1 00:00:00,280 --> 00:00:04,040 Speaker 1: Hi, this is Caroline Hyde from Bloomberg Tech. Today we're 2 00:00:04,080 --> 00:00:07,080 Speaker 1: sharing something a little different in your feed, an episode 3 00:00:07,080 --> 00:00:10,840 Speaker 1: from our colleagues at Here's Why, Bloomberg's weekly show that 4 00:00:10,920 --> 00:00:14,800 Speaker 1: answers one big question in under ten minutes. Host Stephen 5 00:00:14,800 --> 00:00:17,960 Speaker 1: Carroll is joined by our Bloomberg Tech europanker Tom McKenzie 6 00:00:18,360 --> 00:00:20,720 Speaker 1: to dive into a story that's right at the heart 7 00:00:20,760 --> 00:00:24,200 Speaker 1: of the tech world, the massive investments in AI data 8 00:00:24,239 --> 00:00:27,200 Speaker 1: centers and the hidden costs that come with them. If 9 00:00:27,200 --> 00:00:30,280 Speaker 1: you'd like to hear more episodes of Here's Why, you'll 10 00:00:30,320 --> 00:00:32,600 Speaker 1: find a link to the podcast feed in the show notes. 11 00:00:33,120 --> 00:00:34,200 Speaker 1: Hope you enjoy. 12 00:00:36,720 --> 00:00:44,000 Speaker 2: Bloomberg Audio Studios, podcasts, radio news. I'm Stephen Carol and 13 00:00:44,120 --> 00:00:46,640 Speaker 2: this is Here's Why, where we take one new story 14 00:00:46,680 --> 00:00:48,800 Speaker 2: and explain it in just a few minutes with our 15 00:00:48,840 --> 00:00:50,360 Speaker 2: experts here at Bloomberg. 16 00:00:53,640 --> 00:00:56,880 Speaker 3: It's ten thirty pm in this AI party. It started 17 00:00:57,000 --> 00:00:59,880 Speaker 3: nine pm now and that party goes to four am, 18 00:01:00,040 --> 00:01:02,080 Speaker 3: and the reality is like, look, this is going to 19 00:01:02,080 --> 00:01:04,600 Speaker 3: be a two to three year left in this bull 20 00:01:04,680 --> 00:01:07,760 Speaker 3: cycle for tech. The tech sector is very strong because 21 00:01:08,040 --> 00:01:11,920 Speaker 3: artificial intelligence is really a qualitative leap in the kind 22 00:01:11,920 --> 00:01:15,240 Speaker 3: of technology that we've had over the last several decades. 23 00:01:15,400 --> 00:01:19,520 Speaker 3: You're seeing an exponential growth of adoption and use of AI. 24 00:01:19,880 --> 00:01:22,280 Speaker 4: The number of applications that are going to be using 25 00:01:22,280 --> 00:01:23,960 Speaker 4: these AI is also growing. 26 00:01:24,480 --> 00:01:27,600 Speaker 2: Everyone has an opinion on where the AI frenzy is 27 00:01:27,640 --> 00:01:31,839 Speaker 2: going next. While optimism is rampant about the technologies potential, 28 00:01:32,200 --> 00:01:35,720 Speaker 2: more questions are now being asked about AI's running costs. 29 00:01:36,280 --> 00:01:40,680 Speaker 4: We are putting mostly chips silicon into these data centers 30 00:01:41,000 --> 00:01:44,720 Speaker 4: that have a lifespan of perhaps four years, So those 31 00:01:44,760 --> 00:01:46,920 Speaker 4: chips they appreciate very quickly. 32 00:01:47,040 --> 00:01:49,919 Speaker 3: Even in video, there's a new chip every eight months 33 00:01:49,920 --> 00:01:52,000 Speaker 3: and it's ten times as powerful as the earlier ones. 34 00:01:52,160 --> 00:01:54,360 Speaker 4: The thing with the valley this here is that almost 35 00:01:54,360 --> 00:01:57,320 Speaker 4: every investor knows it's all going to turn into pumpkins 36 00:01:57,360 --> 00:02:00,280 Speaker 4: and mice at midnight. Only, as Buffett would say, no 37 00:02:00,320 --> 00:02:01,480 Speaker 4: one in the room as a clock. 38 00:02:02,040 --> 00:02:06,280 Speaker 2: Even with bumper results and bullish revenue forecasts, here's why 39 00:02:06,400 --> 00:02:13,359 Speaker 2: AI casts still worry investors. Tom McKenzie, who hosts Bloomberg 40 00:02:13,400 --> 00:02:16,960 Speaker 2: Tech You're U bum Bloberg Television, joins me now for more. Tom. 41 00:02:17,320 --> 00:02:20,080 Speaker 2: The investor Michael Burry of Big Short fame is among 42 00:02:20,160 --> 00:02:23,280 Speaker 2: those who's worried about these future casts of AI and 43 00:02:23,440 --> 00:02:25,600 Speaker 2: data centers in particular. 44 00:02:25,680 --> 00:02:29,680 Speaker 4: What's the concern, Yeah, absolutely, Michael Burry putting on famously 45 00:02:29,720 --> 00:02:33,320 Speaker 4: short positions, so shortening the stocks of Nvidia and Pallanteer 46 00:02:33,400 --> 00:02:37,880 Speaker 4: before he wrapped up his fund. His concern does focus 47 00:02:38,360 --> 00:02:42,240 Speaker 4: on the depreciation of some of these assets by assets. 48 00:02:42,240 --> 00:02:47,120 Speaker 4: I'm talking about specifically these AI chips, very expensive AI accelerators. 49 00:02:47,240 --> 00:02:51,079 Speaker 4: Ninety percent of the market share is dominated by Nvidia, 50 00:02:51,280 --> 00:02:54,799 Speaker 4: so across the sale of these chips and video has 51 00:02:54,840 --> 00:02:58,720 Speaker 4: that significant market gain versus its rivals. And the concern 52 00:02:58,880 --> 00:03:01,480 Speaker 4: is that as you get newer versions of these chips, 53 00:03:01,480 --> 00:03:05,440 Speaker 4: the older ones essentially become less valuable. And Michael Barrie 54 00:03:05,960 --> 00:03:09,760 Speaker 4: making the argument that companies the hyperscalers, so the Microsofts 55 00:03:09,800 --> 00:03:12,600 Speaker 4: and alphabets and metas of the world, are not properly 56 00:03:12,639 --> 00:03:18,200 Speaker 4: accounting for how quickly these these assets depreciate. The other 57 00:03:18,240 --> 00:03:20,440 Speaker 4: part of the concern and kind of ties into this 58 00:03:20,840 --> 00:03:23,440 Speaker 4: that you hear voice from the skeptics around the AI 59 00:03:23,520 --> 00:03:26,560 Speaker 4: bubble is that there are comparisons, they say, with what 60 00:03:26,680 --> 00:03:29,480 Speaker 4: happened in the late nineteen nineties, nineteen ninety nine, early 61 00:03:29,520 --> 00:03:31,560 Speaker 4: two thousand, the dot com bubble, when it was the 62 00:03:31,639 --> 00:03:36,640 Speaker 4: telecom equipment makers that leading up to all of the 63 00:03:36,760 --> 00:03:40,480 Speaker 4: online expectations around how our digital economy was going to change, 64 00:03:40,640 --> 00:03:43,880 Speaker 4: spent huge amounts of money on building the infrastructure to 65 00:03:44,080 --> 00:03:49,040 Speaker 4: power the dot com era and ended up losing a 66 00:03:49,080 --> 00:03:52,120 Speaker 4: lot of money because the gains didn't come as quickly, 67 00:03:52,160 --> 00:03:55,360 Speaker 4: the technology didn't evolve as rapidly as they had expected. 68 00:03:55,440 --> 00:03:57,200 Speaker 4: Of course, on the back of that, you did get 69 00:03:57,600 --> 00:04:01,280 Speaker 4: some very significant players like Amazon who came through the 70 00:04:01,320 --> 00:04:04,080 Speaker 4: dot com bubble and of course now remain one of 71 00:04:04,080 --> 00:04:06,480 Speaker 4: the most valuable companies on the planet. But there was 72 00:04:06,560 --> 00:04:08,480 Speaker 4: a lot of capital, there was a lot of investment 73 00:04:08,480 --> 00:04:10,720 Speaker 4: that was burnt in that process. And so that is 74 00:04:10,760 --> 00:04:13,200 Speaker 4: another comparison that people are making. It's the depreciation around 75 00:04:13,200 --> 00:04:15,480 Speaker 4: the assets and the chips that they're worried about, but 76 00:04:15,560 --> 00:04:18,320 Speaker 4: also comparisons with what happened during the dot com era 77 00:04:18,400 --> 00:04:20,960 Speaker 4: and the pain that was felt by those telecom equipment 78 00:04:20,960 --> 00:04:24,480 Speaker 4: makers that sunk so much money into which they accumulated 79 00:04:24,560 --> 00:04:25,160 Speaker 4: huge losses. 80 00:04:25,760 --> 00:04:28,799 Speaker 2: So how are the big AI players thinking about these 81 00:04:28,920 --> 00:04:30,080 Speaker 2: casts at the moment? 82 00:04:30,720 --> 00:04:35,240 Speaker 4: So pushback to the depreciation argument would come from in 83 00:04:35,400 --> 00:04:38,120 Speaker 4: video and we've heard that recently from the CEO Jensen Huang, 84 00:04:38,839 --> 00:04:41,800 Speaker 4: and he's made the case that in fact, even their 85 00:04:41,839 --> 00:04:45,400 Speaker 4: older AI chips, one of their older versions is called Hopper, 86 00:04:46,040 --> 00:04:50,000 Speaker 4: has a lifespan of about six years and is very versatile, 87 00:04:50,040 --> 00:04:52,520 Speaker 4: so you can use it not just for the training 88 00:04:53,040 --> 00:04:55,919 Speaker 4: of these large language models, but for the post training 89 00:04:55,960 --> 00:04:58,680 Speaker 4: and for the inference. That's when they're actually being used 90 00:04:58,680 --> 00:05:01,800 Speaker 4: by us, by consumer and by enterprise, and so you 91 00:05:01,839 --> 00:05:05,120 Speaker 4: can move them around. They have different functions and therefore 92 00:05:05,160 --> 00:05:07,800 Speaker 4: they actually have a longer lifespan than some of the 93 00:05:07,800 --> 00:05:11,440 Speaker 4: skeptics are suggesting, and our own analysis suggest that those 94 00:05:11,440 --> 00:05:13,520 Speaker 4: Hopper chips, those older varieties of chips have a life 95 00:05:13,520 --> 00:05:15,960 Speaker 4: span of about six years and are fully utilized by 96 00:05:16,360 --> 00:05:19,240 Speaker 4: most of the companies that own those. So that does 97 00:05:19,279 --> 00:05:25,240 Speaker 4: address some of that concern. The question going forward to 98 00:05:25,279 --> 00:05:27,800 Speaker 4: what extent these companies are going to be able to 99 00:05:27,839 --> 00:05:33,120 Speaker 4: find products that match the investments that they are syncing 100 00:05:33,200 --> 00:05:36,320 Speaker 4: into the AI infrastructure story. A Bane Capital came out 101 00:05:36,320 --> 00:05:40,040 Speaker 4: with a report recently suggesting that by twenty thirty, the 102 00:05:40,160 --> 00:05:43,560 Speaker 4: hyper scalers and other AI giants would have to be 103 00:05:43,600 --> 00:05:47,240 Speaker 4: turning around revenues of about two trillion dollars and that 104 00:05:47,400 --> 00:05:50,080 Speaker 4: right now there's a huge gap, hundreds of billions of 105 00:05:50,080 --> 00:05:53,240 Speaker 4: dollars in terms of the gap between the investments into 106 00:05:53,279 --> 00:05:57,200 Speaker 4: the AI infrastructure and the actual revenues that are coming 107 00:05:57,240 --> 00:06:01,640 Speaker 4: about as customers and as enterprises and companies use the 108 00:06:01,800 --> 00:06:03,720 Speaker 4: end product. So the go to market, the product fit 109 00:06:03,839 --> 00:06:05,599 Speaker 4: is going to be really, really important. And what the 110 00:06:05,600 --> 00:06:08,839 Speaker 4: big AI players say whether or that is the hyperscalers again, 111 00:06:08,880 --> 00:06:11,479 Speaker 4: the likes of Meta and in Alphabet and Amazon say, 112 00:06:11,800 --> 00:06:13,840 Speaker 4: all the likes of open A and Anthropic because we're 113 00:06:13,839 --> 00:06:15,960 Speaker 4: going to be in this world of agentic AI. We're 114 00:06:15,960 --> 00:06:19,320 Speaker 4: going to have AI agents booking our holidays, checking up 115 00:06:19,360 --> 00:06:23,160 Speaker 4: on our healthcare, finding good schools and universities for our students. 116 00:06:23,320 --> 00:06:25,000 Speaker 4: All those kind of things are going to come together. 117 00:06:25,400 --> 00:06:28,520 Speaker 4: Enterprises are going to be embedding AI much more than 118 00:06:28,520 --> 00:06:31,240 Speaker 4: they already are. We're only in the first opening stages 119 00:06:31,279 --> 00:06:33,000 Speaker 4: of that would be the argument. And then there's the 120 00:06:33,120 --> 00:06:36,200 Speaker 4: sovereign AI story where different countries, and we're seeing that 121 00:06:36,240 --> 00:06:38,000 Speaker 4: in the Middle East but also in Europe as well 122 00:06:38,120 --> 00:06:40,520 Speaker 4: and Japan are investing heavily to ensure that they have 123 00:06:40,600 --> 00:06:43,920 Speaker 4: their own AI infrastructure AI clouds. That will very early 124 00:06:44,000 --> 00:06:46,239 Speaker 4: in that story as well. Those are all the cases 125 00:06:46,240 --> 00:06:49,760 Speaker 4: that the big AI players would underscore in terms of 126 00:06:49,920 --> 00:06:53,200 Speaker 4: why this is going to be driving momentum going forward 127 00:06:53,279 --> 00:06:56,359 Speaker 4: at least through twenty twenty six. Our own team at 128 00:06:56,360 --> 00:06:58,800 Speaker 4: Bloomberg Intelligence say the end of twenty twenty six is 129 00:06:58,800 --> 00:07:00,440 Speaker 4: going to be a question mark to whether or not 130 00:07:00,680 --> 00:07:04,400 Speaker 4: investors continue to have patients. Will they continue to invest 131 00:07:04,440 --> 00:07:08,160 Speaker 4: in the hyperscalers if they're not saying real material terms, 132 00:07:08,160 --> 00:07:11,080 Speaker 4: if that product fit and that custom use isn't there 133 00:07:11,160 --> 00:07:13,560 Speaker 4: in a really really significant way. So I think the 134 00:07:13,640 --> 00:07:17,040 Speaker 4: patients of investors and to what extent they can continue 135 00:07:17,520 --> 00:07:20,160 Speaker 4: to lean into the Hyperscalers if they spend these huge 136 00:07:20,160 --> 00:07:22,160 Speaker 4: amounts is going to be a key question mark, and 137 00:07:22,160 --> 00:07:23,880 Speaker 4: our own team think that that's really going to come 138 00:07:23,920 --> 00:07:25,880 Speaker 4: to the fore at the end of twenty twenty six. 139 00:07:25,880 --> 00:07:28,440 Speaker 4: They'll need to answer that question. They've they've spent. Hyper 140 00:07:28,440 --> 00:07:31,520 Speaker 4: Scalers have spent about three hundred billion dollars on air 141 00:07:31,640 --> 00:07:34,360 Speaker 4: infrastructure this year, and the projection is that they could 142 00:07:34,400 --> 00:07:37,000 Speaker 4: be according to Vidia in the video, sees the Hyperscalar 143 00:07:37,040 --> 00:07:40,040 Speaker 4: spending upwards of about six hundred billion dollars next year. 144 00:07:40,480 --> 00:07:41,920 Speaker 2: One of the things that occurs to me in this 145 00:07:42,000 --> 00:07:44,480 Speaker 2: as well. As we're talking about some of the world's 146 00:07:44,640 --> 00:07:48,920 Speaker 2: most valuable companies. They have massive cash piles in a 147 00:07:48,960 --> 00:07:52,360 Speaker 2: lot of cases, Why is their concern at all about 148 00:07:52,360 --> 00:07:54,760 Speaker 2: how they're going to pay for this given their revenue 149 00:07:54,760 --> 00:07:56,240 Speaker 2: streams and how much money they have. 150 00:07:56,640 --> 00:07:59,440 Speaker 4: You're absolutely right. So when we talk about the hyperscalers, 151 00:08:00,000 --> 00:08:04,560 Speaker 4: these are companies with massive balance sheets and huge cash reserves. 152 00:08:04,560 --> 00:08:08,120 Speaker 4: These are incredibly profitable with businesses that come through with 153 00:08:08,320 --> 00:08:13,040 Speaker 4: very strong earnings. These are not nonprofitable major punts and 154 00:08:13,160 --> 00:08:15,520 Speaker 4: risky parts of the market. These are not companies that 155 00:08:15,560 --> 00:08:18,560 Speaker 4: no one's heard of. They're making real product, they're selling 156 00:08:18,560 --> 00:08:22,800 Speaker 4: it to customers, and they've been doing that for decades. Microsoft, Alphabet, Meta, 157 00:08:22,840 --> 00:08:25,440 Speaker 4: and Amazon. They have that balance sheet strength, they have 158 00:08:25,560 --> 00:08:30,280 Speaker 4: that cash on hand. The concern then is around other 159 00:08:30,320 --> 00:08:32,280 Speaker 4: parts of this ecosystem. So if you can think about 160 00:08:32,320 --> 00:08:35,360 Speaker 4: it in different baskets, you have those big ticket blue 161 00:08:35,440 --> 00:08:38,840 Speaker 4: chip names in one basket, and then you have maybe 162 00:08:38,880 --> 00:08:41,640 Speaker 4: neo clouds in the other basket. These are the core 163 00:08:41,679 --> 00:08:45,920 Speaker 4: weaves or the en clouds companies that lease out data 164 00:08:45,960 --> 00:08:48,240 Speaker 4: centers to some of these hyper scalers, and some of 165 00:08:48,240 --> 00:08:51,199 Speaker 4: the large language models who have business models that are 166 00:08:51,280 --> 00:08:55,079 Speaker 4: less proven than the hyperscalers. Then another bucket would be 167 00:08:55,920 --> 00:08:58,280 Speaker 4: maybe some of the key large language models themselves, the 168 00:08:58,320 --> 00:09:01,040 Speaker 4: open eyes and the anthropics that are money on an 169 00:09:01,040 --> 00:09:03,720 Speaker 4: annual basis. Even as they're seeing a lot of growth 170 00:09:04,040 --> 00:09:08,600 Speaker 4: and revenues increase year on year, they're still not profitable. 171 00:09:08,840 --> 00:09:10,920 Speaker 4: So you can break it down into different categories in 172 00:09:11,000 --> 00:09:14,160 Speaker 4: terms of the level of risk. But even amongst the 173 00:09:14,160 --> 00:09:16,880 Speaker 4: big publicly listed companies with those strong balance sheets, you 174 00:09:16,960 --> 00:09:19,840 Speaker 4: have seen examples of then tapping the public markets and 175 00:09:19,960 --> 00:09:22,160 Speaker 4: raising debt on the public markets, and so far that's 176 00:09:22,200 --> 00:09:25,320 Speaker 4: been well received by the markets. But how long is 177 00:09:25,360 --> 00:09:27,440 Speaker 4: that going to continue? And to what extent is the 178 00:09:27,559 --> 00:09:31,120 Speaker 4: leverage that now these companies are starting to tap into 179 00:09:31,640 --> 00:09:34,600 Speaker 4: going to be acceptable to investors. And again I think 180 00:09:34,600 --> 00:09:36,880 Speaker 4: you have to put a different framework over the different 181 00:09:36,880 --> 00:09:39,960 Speaker 4: companies in terms of how you answer that question. Then 182 00:09:39,960 --> 00:09:42,920 Speaker 4: there's the circularity of the financing. So open Ai, for example, 183 00:09:42,960 --> 00:09:46,200 Speaker 4: doing deals with Nvidia and Nvidia investing in open Ai, 184 00:09:46,360 --> 00:09:49,280 Speaker 4: and in response to that, open Ai committing to buying 185 00:09:49,320 --> 00:09:53,000 Speaker 4: a certain number of chips from Nvidia. Those circular financing deals, 186 00:09:53,040 --> 00:09:55,560 Speaker 4: as they've been described by some have also caused some 187 00:09:55,640 --> 00:09:59,840 Speaker 4: concern as all of these companies becoming increasingly enmeshed and 188 00:10:00,040 --> 00:10:03,400 Speaker 4: intertwined in terms of their deals and their investments on 189 00:10:03,440 --> 00:10:05,280 Speaker 4: what is a bet on the future and how the 190 00:10:05,320 --> 00:10:06,400 Speaker 4: future evolves, and. 191 00:10:06,480 --> 00:10:08,880 Speaker 2: An expensive one at that. Tom McKenzie, thank you very 192 00:10:08,960 --> 00:10:11,920 Speaker 2: much for joining us, host of Bloomberg Tech Europe on 193 00:10:11,960 --> 00:10:15,800 Speaker 2: Bloomberg Television. For more explanations like this from our team 194 00:10:15,840 --> 00:10:18,679 Speaker 2: of three thousand journalists and analysts around the world, go 195 00:10:18,679 --> 00:10:22,240 Speaker 2: to Bloomberg dot com slash explainers. I'm Stephen Carroll. This 196 00:10:22,320 --> 00:10:24,480 Speaker 2: is Here's why. I'll be back next week with more. 197 00:10:24,720 --> 00:10:25,480 Speaker 2: Thanks for listening.