few years. 493 00:23:08,920 --> 00:23:10,800 Speaker 3: They already have it there bit a debt, they've already 494 00:23:10,840 --> 00:23:13,119 Speaker 3: run a sort of quite a high debt balance sheet. 495 00:23:13,400 --> 00:23:15,520 Speaker 3: What they're choosing to do is not build the data 496 00:23:15,520 --> 00:23:17,880 Speaker 3: center in the South. They'll get someone else to build 497 00:23:17,920 --> 00:23:20,200 Speaker 3: the physical data center of the show, get the power 498 00:23:20,200 --> 00:23:22,960 Speaker 3: connected up, and they'll lease that off them. And so 499 00:23:23,040 --> 00:23:25,159 Speaker 3: by leasing it off them, they'll say, look, here's a 500 00:23:25,160 --> 00:23:27,879 Speaker 3: twenty year commitment that we're going to pay you an 501 00:23:27,920 --> 00:23:30,840 Speaker 3: annual lease. So you can set up our shell and 502 00:23:30,880 --> 00:23:32,760 Speaker 3: we can go check our compute equipment and chips in 503 00:23:32,800 --> 00:23:34,560 Speaker 3: it and run it. But I want you to manage 504 00:23:34,600 --> 00:23:37,480 Speaker 3: the whole shell, the land, the property, getting the power in. 505 00:23:37,640 --> 00:23:39,080 Speaker 3: I don't want to do any of that. You do it. 506 00:23:39,840 --> 00:23:41,840 Speaker 3: But obviously there's got a big commitment to do that, 507 00:23:41,920 --> 00:23:45,159 Speaker 3: and so Nivity has got I think alone with I 508 00:23:45,200 --> 00:23:46,760 Speaker 3: think it's with open AI that're sort of trying to 509 00:23:46,800 --> 00:23:49,800 Speaker 3: do ten gigawattsworth of expansion. It's like five hundred billion 510 00:23:49,840 --> 00:23:52,800 Speaker 3: dollars over maybe the next ten years and other pieces, 511 00:23:53,280 --> 00:23:56,320 Speaker 3: and so those commitments add up, you know, massive dollars 512 00:23:56,359 --> 00:23:58,239 Speaker 3: and leases. So I think Oracle has got about six 513 00:23:58,359 --> 00:24:01,160 Speaker 3: hundred and two hundred and six the odd billion worth 514 00:24:01,160 --> 00:24:03,320 Speaker 3: of these commitments. They're not on the balance sheet, so 515 00:24:03,320 --> 00:24:04,800 Speaker 3: you go look at the balance sheet, they won't appear 516 00:24:04,800 --> 00:24:07,440 Speaker 3: an Oracle's debt level. But over the next of five 517 00:24:07,480 --> 00:24:10,320 Speaker 3: ten years, as these data sts come online, they will 518 00:24:10,359 --> 00:24:12,480 Speaker 3: come onto the balance sheet. At that point, it's basically 519 00:24:12,520 --> 00:24:16,120 Speaker 3: saying it's really committing heavily, and Moody is aware of this, saying, look, yeah, 520 00:24:16,119 --> 00:24:17,679 Speaker 3: your debt level is this today, but we kind of 521 00:24:17,680 --> 00:24:19,680 Speaker 3: need to look forward a little bit here into what's 522 00:24:19,720 --> 00:24:23,040 Speaker 3: going on and what your debt level maybe in one, two, 523 00:24:23,040 --> 00:24:23,720 Speaker 3: three years time. 524 00:24:24,320 --> 00:24:26,480 Speaker 1: So there was a story recently as well about a 525 00:24:26,480 --> 00:24:28,919 Speaker 1: little research note from an outfit called Satrini, one bit 526 00:24:28,960 --> 00:24:35,040 Speaker 1: of shade thrown on the prospects of AI dislocating the 527 00:24:35,040 --> 00:24:37,159 Speaker 1: amount of employment that's going to be there in the US, 528 00:24:37,560 --> 00:24:39,800 Speaker 1: and suddenly people are selling off all of these companies, 529 00:24:39,840 --> 00:24:42,919 Speaker 1: going oh God, what is going on a with that 530 00:24:43,080 --> 00:24:45,199 Speaker 1: and be in a market where one little note can 531 00:24:45,200 --> 00:24:46,639 Speaker 1: suddenly freak everybody out. 532 00:24:47,040 --> 00:24:51,520 Speaker 3: So Satrinia as a quite a highly regarded called it 533 00:24:51,680 --> 00:24:56,280 Speaker 3: thematical type research house. So they like to think of 534 00:24:56,280 --> 00:24:59,040 Speaker 3: of what's going on, what might be happening in the future. 535 00:24:59,160 --> 00:25:01,360 Speaker 3: But actually what they wrote in the article was things 536 00:25:01,400 --> 00:25:03,840 Speaker 3: we've been talking about and thinking about sort of throughout 537 00:25:03,880 --> 00:25:05,639 Speaker 3: the start of this year, right. Actually a lot of 538 00:25:05,680 --> 00:25:08,080 Speaker 3: the some of the movements and share price section you 539 00:25:08,119 --> 00:25:10,520 Speaker 3: saw happened well before the Ctrinia article came out. And 540 00:25:10,560 --> 00:25:14,480 Speaker 3: so I've i explained what the paper broadly is telling us. 541 00:25:15,480 --> 00:25:17,440 Speaker 3: It sort of starts off by saying, we spent a 542 00:25:17,480 --> 00:25:19,760 Speaker 3: lot of time as investors we kind of think about 543 00:25:19,800 --> 00:25:22,840 Speaker 3: the bare case of AI, it doesn't work, and all 544 00:25:22,880 --> 00:25:25,760 Speaker 3: this over spin doesn't return anything, and even the base 545 00:25:25,800 --> 00:25:28,760 Speaker 3: case it sort of works okay, But until maybe this 546 00:25:28,920 --> 00:25:31,280 Speaker 3: year when we saw how fast clawed code and clud 547 00:25:31,320 --> 00:25:34,080 Speaker 3: cowork has progressed, and everyone's like, wow, this is AI 548 00:25:34,160 --> 00:25:35,560 Speaker 3: is here now, we're always sort of think it's a 549 00:25:35,640 --> 00:25:37,919 Speaker 3: year or two. We haven't too much thought about the 550 00:25:37,920 --> 00:25:41,600 Speaker 3: ball case where AI really is really helpful and really 551 00:25:41,640 --> 00:25:44,600 Speaker 3: changes how we work to the point where it is 552 00:25:44,640 --> 00:25:46,440 Speaker 3: so good that we don't need a lot of the 553 00:25:46,520 --> 00:25:48,720 Speaker 3: childs and all the companies that are sitting there on 554 00:25:48,760 --> 00:25:51,560 Speaker 3: all that with big, large white collar workforces, go, look, 555 00:25:51,560 --> 00:25:54,000 Speaker 3: we don't need this person who's essentially just processing data 556 00:25:54,040 --> 00:25:55,960 Speaker 3: because AI can do it faster than he can do, 557 00:25:56,000 --> 00:25:57,440 Speaker 3: and it can do it better, and it works twenty 558 00:25:57,440 --> 00:25:59,399 Speaker 3: four hours even a day. So we're going to get 559 00:25:59,440 --> 00:26:02,200 Speaker 3: rid of this guy costs US eighty grand a year, 560 00:26:02,240 --> 00:26:04,160 Speaker 3: and we're going to go get an AI that does 561 00:26:04,200 --> 00:26:07,199 Speaker 3: this for a very small margin of the price. At 562 00:26:07,240 --> 00:26:09,440 Speaker 3: the moment we spend I think they sort of said 563 00:26:10,240 --> 00:26:14,280 Speaker 3: round numbers, sixty percent of US GDP is spent on labor, 564 00:26:14,600 --> 00:26:16,880 Speaker 3: and a huge portion of that's the white collar workforce. 565 00:26:17,000 --> 00:26:20,120 Speaker 3: What if AI is so good that we can replace 566 00:26:20,480 --> 00:26:23,640 Speaker 3: a big portion of those jobs with AI. So if 567 00:26:23,640 --> 00:26:26,360 Speaker 3: that happens, we stopped spending the money to the white 568 00:26:26,400 --> 00:26:28,919 Speaker 3: collar people, we stopped getting paid our salaries. I'm a 569 00:26:28,920 --> 00:26:31,440 Speaker 3: white collar worker. I stopped getting paid by salary, for example. 570 00:26:31,520 --> 00:26:33,280 Speaker 3: So they all of a sudden, you're not paying those humans, 571 00:26:33,600 --> 00:26:37,560 Speaker 3: they lose their jobs, unemployment levels go higher. They stopped 572 00:26:37,560 --> 00:26:40,480 Speaker 3: getting paid. So previously, you might have had a tech 573 00:26:40,520 --> 00:26:43,600 Speaker 3: revolution which comes along and it impacts a very smissit 574 00:26:43,640 --> 00:26:46,160 Speaker 3: specific part. I mean, computers came along with something, and 575 00:26:46,520 --> 00:26:48,159 Speaker 3: it made it hard to kind of like people that 576 00:26:48,160 --> 00:26:50,479 Speaker 3: did manual calculations got replaced or you didn't need as 577 00:26:50,520 --> 00:26:52,280 Speaker 3: many of them. I mean, AI might be able to 578 00:26:52,320 --> 00:26:54,960 Speaker 3: replace all sorts of industries or actually have a big impact. 579 00:26:55,040 --> 00:26:56,919 Speaker 3: So you can't go find another white collar job. So 580 00:26:56,960 --> 00:26:58,800 Speaker 3: then you actually have to go find a blue collar job. 581 00:26:59,080 --> 00:27:01,520 Speaker 3: You go be a builder, but there's already heaps the builders, 582 00:27:01,520 --> 00:27:03,120 Speaker 3: so there's more he runs fighting over the same job. 583 00:27:03,200 --> 00:27:06,080 Speaker 3: So it sort of entertains this sort of quite doomsday 584 00:27:06,119 --> 00:27:09,119 Speaker 3: type scenario where AI is actually so good it causes 585 00:27:09,119 --> 00:27:11,800 Speaker 3: a big societal disruption and in the way we work 586 00:27:11,840 --> 00:27:13,960 Speaker 3: and the debate. There's been a lot of sort of 587 00:27:13,960 --> 00:27:16,320 Speaker 3: following since that article of people take on the other side. 588 00:27:16,359 --> 00:27:19,119 Speaker 3: The debate is now firmly sitting lot. Is it going 589 00:27:19,160 --> 00:27:22,080 Speaker 3: to disrupting so much that there's a big sort of 590 00:27:22,119 --> 00:27:25,480 Speaker 3: unemployment levels recession that gets caused for it? We can't 591 00:27:25,520 --> 00:27:28,000 Speaker 3: get these white collar people retrained to work somewhere all 592 00:27:28,040 --> 00:27:30,000 Speaker 3: will Will it go at a pace? Who We've got 593 00:27:30,000 --> 00:27:32,679 Speaker 3: time for people to reskill work with AI, come up 594 00:27:32,680 --> 00:27:35,720 Speaker 3: with new companies, and also jobs get created with new companies. 595 00:27:35,720 --> 00:27:37,520 Speaker 3: And I think the answer is going to be a 596 00:27:37,520 --> 00:27:40,280 Speaker 3: bit of both. And you know how we are viewing 597 00:27:40,320 --> 00:27:42,119 Speaker 3: it is increasingly we think there is going to be 598 00:27:42,119 --> 00:27:44,040 Speaker 3: some disruption to the white collar workforce. 599 00:27:44,200 --> 00:27:46,840 Speaker 1: We're talking before about scarcity and how that's the opportunity 600 00:27:46,840 --> 00:27:49,720 Speaker 1: in this game in some cases, where are the bottlenecks 601 00:27:49,760 --> 00:27:52,600 Speaker 1: and who are the businesses, the companies the opportunities to 602 00:27:52,680 --> 00:27:55,000 Speaker 1: invest that are potentially exposed to that. 603 00:27:55,560 --> 00:27:57,399 Speaker 3: Yeah, I can talk about a few I won't give 604 00:27:57,440 --> 00:27:59,439 Speaker 3: investment advice, but I'll give you a few perspectives of 605 00:28:00,080 --> 00:28:04,000 Speaker 3: a sort of areas where we see constraints in the 606 00:28:04,040 --> 00:28:07,840 Speaker 3: AI supply chain. And this is sort of areas that's 607 00:28:07,840 --> 00:28:10,280 Speaker 3: been talked about. It's nothing particularly new, but I think 608 00:28:10,280 --> 00:28:13,080 Speaker 3: there are a few areas which maybe some people won't 609 00:28:13,080 --> 00:28:16,960 Speaker 3: be aware of and becoming bigger constraints. So the sort 610 00:28:17,000 --> 00:28:19,119 Speaker 3: of most talked about area has been power. Over the 611 00:28:19,160 --> 00:28:23,240 Speaker 3: last lasted eighteen months, it's been sort of the big 612 00:28:23,280 --> 00:28:25,800 Speaker 3: topic of is there enough power and when we're going 613 00:28:25,840 --> 00:28:27,640 Speaker 3: to have to fuel all these data centers in each 614 00:28:27,680 --> 00:28:29,840 Speaker 3: gig of what they require? These power needs to come 615 00:28:29,840 --> 00:28:30,399 Speaker 3: from somewhere. 616 00:28:30,400 --> 00:28:32,719 Speaker 1: And it's a political problem in the States, right. I mean, 617 00:28:32,720 --> 00:28:34,440 Speaker 1: you've got out the company spending the thick end of 618 00:28:34,480 --> 00:28:37,320 Speaker 1: one hundred million New Zealand dollars trying to get favorable 619 00:28:37,359 --> 00:28:39,520 Speaker 1: legislation and states so that they can do the build 620 00:28:39,520 --> 00:28:41,840 Speaker 1: out because they know that there's this pushback coming from 621 00:28:41,840 --> 00:28:45,200 Speaker 1: the locals about you sticking that great, big data center 622 00:28:45,280 --> 00:28:46,960 Speaker 1: up there that's going to make my passion. 623 00:28:46,800 --> 00:28:49,280 Speaker 3: Power prices go up. And Trump's actually even just recently 624 00:28:49,320 --> 00:28:51,560 Speaker 3: called the CEOs of all the largest tech firms into 625 00:28:51,600 --> 00:28:54,320 Speaker 3: his office and he's basically trying to push some of 626 00:28:54,320 --> 00:28:57,520 Speaker 3: the onus on them to build their power or protect 627 00:28:57,560 --> 00:28:59,400 Speaker 3: the power. And you need a huge amount of power. 628 00:28:59,480 --> 00:29:01,440 Speaker 3: I think casts has sort of over the next number 629 00:29:01,440 --> 00:29:04,000 Speaker 3: of years, that AI data center power consumption will go 630 00:29:04,040 --> 00:29:06,360 Speaker 3: to I think somewhere from a low couple percent two 631 00:29:06,360 --> 00:29:10,680 Speaker 3: percent area to ten percent of US power consumption. And 632 00:29:10,720 --> 00:29:13,800 Speaker 3: you've got to remember that that power use in the 633 00:29:13,920 --> 00:29:16,760 Speaker 3: US and most developed nations globally has been declining for 634 00:29:16,760 --> 00:29:19,080 Speaker 3: the last twenty years. And there's the fact that until 635 00:29:19,320 --> 00:29:21,000 Speaker 3: I sort of looked into this over the last couple years, 636 00:29:21,000 --> 00:29:23,000 Speaker 3: I didn't realize but power consumption didn't tend to go 637 00:29:23,040 --> 00:29:25,760 Speaker 3: down because things are becoming more efficient. 638 00:29:26,280 --> 00:29:29,400 Speaker 1: And the industrialization, all the factories and stuff moving off. 639 00:29:29,320 --> 00:29:33,040 Speaker 3: SI exactly, and so pipelines and everything was set up 640 00:29:33,080 --> 00:29:36,120 Speaker 3: for what wasn't really a growing demand for power, now 641 00:29:36,280 --> 00:29:38,560 Speaker 3: that that it has grown. But when we sort of 642 00:29:38,560 --> 00:29:42,480 Speaker 3: around the numbers at Milford about six months ago, based 643 00:29:42,480 --> 00:29:44,560 Speaker 3: on sort of reasonable estimates of where we thought our 644 00:29:44,680 --> 00:29:46,880 Speaker 3: capics was going the power that there's enough power on 645 00:29:46,920 --> 00:29:49,480 Speaker 3: the US over the next five years and the grid 646 00:29:49,520 --> 00:29:51,760 Speaker 3: that's expected to come on. The problem was getting that 647 00:29:51,760 --> 00:29:54,280 Speaker 3: power online at the right place the right time to 648 00:29:54,280 --> 00:29:56,440 Speaker 3: get that data center. And we have to do it 649 00:29:56,480 --> 00:29:59,160 Speaker 3: quickly because it can take five years to get new 650 00:29:59,200 --> 00:30:01,360 Speaker 3: power online. All that is a whole huge amount of 651 00:30:01,400 --> 00:30:06,120 Speaker 3: regulatory compliance hurdles to get through. Over the last four 652 00:30:06,200 --> 00:30:10,640 Speaker 3: or five months, our estimate of sort of where AI 653 00:30:10,800 --> 00:30:12,560 Speaker 3: capex and the amount of gig what's coming on is 654 00:30:12,600 --> 00:30:14,720 Speaker 3: starting to go up higher. So you're now starting to 655 00:30:14,760 --> 00:30:16,400 Speaker 3: get to that point that actually there may not be 656 00:30:16,520 --> 00:30:19,160 Speaker 3: enough power generation coming on. So I think that's an 657 00:30:19,200 --> 00:30:21,720 Speaker 3: area I think where you've got to see sort of 658 00:30:21,720 --> 00:30:24,120 Speaker 3: more focus. And then, of course chips. I haven't talked 659 00:30:24,160 --> 00:30:25,880 Speaker 3: about ships, which is one of the oldest areas you 660 00:30:25,920 --> 00:30:29,920 Speaker 3: talk to. Chips is a big beneficiary until you really 661 00:30:30,000 --> 00:30:32,360 Speaker 3: until the last six months or so, as much as 662 00:30:32,360 --> 00:30:34,680 Speaker 3: everyone says it's side demand for chips, demand for chips, 663 00:30:34,720 --> 00:30:37,640 Speaker 3: there weren't really supply constrained. The constraint was more getting 664 00:30:37,640 --> 00:30:40,560 Speaker 3: the data centers online up to fill the chips. As in, 665 00:30:40,760 --> 00:30:42,240 Speaker 3: if you've got the space and the data center, you 666 00:30:42,240 --> 00:30:43,840 Speaker 3: can get the chips in there. That's sort of been 667 00:30:44,120 --> 00:30:47,360 Speaker 3: where it's at now very firmly starting to move that 668 00:30:47,440 --> 00:30:49,360 Speaker 3: we can't get enough chips. There's actually quite a few 669 00:30:49,400 --> 00:30:53,080 Speaker 3: data centers coming on. They're still constrained, but we're actually 670 00:30:53,160 --> 00:30:55,200 Speaker 3: hitting the limits of how many chips we can produce. 671 00:30:55,240 --> 00:30:57,160 Speaker 3: And there's two key areas in this There's one that 672 00:30:57,280 --> 00:31:02,160 Speaker 3: is what's called leading air foundry. There's a business called 673 00:31:02,240 --> 00:31:07,120 Speaker 3: t s MC Taiwan Semiconductor Manufacturing Company based out of Taiwan. 674 00:31:07,160 --> 00:31:09,280 Speaker 3: It's got about a two trillion US dollar market cap 675 00:31:09,280 --> 00:31:11,640 Speaker 3: and it makes almost one hundred percent of the AI 676 00:31:11,720 --> 00:31:14,160 Speaker 3: chips in the world. That company is now basically sold 677 00:31:14,160 --> 00:31:16,280 Speaker 3: out and what it can do for the next two years. 678 00:31:17,360 --> 00:31:20,600 Speaker 3: So and they called wafers. They produce wafers and that's 679 00:31:20,600 --> 00:31:23,600 Speaker 3: what these real tiny transistors gone to produce. Chips sold 680 00:31:23,640 --> 00:31:26,120 Speaker 3: up for abound about two years. They can't produce much more. 681 00:31:26,160 --> 00:31:28,000 Speaker 3: They're doing everything they can to get more online, but 682 00:31:28,000 --> 00:31:29,600 Speaker 3: it takes about three or four years to build a 683 00:31:29,640 --> 00:31:34,000 Speaker 3: new new manufacturing facility for these chips. So they're starting 684 00:31:34,040 --> 00:31:37,600 Speaker 3: to hold this supply chain back. Going with that, for 685 00:31:37,640 --> 00:31:41,160 Speaker 3: every compute powered chip, they now attach what's called a 686 00:31:41,240 --> 00:31:44,640 Speaker 3: memory chip to it. Think of the the RAM in 687 00:31:44,640 --> 00:31:46,240 Speaker 3: your laptop, and you know, we always get sick of 688 00:31:46,240 --> 00:31:48,240 Speaker 3: slow laptops and like any more RAM. I've got six 689 00:31:48,240 --> 00:31:49,520 Speaker 3: and in Gigabyt it's going to be a thirty two 690 00:31:49,560 --> 00:31:52,800 Speaker 3: Google like. RAM is kind of like the brains. It's 691 00:31:52,840 --> 00:31:56,120 Speaker 3: what stores short term memory. Now this this ram, it's 692 00:31:56,120 --> 00:31:59,000 Speaker 3: called de RAM, and the technical part of the industry, 693 00:31:59,080 --> 00:32:02,680 Speaker 3: this this de RAM is now basically completely sold out 694 00:32:02,680 --> 00:32:04,840 Speaker 3: for the next two years as well. You're getting the 695 00:32:04,840 --> 00:32:06,960 Speaker 3: biggest customers in the world only getting filled for about 696 00:32:06,960 --> 00:32:10,400 Speaker 3: fifty percent to two thirds of their orders. Prices have 697 00:32:10,440 --> 00:32:12,720 Speaker 3: gone up somewhere around about ten times in the last 698 00:32:13,480 --> 00:32:17,360 Speaker 3: six months alone, and it's becoming a massive constraint. All 699 00:32:17,400 --> 00:32:19,480 Speaker 3: really interesting areas to look at in terms of this 700 00:32:20,280 --> 00:32:23,200 Speaker 3: sort of supply chain constraint around AO buildout. 701 00:32:23,640 --> 00:32:26,960 Speaker 1: Andrew Kurtain, I feel like I've learned a lot today 702 00:32:27,440 --> 00:32:28,640 Speaker 1: just listening to all of this, so I'm going to 703 00:32:28,640 --> 00:32:29,960 Speaker 1: have to go away and do even more homework to 704 00:32:30,040 --> 00:32:31,880 Speaker 1: try and grasp But I just want to leave you 705 00:32:32,120 --> 00:32:34,600 Speaker 1: with one question and you can try and sum it 706 00:32:34,680 --> 00:32:36,200 Speaker 1: up if you like. What's the one thing you think 707 00:32:36,240 --> 00:32:40,880 Speaker 1: that retail investors most misunderstand about AI investment right now? 708 00:32:41,480 --> 00:32:42,960 Speaker 3: One thing a lot of people don't appreciate. And I 709 00:32:42,960 --> 00:32:45,240 Speaker 3: wouldn't say this is just retail investors. I think this 710 00:32:45,360 --> 00:32:51,600 Speaker 3: is professional investors as well. Is how uncertain technology revolutions are. 711 00:32:51,960 --> 00:32:53,440 Speaker 3: I think when you're sitting in that you sort of 712 00:32:53,440 --> 00:32:55,600 Speaker 3: feel like you sort of see what's going on and 713 00:32:55,640 --> 00:32:57,479 Speaker 3: it sort of feels really clear who the winners are 714 00:32:57,480 --> 00:33:00,120 Speaker 3: going to be. And we look at the video, we 715 00:33:00,120 --> 00:33:03,360 Speaker 3: look at open AI andthropic. If you go through your history, 716 00:33:05,480 --> 00:33:08,720 Speaker 3: history will show it's very difficult to predict how the 717 00:33:08,800 --> 00:33:11,760 Speaker 3: technology will develop over five ten years and who the 718 00:33:11,800 --> 00:33:15,040 Speaker 3: winners will be. In most technologic revolutions, companies you thought 719 00:33:15,080 --> 00:33:16,640 Speaker 3: were going to be the winners at the start don't 720 00:33:16,680 --> 00:33:17,840 Speaker 3: end up being the winners. And it might be a 721 00:33:17,840 --> 00:33:20,560 Speaker 3: company that you never heard of that comes out and wins. 722 00:33:20,560 --> 00:33:22,320 Speaker 3: And I just apply that to AI at the moment. 723 00:33:22,400 --> 00:33:25,600 Speaker 3: I mean, we have very little idea of exactly how 724 00:33:25,640 --> 00:33:27,040 Speaker 3: this is going to play out in a five year 725 00:33:27,080 --> 00:33:29,920 Speaker 3: of view. We don't know who's going to monetize things. Well, 726 00:33:30,040 --> 00:33:32,880 Speaker 3: are the LM's going to rule everything? Are they too 727 00:33:32,880 --> 00:33:35,120 Speaker 3: competitive because there's five of them all competition? They don't 728 00:33:35,160 --> 00:33:37,840 Speaker 3: make any money. But one thing you tend to get 729 00:33:37,840 --> 00:33:40,520 Speaker 3: in technology revolutions is value. A cruise to the consumer. 730 00:33:40,880 --> 00:33:44,760 Speaker 3: We get better technology, cheaper prices, and makes our life better. 731 00:33:45,040 --> 00:33:47,560 Speaker 3: And then the investment side of thing is a bit 732 00:33:47,560 --> 00:33:49,560 Speaker 3: more complex. Who wins? So I think the saying's going 733 00:33:49,560 --> 00:33:51,280 Speaker 3: to apply here for this Ai revolution. 734 00:33:52,720 --> 00:33:54,600 Speaker 1: Well, thanks for giving us a front row seat on 735 00:33:54,880 --> 00:33:58,080 Speaker 1: the next six months, twelve months and what's going on 736 00:33:58,200 --> 00:34:00,880 Speaker 1: inside your here at the moment. Andrew Contain from Milford 737 00:34:00,920 --> 00:34:05,000 Speaker 1: Esset Management, thank you very much, and you for listening watching, 738 00:34:05,120 --> 00:34:08,320 Speaker 1: Thank you for your attention, Koma too. That's the shared 739 00:34:08,400 --> 00:34:09,319 Speaker 1: lunch for this week.