1 00:00:02,800 --> 00:00:07,760 Speaker 1: Zone, media, superstition, fear, and jealousy. Welcome to this week's 2 00:00:07,760 --> 00:00:22,599 Speaker 1: Better Offline monologue. I'm your host ed ze Tron. As ever, 3 00:00:22,640 --> 00:00:25,120 Speaker 1: please subscribe to the newsletter. We're running a one year 4 00:00:25,120 --> 00:00:27,240 Speaker 1: anniversary deal right now. The link will be in the 5 00:00:27,280 --> 00:00:29,760 Speaker 1: episode notes, and we're giving you ten bucks off the 6 00:00:29,800 --> 00:00:32,760 Speaker 1: annual rate for life and even if you die and 7 00:00:32,760 --> 00:00:34,519 Speaker 1: come back to life, I'll give it to you anyway. 8 00:00:34,760 --> 00:00:37,040 Speaker 1: The deal runs through midnight on Sunday. But if you 9 00:00:37,040 --> 00:00:39,120 Speaker 1: happen to hear this after this point, email me at 10 00:00:39,120 --> 00:00:41,880 Speaker 1: easy at Better offline dot com or ez at better 11 00:00:41,920 --> 00:00:44,479 Speaker 1: Offline dot com for the British and Canadians in the audience, 12 00:00:44,640 --> 00:00:46,800 Speaker 1: and I'll sort you out. I'm generous like that, and 13 00:00:46,880 --> 00:00:48,920 Speaker 1: I love my premium mates. Really some of the best 14 00:00:48,960 --> 00:00:52,080 Speaker 1: analysts work out there. It's very deep. I got the 15 00:00:52,159 --> 00:00:54,160 Speaker 1: terminal going now and pulling all sorts of data. The 16 00:00:54,240 --> 00:00:57,320 Speaker 1: many numbers I read every week are driving me truly insane. 17 00:00:57,960 --> 00:01:01,000 Speaker 1: And with that in mind, to day's monologue starts with 18 00:01:01,000 --> 00:01:05,520 Speaker 1: a simple question, what the fuck are hyperscaler is doing? Now? 19 00:01:05,560 --> 00:01:08,160 Speaker 1: I realize I've asked this question maybe one hundred thousand 20 00:01:08,200 --> 00:01:10,200 Speaker 1: times in the last two years, but it's begun to 21 00:01:10,240 --> 00:01:15,280 Speaker 1: get so incredibly stupid that I'm actually a little exasperated. McKinsey. 22 00:01:15,319 --> 00:01:19,280 Speaker 1: Spawn and Google CEO Sundar Pashai announced in Wednesday's earnings 23 00:01:19,280 --> 00:01:22,200 Speaker 1: that Google would raise its year end capital expenditures to 24 00:01:22,240 --> 00:01:24,640 Speaker 1: somewhere between one hundred and ninety five billion dollars and 25 00:01:24,680 --> 00:01:27,880 Speaker 1: two hundred and five billion dollars after burning eighty point 26 00:01:28,000 --> 00:01:30,000 Speaker 1: six billion dollars in the first half of the year, 27 00:01:30,319 --> 00:01:32,720 Speaker 1: and this suggests that Google will spend another one hundred 28 00:01:32,760 --> 00:01:35,000 Speaker 1: and fourteen billion dollars by the end of twenty twenty six. 29 00:01:35,040 --> 00:01:37,200 Speaker 1: They're going to be doing what fifty five sixty billion 30 00:01:37,240 --> 00:01:41,199 Speaker 1: dollars a quarter. It's fucking insane, it's fucking stupid. I'm 31 00:01:41,240 --> 00:01:46,040 Speaker 1: sorry for swearing, but it's absolutely deranged. What are you doing, Sunda. 32 00:01:46,160 --> 00:01:48,720 Speaker 1: You've already spent two hundred and eighty eight billion dollars 33 00:01:48,720 --> 00:01:51,040 Speaker 1: in capex since twenty twenty two, and all you've got 34 00:01:51,040 --> 00:01:53,760 Speaker 1: to show for it is several also RAN llms and 35 00:01:53,840 --> 00:01:58,080 Speaker 1: one indeterminately large customer in Anthropic, who you've also invested 36 00:01:58,120 --> 00:02:02,600 Speaker 1: thirteen billion dollars in. Now. One answer might be that 37 00:02:02,840 --> 00:02:06,240 Speaker 1: Anthropic helped choose Google cloud's growth by eighty two percent 38 00:02:06,320 --> 00:02:08,720 Speaker 1: year of a year in this qure But considering that 39 00:02:08,760 --> 00:02:10,840 Speaker 1: Google doesn't actually break out how much of that is 40 00:02:10,840 --> 00:02:14,040 Speaker 1: from AI, or from Anthropic, or really from anything, it's 41 00:02:14,080 --> 00:02:16,280 Speaker 1: difficult to say what exactly this money is doing, other 42 00:02:16,320 --> 00:02:18,480 Speaker 1: than turning its free cash flow negative for the first 43 00:02:18,560 --> 00:02:21,840 Speaker 1: time since Google went public. While I can't say for 44 00:02:21,880 --> 00:02:24,440 Speaker 1: sure how much Anthropic is spending, I do believe it's 45 00:02:24,480 --> 00:02:27,480 Speaker 1: taking up the vast majority of Google's AI infrastructure and 46 00:02:27,560 --> 00:02:31,680 Speaker 1: representing most of its cloud growth. Google Cloud was, up 47 00:02:31,760 --> 00:02:34,119 Speaker 1: until fairly recently the last year or so, are much 48 00:02:34,120 --> 00:02:37,680 Speaker 1: smaller business than Microsoft. As euroamason web services have been 49 00:02:37,720 --> 00:02:40,639 Speaker 1: growing somewhere between twenty two percent and thirty percent year 50 00:02:40,680 --> 00:02:43,240 Speaker 1: of a year, hitting thirty three point nine percent year 51 00:02:43,280 --> 00:02:44,839 Speaker 1: of a year growth in the third quarter of twenty 52 00:02:44,919 --> 00:02:49,840 Speaker 1: twenty five. That very same quarter, Google and Anthropic announced 53 00:02:49,840 --> 00:02:52,120 Speaker 1: the deal to expand its use of TPUs, which is 54 00:02:52,160 --> 00:02:55,280 Speaker 1: Google's special GPUs, in a contract worth than I quote 55 00:02:55,360 --> 00:02:59,680 Speaker 1: tens of billions of dollars. Mysteriously can't imagine how Google's 56 00:02:59,720 --> 00:03:01,880 Speaker 1: next time three quarters of cloud revenue growth showed forty 57 00:03:01,919 --> 00:03:04,560 Speaker 1: seven point seven percent sixty three point four percent and 58 00:03:04,600 --> 00:03:06,840 Speaker 1: eighty two point one percent year of a year growth. 59 00:03:07,080 --> 00:03:10,240 Speaker 1: Whatever could that be? Where did that come from? I 60 00:03:10,320 --> 00:03:13,440 Speaker 1: can't work it out, can you, dear listener? What could 61 00:03:13,480 --> 00:03:18,120 Speaker 1: it possively? It's anthropic, It's it's obviously anthropic. Except in 62 00:03:18,200 --> 00:03:22,680 Speaker 1: those same quarters, Google's capital expenditures went up ninety five percent, 63 00:03:22,919 --> 00:03:25,079 Speaker 1: one hundred and seven percent, and one hundred percent year 64 00:03:25,120 --> 00:03:28,240 Speaker 1: of a year, absorbing every dollar of cloud revenue and 65 00:03:28,280 --> 00:03:30,720 Speaker 1: then some. In fact, outside of the fourth quarter of 66 00:03:30,720 --> 00:03:34,560 Speaker 1: twenty twenty four, Google's capital expenditures have outpaced cloud growth 67 00:03:34,600 --> 00:03:36,520 Speaker 1: by at least twelve percent year of a year for 68 00:03:36,560 --> 00:03:40,680 Speaker 1: two goddamn years. And if your pushback is, well, Google 69 00:03:40,760 --> 00:03:44,280 Speaker 1: spending that CAPEX on something else, No, they're not. They're not. 70 00:03:44,680 --> 00:03:46,320 Speaker 1: Maybe they've got a couple of billy coming in from 71 00:03:46,400 --> 00:03:51,200 Speaker 1: Google Workspace and cucking their various subscribers with annoying things. 72 00:03:52,000 --> 00:03:55,680 Speaker 1: It's coming from anthropic. Google is not making that money 73 00:03:55,680 --> 00:03:58,560 Speaker 1: from Gemini. Your ass must be crazy if you think 74 00:03:58,640 --> 00:04:01,560 Speaker 1: that's that's the point. But all of this begs the question, 75 00:04:02,160 --> 00:04:06,480 Speaker 1: when exactly does all of this payoff? How does it payoff? 76 00:04:06,800 --> 00:04:08,960 Speaker 1: What does the payoff even look like? And why am 77 00:04:08,960 --> 00:04:13,360 Speaker 1: I one of the few people asking these questions, hyperscalas 78 00:04:13,360 --> 00:04:15,960 Speaker 1: have now spent I asked them at about one point 79 00:04:15,960 --> 00:04:20,240 Speaker 1: one trillion dollars. I haven't seen Amazon and Microsoft's quarters yet, 80 00:04:20,279 --> 00:04:21,719 Speaker 1: and I'm going to guess they spend a good amount 81 00:04:21,720 --> 00:04:24,560 Speaker 1: on capex. They now spent one point one trillion bloody 82 00:04:24,600 --> 00:04:27,120 Speaker 1: dollars on capital expenditures. And when you ask them to 83 00:04:27,160 --> 00:04:29,880 Speaker 1: tell you how much revenue they made from said expenditures, 84 00:04:29,920 --> 00:04:32,039 Speaker 1: they look at you like that meme of Brendan Fraser 85 00:04:32,040 --> 00:04:35,120 Speaker 1: from the whale. Anyway, here's a fun fact for you. 86 00:04:35,640 --> 00:04:38,720 Speaker 1: Thanks to my own reporting and open ais auditive financials, 87 00:04:39,000 --> 00:04:42,040 Speaker 1: we know that It's spent seventeen point two billion dollars 88 00:04:42,080 --> 00:04:44,640 Speaker 1: on Microsoft a zure in colendar year twenty twenty five, 89 00:04:44,960 --> 00:04:47,960 Speaker 1: a year when Microsoft's intelligent cloud segment made one hundred 90 00:04:47,960 --> 00:04:50,640 Speaker 1: and twenty point four billion dollars in revenue. The segment 91 00:04:50,680 --> 00:04:54,360 Speaker 1: grew as a result twenty six percent year of a year. Yeah, 92 00:04:54,400 --> 00:04:56,719 Speaker 1: I'm a curious little Chris or around the numbers, and 93 00:04:56,839 --> 00:05:00,640 Speaker 1: open Ay's revenue represented fourteen point three percent of all 94 00:05:00,839 --> 00:05:04,240 Speaker 1: Microsoft's cloud revenues for that period. And to be clear, 95 00:05:04,440 --> 00:05:07,080 Speaker 1: it's calendar year twenty twenty five, they have a fiscal 96 00:05:07,120 --> 00:05:09,520 Speaker 1: core thing. But I went and worked it out. But 97 00:05:09,600 --> 00:05:13,359 Speaker 1: this means the open Aiy's money represented eighteen percent of 98 00:05:13,400 --> 00:05:16,520 Speaker 1: the year of year growth of Microsoft's cloud. Without it, 99 00:05:17,040 --> 00:05:19,479 Speaker 1: their cloud revenue growth was a mere eight percent year 100 00:05:19,520 --> 00:05:22,680 Speaker 1: of a year. That's pathetic. That's a bonable. That's barely 101 00:05:22,760 --> 00:05:26,320 Speaker 1: keeping up with inflation. That's dog shit. Satya, Satchy, you 102 00:05:26,320 --> 00:05:29,080 Speaker 1: should be ashamed of yourself. Sacha, come on my show 103 00:05:29,200 --> 00:05:32,400 Speaker 1: so I can ask you very reasonable questions. I'll be nice. 104 00:05:32,480 --> 00:05:35,640 Speaker 1: I swear you won't. You're scared, just let the rest 105 00:05:35,680 --> 00:05:39,520 Speaker 1: of them, just like Clammy Sam on one Clamuel, your 106 00:05:39,520 --> 00:05:42,159 Speaker 1: public relations people said you're too scared to speak to me. Well, 107 00:05:42,200 --> 00:05:43,599 Speaker 1: they said you wouldn't speak to me, and I know 108 00:05:43,640 --> 00:05:46,599 Speaker 1: you're scared. Same with weird Daria. Where are you boys? 109 00:05:46,640 --> 00:05:49,240 Speaker 1: You're too scared of better offline? Anyway, Enough of the 110 00:05:49,279 --> 00:05:53,320 Speaker 1: Trump voice, though, there is actually a reason that hyperscalers 111 00:05:53,320 --> 00:05:57,000 Speaker 1: are still spending all that capex, open AI and Anthropic 112 00:05:57,040 --> 00:06:00,000 Speaker 1: allow hyperscalers to boost growth by feeding their own money 113 00:06:00,120 --> 00:06:03,400 Speaker 1: back into their businesses, a short term solution to the 114 00:06:03,480 --> 00:06:06,520 Speaker 1: long term problem of the end of the era of hypergrowth. 115 00:06:06,760 --> 00:06:09,400 Speaker 1: One of the reasons they're so shy about AI revenues 116 00:06:09,520 --> 00:06:12,279 Speaker 1: is that they know it'll become obvious where those revenues 117 00:06:12,320 --> 00:06:16,680 Speaker 1: are coming from two unprofitable and unsustainable AI labs and 118 00:06:16,720 --> 00:06:20,920 Speaker 1: a little trickle, little spurt from AI services they dump 119 00:06:21,000 --> 00:06:24,279 Speaker 1: on their enterprise customers through confusing subscriptions up cells and 120 00:06:24,400 --> 00:06:28,839 Speaker 1: just adding AI to random things and being like, look, folks, 121 00:06:29,120 --> 00:06:31,479 Speaker 1: don't we love this? Don't you all love this? You 122 00:06:31,560 --> 00:06:36,520 Speaker 1: little shits? And hyperscalers also know that the second they 123 00:06:36,560 --> 00:06:41,200 Speaker 1: cut their capex spend, their AI trade unwinds. If Microsoft, Google, 124 00:06:41,320 --> 00:06:44,320 Speaker 1: Amazon and Meta aren't buying hundreds of billions of dollars 125 00:06:44,320 --> 00:06:47,880 Speaker 1: of GPUs a year, that means Taiwanese server manufacturers like 126 00:06:47,960 --> 00:06:50,640 Speaker 1: Quanta and on High won't sell as many servers and 127 00:06:51,000 --> 00:06:53,919 Speaker 1: thus won't need as many GPUs broad common in video 128 00:06:54,000 --> 00:06:57,039 Speaker 1: won't need as much high bandwid memory for those TPUs 129 00:06:57,080 --> 00:06:59,960 Speaker 1: and GPUs, and they are sent speculators across the world 130 00:07:00,160 --> 00:07:03,600 Speaker 1: will suddenly realize that they conflated hyperscalers being stupid with 131 00:07:03,640 --> 00:07:06,360 Speaker 1: their money with the existence of massive amounts of demand 132 00:07:06,400 --> 00:07:09,960 Speaker 1: for AI compute. But another way, hyperscalers are doing k 133 00:07:10,160 --> 00:07:12,560 Speaker 1: fape to keep the crowd riled up, but everybody slowly 134 00:07:12,560 --> 00:07:14,680 Speaker 1: working out that the wrestlers aren't really hitting each other 135 00:07:14,720 --> 00:07:18,360 Speaker 1: and everything is scripted. I also should add I can 136 00:07:18,440 --> 00:07:20,600 Speaker 1: just anticipate I'm going to get an email from someone 137 00:07:20,640 --> 00:07:23,400 Speaker 1: being like, well, actually, this is good. The cattle expenditures 138 00:07:23,400 --> 00:07:26,440 Speaker 1: are good because it showed growth, even though it's short term. 139 00:07:26,600 --> 00:07:30,400 Speaker 1: That's a silly, silly perspective to take, considering they're feeding 140 00:07:30,440 --> 00:07:32,720 Speaker 1: all the revenues back into the business in the capex 141 00:07:32,720 --> 00:07:35,480 Speaker 1: for stuff that won't be useful in the future. Mathematics. 142 00:07:35,480 --> 00:07:41,040 Speaker 1: My friend try it sometime. Anyway. It is a funny 143 00:07:41,040 --> 00:07:42,760 Speaker 1: old time for me right now, because I feel like 144 00:07:42,760 --> 00:07:45,960 Speaker 1: I'm on three or four podcasts or TV shows or 145 00:07:45,960 --> 00:07:48,800 Speaker 1: interviews a week, as the system appears to be trying 146 00:07:48,840 --> 00:07:51,200 Speaker 1: to cover its ass by having me on to explain 147 00:07:51,240 --> 00:07:53,800 Speaker 1: stuff I said two years ago, but a little more 148 00:07:53,840 --> 00:07:57,920 Speaker 1: insistently and clearly than before. While the mainstream media seems 149 00:07:57,960 --> 00:08:00,360 Speaker 1: hesitant to fully commit to the graveyard small of the 150 00:08:00,360 --> 00:08:03,160 Speaker 1: AI bubble. It feels as if there's a conscious attempt 151 00:08:03,200 --> 00:08:05,840 Speaker 1: to represent my work as more than just the other side, 152 00:08:05,960 --> 00:08:08,440 Speaker 1: or a hater or a skeptic or a cynic, which 153 00:08:08,480 --> 00:08:12,080 Speaker 1: is both edifying and enraging because we are now so 154 00:08:12,240 --> 00:08:17,120 Speaker 1: far past past the point of no Returnemy. Yeah, we 155 00:08:17,240 --> 00:08:20,560 Speaker 1: really can't avoid what's coming. As I wrote in my 156 00:08:20,600 --> 00:08:23,120 Speaker 1: newsletter this week, I think there's a subprime data center 157 00:08:23,200 --> 00:08:26,160 Speaker 1: crisis brewing right now, with over five hundred billion dollars 158 00:08:26,160 --> 00:08:28,920 Speaker 1: in data center SPVs that can only ever make investors 159 00:08:28,920 --> 00:08:31,360 Speaker 1: whole if there's hundreds of billions of dollars of AI 160 00:08:31,720 --> 00:08:34,280 Speaker 1: compute demand in the next two years. And to be clear, 161 00:08:34,320 --> 00:08:36,520 Speaker 1: when I say hundreds of billions of dollars demand, I 162 00:08:36,600 --> 00:08:39,480 Speaker 1: mean a year. I mean literally a year. I ran 163 00:08:39,520 --> 00:08:42,000 Speaker 1: the numbers, and for the what I think it's one 164 00:08:42,040 --> 00:08:44,839 Speaker 1: hundred ninety giga wats of supposedly planned data centers, you'd 165 00:08:44,880 --> 00:08:47,240 Speaker 1: need one point six trillion dollars a year in annual 166 00:08:47,280 --> 00:08:51,200 Speaker 1: compute revenue to make it worth worthwhile. And by the way, 167 00:08:51,200 --> 00:08:54,360 Speaker 1: the global software market is under eight hundred billion dollars 168 00:08:54,400 --> 00:08:56,480 Speaker 1: a year, So I don't know where that fucking money's 169 00:08:56,520 --> 00:08:59,040 Speaker 1: coming from, and no one else seems to really want 170 00:08:59,040 --> 00:09:01,480 Speaker 1: to think about him. And I deeply worry about what 171 00:09:01,559 --> 00:09:04,080 Speaker 1: happens when the demand doesn't arrive. As these data center 172 00:09:04,120 --> 00:09:06,679 Speaker 1: debt deals are often funded by private credit funds, which 173 00:09:06,760 --> 00:09:10,240 Speaker 1: are in turn funded by both public and private pensions 174 00:09:10,240 --> 00:09:14,360 Speaker 1: and insurance funds. Anyway, do read the free newsletter to 175 00:09:14,440 --> 00:09:16,800 Speaker 1: learn more. It is free this one. I have a 176 00:09:16,800 --> 00:09:20,440 Speaker 1: premium one coming out today that will be about well, 177 00:09:20,520 --> 00:09:22,800 Speaker 1: the fact that Oracle's dying, which I mean, hey, you 178 00:09:22,840 --> 00:09:26,080 Speaker 1: in the back, stop laughing. Stop it, poor Larry Ellison 179 00:09:26,280 --> 00:09:30,079 Speaker 1: Paul who gives a fuck in all seriousness. Though, it's 180 00:09:30,120 --> 00:09:32,040 Speaker 1: going to be a banger. The Free ones a banger. 181 00:09:32,080 --> 00:09:34,760 Speaker 1: There's bangers everywhere. And I'll be back next week with 182 00:09:34,800 --> 00:09:37,120 Speaker 1: my good friend Nick Saresh to talk about how AI 183 00:09:37,280 --> 00:09:40,520 Speaker 1: mania is eviscerating global decision making, creating a cult lug 184 00:09:40,640 --> 00:09:44,400 Speaker 1: environment where once participation in AI is almost a religious act, 185 00:09:44,480 --> 00:09:47,640 Speaker 1: and if you talk badly about AI, it's considered a 186 00:09:47,679 --> 00:09:50,440 Speaker 1: cardinal sin or a cardinal sin, one of those two. 187 00:09:50,640 --> 00:09:53,880 Speaker 1: Sure one of you will correct me on this one. Nevertheless, 188 00:09:54,240 --> 00:09:56,600 Speaker 1: it's a really he it's based on this incredible piece 189 00:09:56,640 --> 00:09:58,600 Speaker 1: he wrote. I can't wait for you to read it. 190 00:09:58,640 --> 00:10:00,960 Speaker 1: I can't wait for you Timmy talk to him. Oh no, 191 00:10:01,160 --> 00:10:04,000 Speaker 1: can't wait for you to consume my content. What a 192 00:10:04,000 --> 00:10:07,160 Speaker 1: peculiar thing to say. Anyway, I'll got you Next week 193 00:10:07,559 --> 00:10:08,240 Speaker 1: Zitron out