WEBVTT - Monologue: We Live Inside A Dream, But Who Is The Dreamer? 0:00:02.800 --> 0:00:07.880 Zo Media. Hello and welcome to this week's Better Offline Monologue. 0:00:07.920 --> 0:00:22.520 I'm your host ed Zetron. It's been a weird week. 0:00:22.600 --> 0:00:25.479 I'm recording this on Thursday to thirtieth of July. So 0:00:25.560 --> 0:00:28.120 we've only had Microsoft and Meta report earning so far. 0:00:29.160 --> 0:00:31.480 So let's start with Meta, who saw it stock get 0:00:31.480 --> 0:00:34.000 hammered to day after missing on earnings for share narrowing 0:00:34.080 --> 0:00:36.880 by which I mean increasing by another word, it's cattle 0:00:36.920 --> 0:00:39.400 expenditures guidance for the year to Arrange of about one 0:00:39.479 --> 0:00:41.480 hundred and thirty five billion to one hundred and forty 0:00:41.520 --> 0:00:45.040 five billion dollars and guiding as in projecting sixty one 0:00:45.120 --> 0:00:47.720 billion dollars to sixty four billion dollars of revenue in 0:00:47.800 --> 0:00:51.520 Q three, which was below estimates. Cash Flow also deteriorated 0:00:51.560 --> 0:00:54.640 to a little under eight hundred million dollars, and I 0:00:54.720 --> 0:00:57.279 must be clear how strange it is watching these companies 0:00:57.320 --> 0:00:59.560 do this. Meta used to just print money. They used 0:00:59.560 --> 0:01:02.960 to have ton free cash flow. But this dedication to 0:01:03.000 --> 0:01:06.560 AI is questionable and wasteful, and the market is pissed, 0:01:06.720 --> 0:01:09.039 mostly because Meta is yet to show any proof that's 0:01:09.040 --> 0:01:12.120 spending over two hundred billion dollars in Capex did anything 0:01:12.120 --> 0:01:15.959 for the company. Conversely, Microsoft's stock is ripping because it 0:01:16.000 --> 0:01:18.800 beat on analyst estimates across the board, with its as 0:01:18.880 --> 0:01:21.319 Your Cloud platform hitting one hundred billion dollars in revenue 0:01:21.360 --> 0:01:23.720 for the fiscal year that just ended, and to be clear, 0:01:23.800 --> 0:01:27.280 Microsoft's fiscal year twenty twenty six started on July first, 0:01:27.280 --> 0:01:31.039 twenty twenty five. Annoying, so annoying, I hate it, but 0:01:31.480 --> 0:01:35.160 that's the ballgame. What's funny is that neither company actually 0:01:35.240 --> 0:01:38.360 proved that AI is paying off outside of brainless analysts 0:01:38.360 --> 0:01:40.720 and a captured business and tech media reading between the 0:01:40.760 --> 0:01:44.360 lines and saying, ooh, AI's paying off big time. Last quarter, 0:01:44.440 --> 0:01:47.400 Microsoft disclosed its AI run rate month times twelve, which 0:01:47.440 --> 0:01:50.440 was thirty seven billion dollars about three billion dollars a month, 0:01:50.520 --> 0:01:53.040 and then simply chose not to report the figure at 0:01:53.080 --> 0:01:56.080 all in this quarter. I assume because the number was lower, 0:01:56.160 --> 0:01:58.680 but that doesn't matter to anyone covering earnings. Put the 0:01:58.720 --> 0:02:02.240 AI hype in the bag. No thinking only number go up, please, 0:02:02.760 --> 0:02:05.680 no thinking at all. We cannot think here. It's not 0:02:05.720 --> 0:02:08.800 like people are gonna buy Microsoft's stock because of this. 0:02:09.840 --> 0:02:12.360 I'll also add that nobody seems to want to discuss 0:02:12.400 --> 0:02:16.160 that Microsoft Azure's largest customer and likely the vast majority 0:02:16.200 --> 0:02:19.840 of its growth, is open Ai, an unprofitable, unsustainable company 0:02:19.960 --> 0:02:23.320 that accounted for, per my own reporting, sixty nine percent 0:02:23.360 --> 0:02:26.280 of all year over year growth of Microsoft's cloud in 0:02:26.320 --> 0:02:30.959 calendar year twenty twenty five. Microsoft's cloud revenue mysteriously seems 0:02:30.960 --> 0:02:34.160 to grow every quarter with open AI's increasing spent on 0:02:34.200 --> 0:02:37.840 AI compute, a fact mentioned in exactly zero of the 0:02:37.880 --> 0:02:40.320 stories I read about earnings, and god damn did I look. 0:02:41.720 --> 0:02:44.440 The inflation of the AI bubble was only made possible 0:02:44.480 --> 0:02:47.639 by the continued unwillingness of the media to actually explain 0:02:47.840 --> 0:02:51.640 or even try to understand what's going on. Microsoft's growth 0:02:51.720 --> 0:02:54.920 was immediately credited to its AI bets paying off, ignoring 0:02:54.919 --> 0:02:57.919 both open AI's compute spend and the massive Microsoft three 0:02:58.000 --> 0:03:01.240 sixty five price increases that hit onto first, because it's 0:03:01.280 --> 0:03:04.119 convenient and rewarding to coddle the powerful and celebrate their 0:03:04.280 --> 0:03:07.519 supposed victories rather than actually seeking the truth. And what's 0:03:07.520 --> 0:03:09.440 funny is people are going to say, well, open ai 0:03:09.600 --> 0:03:13.120 is their bet paying off? Is it? Is it? When 0:03:13.160 --> 0:03:15.960 it's a venture capital backed company? When Microsoft has had 0:03:15.960 --> 0:03:19.359 to sink over thirteen billion dollars into the company when 0:03:19.480 --> 0:03:21.880 there are reports in twenty twenty five that that compute 0:03:21.919 --> 0:03:24.359 was served at cost. How much is their investments to 0:03:24.440 --> 0:03:28.000 zero credits? Who knows? Nobody bothers to look, no one 0:03:28.080 --> 0:03:31.520 bothers to ask. The analysts on these calls should not 0:03:31.639 --> 0:03:34.720 have jobs. They do not exist to protect investors. They 0:03:34.760 --> 0:03:39.000 exist to perpetuate hype cycles. And what's actually happening to 0:03:39.040 --> 0:03:42.240 this company is pretty simple. As open AI's largest provider 0:03:42.280 --> 0:03:44.680 of AI compute AZ, your revenue growth will continue to 0:03:44.720 --> 0:03:48.280 spike as long as venture capital continues to funnel cash 0:03:48.480 --> 0:03:52.320 to Clammy Sam Altman. Microsoft has not found some magic 0:03:52.360 --> 0:03:55.320 AI money chie it is not unlocked incredible value for 0:03:55.360 --> 0:03:58.440 its Microsoft three sixty five copilot subscribers, and it's not 0:03:58.520 --> 0:04:02.200 got some secret saw. While three sixty five co pilot 0:04:02.240 --> 0:04:05.240 subscriptions increase by ten million in this quarter, keep in 0:04:05.240 --> 0:04:07.960 mind that Microsoft regularly discounts l simpathly adds it for 0:04:08.080 --> 0:04:12.320 free to pump these numbers up. We're far past the 0:04:12.360 --> 0:04:15.480 point in the AI bubble when naked deception is necessary, 0:04:15.480 --> 0:04:17.320 because the media is more than willing to keep filling 0:04:17.320 --> 0:04:19.800 in the blanks at the preference of the company's Keep 0:04:19.839 --> 0:04:21.839 in mind that a mere twenty four hours before Meta's 0:04:21.880 --> 0:04:23.839 horror show earnings that The Wall Street Journal in New 0:04:23.920 --> 0:04:26.880 York Times both ran separate stories about Mark zuckerbog wanting 0:04:26.920 --> 0:04:30.360 superintelligence for all evacuous statement that exists only to get 0:04:30.400 --> 0:04:33.640 an irrelevant company attention and to paper over its meaningless 0:04:33.680 --> 0:04:37.320 and wasteful AI play. What super intelligence, Mark, You're building 0:04:37.360 --> 0:04:41.760 fucking large language models? Who cares? Who fucking cares? None 0:04:41.760 --> 0:04:45.279 of this actually connects to what the models do. I 0:04:45.360 --> 0:04:49.400 feel like I'm going insane, And just just look at 0:04:49.440 --> 0:04:52.479 the coverage around in Nvidia's potential backstop of two hundred 0:04:52.520 --> 0:04:55.719 and fifty billion dollars in compute spend on the still 0:04:55.800 --> 0:04:58.719 theoretical ten giga. What data center in Ohio that a 0:04:58.760 --> 0:05:01.919 soft bank subsidiary want to build for open Ai? Not 0:05:02.040 --> 0:05:04.039 once in any of the stories did I read a 0:05:04.120 --> 0:05:07.400 single thing about how open ai can't afford that spend, 0:05:07.480 --> 0:05:10.359 that debt markets are becoming exhausted with AI loans, that 0:05:10.520 --> 0:05:12.800 sp Energy has never built a data center before, and 0:05:12.839 --> 0:05:15.719 that none of this makes any financial sense. No need 0:05:15.800 --> 0:05:17.839 put the AI hype in the bag. No thinking only 0:05:17.839 --> 0:05:20.520 growth number go up, No need to think at all, 0:05:21.160 --> 0:05:24.520 Just run story. Numbers are big. We love number Again 0:05:24.560 --> 0:05:26.960 and again and again. The media foils investors in the 0:05:26.960 --> 0:05:30.440 general public through either a misplaced attachment to fictional objectivity 0:05:30.680 --> 0:05:33.400 or a desperation to appeal to the powerful. It doesn't 0:05:33.440 --> 0:05:35.359 really matter which one is true, because they both have 0:05:35.440 --> 0:05:37.560 the same effect, creating the appearance that any of this 0:05:37.640 --> 0:05:40.040 is possible in a way that only benefits the rich 0:05:40.080 --> 0:05:57.840 and powerful. And what's incredibly frustrating is it's now, for 0:05:57.880 --> 0:06:00.920 the most part, commonly held consensus we're in a bubble. 0:06:01.440 --> 0:06:03.840 Yet the media continues to write about this stuff as 0:06:03.839 --> 0:06:06.719 if that isn't the case, glazing every single impossible deal 0:06:06.800 --> 0:06:10.720 or blatant active circular financing, salivating over every bizarre new 0:06:10.760 --> 0:06:13.839 model drop or suppose an achievement without ever really explaining 0:06:13.839 --> 0:06:16.719 why it matters, and gooning over every obtuse statement that 0:06:16.800 --> 0:06:20.000 Sam Altman makes. We're not in the singularity, a fucking asshole. 0:06:20.640 --> 0:06:22.880 And this is because we live, on some level in 0:06:22.920 --> 0:06:26.320 an entirely imaginary and manufactured reality, at least when the 0:06:26.360 --> 0:06:30.719 media is concerned. Hyperscalas have spent over a trillion dollars 0:06:30.720 --> 0:06:33.240 in ai Capex and have across the board refuse to 0:06:33.240 --> 0:06:37.159 give us hard numbers about its return on investment. Open 0:06:37.160 --> 0:06:40.800 AI and Anthropic are unprofitable, unsustainable, and we'll need over 0:06:40.839 --> 0:06:43.360 a trillion dollars of funding in the next four years 0:06:43.440 --> 0:06:46.200 because they've made over a trillion dollars of commitments than 0:06:46.200 --> 0:06:48.839 neither of them can afford without a nearest infinite sum 0:06:49.080 --> 0:06:52.560 of venture capital. AI data centers are getting built at 0:06:52.600 --> 0:06:55.040 a snail's pace, and debt to build them is becoming 0:06:55.040 --> 0:06:57.719 more and more expensive. With core Weave hitting goat sea level. 0:06:57.800 --> 0:07:00.520 Credit spreads trading from eight point three percent to nine 0:07:00.520 --> 0:07:03.440 point three percent above US treasuries, which is way higher 0:07:03.480 --> 0:07:06.240 than the average JUG pond, meaning that investors are valuing 0:07:06.279 --> 0:07:08.559 them like the company is a very real chance of death. 0:07:09.760 --> 0:07:12.920 And you'll notice that this isn't really something. You'll notice 0:07:12.960 --> 0:07:16.080 that the spreads are being covered, people thinking, Oh, that's strange. 0:07:16.600 --> 0:07:19.640 Oh that doesn't look good. Ah, wonder what that's about. 0:07:19.880 --> 0:07:22.720 What it's about is that the bond market is screaming 0:07:23.000 --> 0:07:27.200 Core We've might die. Core We've might die. And if 0:07:27.240 --> 0:07:31.480 Core We've dies, every Neocloud dies, Iron dies, Nebuis dies, 0:07:31.560 --> 0:07:35.280 and Scale dies, Lambda dies, all of them die. And 0:07:35.400 --> 0:07:37.960 even if I might be wrong, which I'm not. I 0:07:38.040 --> 0:07:40.480 really am quite confident on this. The fact that this 0:07:40.600 --> 0:07:43.600 is not even a discuss point across the media is 0:07:43.600 --> 0:07:45.280 a sign that no one really gives a fuck about 0:07:45.280 --> 0:07:48.160 any investors, the public, anyone who might get tied up 0:07:48.200 --> 0:07:50.400 in this. It's a sign that no one's thinking about 0:07:50.400 --> 0:07:53.760 this too hard, either through a dramatic ignorance or just 0:07:53.800 --> 0:07:57.120 a lack of given a shit about anything. Yet, the 0:07:57.160 --> 0:07:59.760 irony of all these blaring warning signs is that it's 0:08:00.000 --> 0:08:01.680 possible for most of the media to take them to 0:08:01.720 --> 0:08:04.480 their logical endpoint, as doing so would mean accepting that 0:08:04.520 --> 0:08:07.840 we live in an era of outright insanity. To acknowledge 0:08:07.880 --> 0:08:10.720 open AI's financial condition is to acknowledge that the largest 0:08:10.720 --> 0:08:13.360 customer of Azure might die and that a large chunk 0:08:13.360 --> 0:08:16.360 of Microsoft revenue growth is a fiction created by venture capital. 0:08:16.840 --> 0:08:19.840 To acknowledge Corweve's condition is to acknowledge that a company 0:08:19.880 --> 0:08:22.280 pumped up by most of the media is going to explode. 0:08:22.720 --> 0:08:25.440 To acknowledge that Anthropic is basically the same company and 0:08:25.520 --> 0:08:28.040 in the same financial position as open Ai is to 0:08:28.120 --> 0:08:30.840 accept that Google Cloud and Amazon Web Services are as 0:08:30.880 --> 0:08:34.040 inflated as Microsoft, as your and to acknowledge these things 0:08:34.080 --> 0:08:36.840 would require the media to acknowledge that they ignored basically 0:08:36.920 --> 0:08:39.920 every warning sign right up until the end. I can 0:08:39.960 --> 0:08:41.679 imagine that the bubble will burst and they'll be out 0:08:41.760 --> 0:08:43.400 let's the world attempt to pretend that they were ahead 0:08:43.440 --> 0:08:46.080 of it. But those people don't realize how detailed my 0:08:46.160 --> 0:08:50.679 notes are and how unflinching I'll be in publishing them. 0:08:50.960 --> 0:08:55.000 I will, I will be doing so. I will be 0:08:55.120 --> 0:08:59.120 publishing everything. I know everyone who pumped this up. I 0:08:59.200 --> 0:09:02.160 know what everyone did, I know how many times they 0:09:02.280 --> 0:09:05.439 ignored the warning science and I can see the articles 0:09:05.480 --> 0:09:09.120 where they helped pump this bubble up. And I'm not 0:09:09.160 --> 0:09:11.320 going to get too personal. I'm just going to publish 0:09:11.360 --> 0:09:13.480 the fucking articles and I'm going to point. I'm going 0:09:13.520 --> 0:09:16.800 to say it was obvious, then why didn't you do something? 0:09:17.800 --> 0:09:20.400 And the answer is they didn't give a shit. Either 0:09:20.440 --> 0:09:22.640 they didn't give a shit enough to learn, or maybe 0:09:22.720 --> 0:09:25.120 they and it's much worse if they understood and they 0:09:25.160 --> 0:09:27.360 just chose to ignore it. If the answer is, well, 0:09:27.440 --> 0:09:30.000 the largest companies in the world wouldn't make such a mistake, 0:09:31.000 --> 0:09:34.760 You're just very, very gullible. I guess I'll know. I 0:09:34.800 --> 0:09:37.839 hope that someone learns after this. I don't think they 0:09:37.920 --> 0:09:41.720 will really gonna try. I'm really gonna try and explain 0:09:41.720 --> 0:09:43.680 how people got this wrong. I feel like I've been trying, 0:09:44.040 --> 0:09:46.400 and the bile you can hear, the poison in my 0:09:46.520 --> 0:09:50.280 voice comes from a place of It's funny. People say, oh, 0:09:50.280 --> 0:09:52.880 he's just a form of pr person. He doesn't know 0:09:52.880 --> 0:09:55.080 anything about finance. The only thing that it takes to 0:09:55.120 --> 0:09:58.120 become a financial expert is to become an expert in finance, 0:09:58.120 --> 0:10:00.959 to sit down and learn. Everyone else you can fuck it. 0:10:01.000 --> 0:10:02.760 I'm using the same shit as everyone else. You just 0:10:02.840 --> 0:10:04.920 learned to read. And I don't know why more people 0:10:05.000 --> 0:10:06.800 don't do it. For the answer is it's a lot 0:10:06.840 --> 0:10:08.120 of work, and you don't really have to do it 0:10:08.160 --> 0:10:10.240 when your editors pro ai and you can just do 0:10:10.320 --> 0:10:12.640 whatever you want. I know some journalists listen, and hopefully 0:10:12.679 --> 0:10:15.000 I'm not talking about you. Hopefully you made the effort. 0:10:15.200 --> 0:10:18.480 I really hope you did. And I must be clear 0:10:18.480 --> 0:10:21.200 on I'll probably mostly be coming after outlets rather than 0:10:21.240 --> 0:10:23.520 people directly. That would be mean. But at the same time, 0:10:23.760 --> 0:10:27.520 please God, when this explodes, journalists listening to this, please 0:10:27.559 --> 0:10:31.160 fucking try harder or I don't know, try trying harder beforehand. 0:10:31.320 --> 0:10:34.120 And if you think you are pushing back, if you 0:10:34.240 --> 0:10:36.800 truly think you are pushing back by all means, reach 0:10:36.880 --> 0:10:39.560 out and find out if you actually are, by all means, 0:10:39.600 --> 0:10:42.000 email me, signal me you have my information? Is it 0:10:42.040 --> 0:10:46.040 froet seventy six and ask me directly, because pretty much 0:10:46.080 --> 0:10:47.600 if you have to ask, you probably not. And if 0:10:47.600 --> 0:10:49.520 you think you are already, I don't know. Have you 0:10:49.559 --> 0:10:51.600 talked about core weave. Have you talked about open ai 0:10:51.720 --> 0:10:54.319 running out of money? Or when you did so, did 0:10:54.360 --> 0:10:57.439 you talk about it saying yeah, open Ai isn't profitable. Yeah, 0:10:57.600 --> 0:11:00.440 but you know, companies take a while to get comfortable, 0:11:00.520 --> 0:11:02.880 and Uber lost a lot of money, in Aws lost 0:11:02.920 --> 0:11:04.760 a lot of money. If that's what you're thinking, you 0:11:04.800 --> 0:11:07.400 are wrong. You are wrong. You are misleading your readers, 0:11:07.480 --> 0:11:09.920 you're misleading the general public, and you should be kind 0:11:09.920 --> 0:11:12.200 of fucking ashamed of yourself. I'm I am sorry to 0:11:12.240 --> 0:11:14.640 be so merciless right now, but I am sick and 0:11:14.720 --> 0:11:17.600 tired of watching these cycles, and I know how many 0:11:17.640 --> 0:11:19.800 people are going to get washed out. For a real 0:11:19.880 --> 0:11:22.240 preview of that, go and look at the Korean stock 0:11:22.280 --> 0:11:24.440 market right now, and the many stories of people talking 0:11:24.440 --> 0:11:29.400 about fucking killing themselves. That is the real consequence of bubbles, 0:11:29.600 --> 0:11:32.720 That is the real thing that happens, and everyone trying 0:11:32.720 --> 0:11:37.560 to downplay this is responsible for people like that suffering 0:11:37.679 --> 0:11:42.440 and eventually American and British and other investors too. It's sickening, 0:11:43.000 --> 0:11:44.920 and until something changes, we're all going to live in 0:11:44.920 --> 0:11:49.240 a paradoxical media environment that simultaneously publishes proof point after 0:11:49.280 --> 0:11:52.520 proof point after proof point that this is financially unstable, 0:11:52.720 --> 0:11:54.800 all while saying that AI is going to be the biggest, 0:11:54.840 --> 0:11:57.400 most hugest thing in history, right up until the bottom 0:11:57.440 --> 0:12:01.720 falls out. Hope a AI collapses, data centers get canceled, 0:12:01.800 --> 0:12:04.400 core we've shits its pants and everything falls apart, at 0:12:04.400 --> 0:12:06.200 which point they'll throw up their hands and say, Wow, 0:12:06.200 --> 0:12:09.040 how'd that all happen? It happened because we live in 0:12:09.080 --> 0:12:13.000 an era where responsibility often correlates with cowardice. Those who 0:12:13.080 --> 0:12:15.600 have the most power to critique the powerful choose instead 0:12:15.640 --> 0:12:18.559 to jerk their shit from first principles, helping them create 0:12:18.640 --> 0:12:22.160 fantastical mediated realities that enrich the already rich and fuck 0:12:22.240 --> 0:12:27.319 over retail investors at scale. It darkens my heart. When 0:12:27.320 --> 0:12:30.320 I think about it, it really pisses me off because 0:12:30.360 --> 0:12:33.400 you can already see what's gonna happen. You can already 0:12:33.440 --> 0:12:35.600 see that the people invested in the markets are going 0:12:35.679 --> 0:12:38.920 to get washed out. And then there's the widespread problems 0:12:38.960 --> 0:12:41.720 caused by all of these private credit funds sinking money 0:12:41.800 --> 0:12:44.120 into data centers that will never get built, will never 0:12:44.160 --> 0:12:47.000 provide the yield, and will fuck over anyone who is 0:12:47.080 --> 0:12:51.480 unfortunate to have insurance annuities or pension funds involved in 0:12:51.520 --> 0:12:55.760 this bullshit. Next week, i'll've ed Elson back on to 0:12:55.760 --> 0:12:57.880 talk about tech earnings and continue to fight to give 0:12:57.880 --> 0:13:00.240 you as clear a picture of reality as possible, even 0:13:00.280 --> 0:13:02.280 as we battled the lead one created by and for 0:13:02.400 --> 0:13:04.520 the powerful. Thank you for listening.