WEBVTT - AI Is Worse Than The Dot Com Bubble: Part Four 0:00:03.160 --> 0:00:07.360 Media. Hello and welcome to bear off Line. I'm your 0:00:07.360 --> 0:00:22.160 host ed Zetron. Today is the final part of Bubble 0:00:22.200 --> 0:00:25.279 Week and our four part dot com Bubble special. I 0:00:25.320 --> 0:00:28.160 am worryingly invested in this show. I've put so many 0:00:28.160 --> 0:00:30.080 hours into it. This is the most I've done in 0:00:30.120 --> 0:00:32.760 the January. I'm having the time of my fucking life. 0:00:33.320 --> 0:00:36.120 But this four part has been rough, and it's not 0:00:36.159 --> 0:00:38.639 been rough for the recording. It's rough because every time 0:00:38.680 --> 0:00:42.159 I go over these details, they fill me full of anxiety. 0:00:42.280 --> 0:00:46.520 I don't enjoy being right here. Sure I get the satisfaction, 0:00:46.680 --> 0:00:48.800 Sure I get the clicks and all that, but like 0:00:51.080 --> 0:00:55.480 it sucks, It sucks. I'm really fucking worried. I'm not 0:00:55.600 --> 0:00:58.280 doing this to be contrarian. I'm seeing these things and 0:00:58.320 --> 0:01:00.400 I'm worried more people aren't seeing them, and I'm worried 0:01:00.400 --> 0:01:04.120 that the after effects will be so much worse than 0:01:04.120 --> 0:01:08.040 the dot com bubble. So this Foe part was somewhat 0:01:08.120 --> 0:01:09.920 drawn from a premium piece I did a few weeks 0:01:09.959 --> 0:01:12.959 ago called Inventively. This is worse than the dot com bubble. 0:01:13.000 --> 0:01:15.880 I really suggest you subscribe to my premium newsletter or 0:01:15.959 --> 0:01:17.880 be in there. But if you don't. I've had a 0:01:17.880 --> 0:01:20.560 little mustard on it and truncated it for this because 0:01:20.920 --> 0:01:24.679 you needed to be different in voice. Nevertheless, you should subscribe. 0:01:24.920 --> 0:01:27.600 But when I was writing that, I actually went into 0:01:27.640 --> 0:01:31.039 it trying to prove myself wrong. I spent hours and 0:01:31.120 --> 0:01:34.480 hours reading old articles through broken paywalls and archive dot 0:01:34.560 --> 0:01:36.920 org links, in the hopes that I'd find some sort 0:01:36.959 --> 0:01:39.640 of positive spin, a way that the AI bubble wasn't 0:01:39.680 --> 0:01:42.399 as bad, some sliver or nugget of joy, some shread 0:01:42.440 --> 0:01:45.160 of hope. And everything I've read has made me more worried, 0:01:45.280 --> 0:01:49.120 not less. The real problem is that there are similarities, 0:01:49.160 --> 0:01:52.280 but only with the worst parts of both bubbles. We've 0:01:52.280 --> 0:01:55.200 got stupid amounts of money being sunk into stuff. We've 0:01:55.200 --> 0:01:58.680 got stupid companies that want to ypo. We've got hopeless 0:01:58.680 --> 0:02:02.360 startups duced up by edulous venture capitalists and marketed by 0:02:02.400 --> 0:02:05.640 even more credulous journalists. Except this time we have the 0:02:05.640 --> 0:02:09.520 worst margins this side of a vineyard planted in Las Vegas, Nevada, 0:02:09.560 --> 0:02:11.760 and despicable boosters that will go to war with you 0:02:11.840 --> 0:02:14.440 to try and improve You're wrong. Using ghost stories, fairy tales, 0:02:14.440 --> 0:02:17.880 and Bixie dust, I realize I'm being a little cavalier 0:02:17.880 --> 0:02:22.400 and colloquial about this, and that's because I feel emotionally drained, 0:02:22.680 --> 0:02:25.359 is the way i'd put in. Things are so much 0:02:25.400 --> 0:02:28.720 worse than I thought. I really thought i'd find something good. 0:02:28.760 --> 0:02:30.920 I'd dig into the past and find that there were direct, 0:02:31.280 --> 0:02:35.720 obvious similarities that would say, Okay, these GPUs will be useful. 0:02:35.760 --> 0:02:40.000 There will be something we've built here. I can't find it. 0:02:40.080 --> 0:02:42.239 I'm actually now convinced that things are going to be 0:02:42.320 --> 0:02:45.600 much worse. The dot com bubble was relatively distributed and 0:02:45.639 --> 0:02:49.080 created useful infrastructure. The ideas of some of the most 0:02:49.120 --> 0:02:52.520 prominent dot com startups were valid, but based on stupid 0:02:52.600 --> 0:02:55.200 hypergrowth business models that didn't make sense and were only 0:02:55.240 --> 0:02:59.959 sustained through mass hysteria and ignorance. By comparison, AI startups 0:03:00.120 --> 0:03:03.880 fundamentally illogical or selling unreliable software that's best known for 0:03:03.919 --> 0:03:06.919 its consistent mistakes and active harms, backed by the corrosive 0:03:07.000 --> 0:03:10.760 and ever worsening infrastructural costs of GPUs, and they're the 0:03:10.800 --> 0:03:13.440 customers of the largest company on the stock market that 0:03:13.520 --> 0:03:16.520 can only keep its valuation on which the markets depend 0:03:16.800 --> 0:03:18.919 by selling more of it. An expensive to buy and 0:03:19.000 --> 0:03:21.720 more expensive to install item that only appears to lose 0:03:21.760 --> 0:03:25.760 its customers money. There is no squaring in Vidia's circle. 0:03:26.200 --> 0:03:29.600 It is inevitable even if AI somehow worked out that 0:03:29.720 --> 0:03:32.840 companies will have to stop buying as many GPUs, and 0:03:32.840 --> 0:03:35.880 in Vidia cannot afford for them to slow down at all. 0:03:36.000 --> 0:03:38.600 They must speed up, and they must do so forever. 0:03:39.640 --> 0:03:43.440 There's also no other business model for GPUs that scales none. 0:03:43.480 --> 0:03:46.520 There is no use case. This is not useful infrastructure. 0:03:46.920 --> 0:03:49.680 Once the AI bubble bursts, these data centers cannot be 0:03:49.720 --> 0:03:51.800 used for other things, at least not things that are 0:03:51.840 --> 0:03:53.800 going to make more than a couple of single digit 0:03:53.880 --> 0:03:56.560 millions in total. They will have to be emptied out 0:03:56.600 --> 0:03:59.040 and replaced with useful gear. They will be those that 0:03:59.080 --> 0:04:00.760 are finished, and I think going to be a bunch 0:04:00.800 --> 0:04:04.080 of undone construction. They're going to be powered shells, which 0:04:04.080 --> 0:04:07.080 are just empty buildings with power attached. And really, I 0:04:07.120 --> 0:04:10.080 don't think that's gonna be. That's gonna be across the board. 0:04:10.080 --> 0:04:11.880 I think we're gonna have a lot of failed construction. 0:04:12.520 --> 0:04:14.680 And throughout all of this bullshit, the rest of them, 0:04:14.760 --> 0:04:17.680 Magnificent seven has seen their stock values saw off the 0:04:17.680 --> 0:04:20.559 back of pure hype because they weren't making a dime 0:04:20.600 --> 0:04:24.359 of profit from selling artificial intelligence services, and they've done 0:04:24.400 --> 0:04:27.800 so by it acquiring hundreds of billions of dollars of 0:04:27.880 --> 0:04:31.400 physical assets. The whole point of these fucking companies was 0:04:31.440 --> 0:04:35.159 that they were asset light and cash heavy. Now they're 0:04:35.200 --> 0:04:39.240 full of fucking GPUs to depreciate. This is a fucking calamity. 0:04:39.480 --> 0:04:42.640 I am fucking terrified because everything I've read about the 0:04:42.680 --> 0:04:44.960 past makes me certain that this will be worse. The 0:04:45.040 --> 0:04:47.440 hands will be longer, and nobody seems to want to 0:04:47.480 --> 0:04:51.560 accept the true logical endpoint of my arguments, So I'm 0:04:51.600 --> 0:04:54.560 gonna kind of lay them out depending on how right 0:04:54.640 --> 0:04:57.760 I am. I'm worrying about the following things. And Nvidia 0:04:57.800 --> 0:05:00.080 will lose anywhere between forty percent and night besore and 0:05:00.160 --> 0:05:02.280 of its revenue in a relatively short amount of time, 0:05:02.360 --> 0:05:05.239 a couple of quarters maybe, which will annihilate the value 0:05:05.240 --> 0:05:07.200 of the S and P five hundred because the stock 0:05:07.240 --> 0:05:11.360 will go down because the market will go Ah, Microsoft, 0:05:11.400 --> 0:05:15.000 Google and Amazon will face massive multi billion dollar impairments 0:05:15.080 --> 0:05:17.240 within the next few years. For God to say, meta, 0:05:17.279 --> 0:05:20.760 but they are included too. Now, this one is funny 0:05:21.120 --> 0:05:24.880 but will still be bad. Oracle will potentially face bankruptcy 0:05:24.960 --> 0:05:27.479 or at least a severe horrendous series of layoffs and 0:05:27.520 --> 0:05:31.080 restructuring efforts to try and write the shift after agreeing 0:05:31.080 --> 0:05:33.240 to two hundred and forty eight billion dollars in lease 0:05:33.279 --> 0:05:37.000 obligations and fifty six billion dollars in debt primarily to 0:05:37.040 --> 0:05:39.120 build data centers to get the revenue from its three 0:05:39.200 --> 0:05:42.240 hundred billion dollar five year long deal with open Ai, 0:05:42.560 --> 0:05:44.560 a company that's going to run out of money in 0:05:44.600 --> 0:05:47.800 the next year. Venture Capital is going to take such 0:05:47.800 --> 0:05:50.320 a severe haircut from the AI bubble that it will 0:05:50.320 --> 0:05:52.920 effectively kill any chances of raising a round off the 0:05:52.960 --> 0:05:56.880 Series B for many, many years. I'm really worried about 0:05:56.880 --> 0:05:59.160 the reported one hundred and seventy eight point five billion 0:05:59.200 --> 0:06:03.400 dollars in US date center deals going underwater. I don't 0:06:03.440 --> 0:06:04.800 think they're going to be paid, and it's going to 0:06:04.920 --> 0:06:07.320 lead to massive losses with private equity and banks, which 0:06:07.360 --> 0:06:10.040 will lead to a chill in effectively all debt markets 0:06:10.279 --> 0:06:13.720 and massive job losses across the many many construction projects 0:06:13.760 --> 0:06:15.880 that have started as a result of the AI bubble, 0:06:16.440 --> 0:06:18.680 and if open Ai and Anthropic can actually manage to 0:06:18.680 --> 0:06:21.560 go public. I think every investor is going to lose 0:06:21.560 --> 0:06:24.200 out as they collapse under the weight of their bullshit assumptions. 0:06:24.600 --> 0:06:28.560 Every partner, every company relying on them, every infrastructure partner, 0:06:28.800 --> 0:06:31.360 every AI starter built on their tech will be burned. 0:06:32.560 --> 0:06:35.320 I also believe that we as a society will face 0:06:35.360 --> 0:06:38.080 a violent paradigm shift as the result of the AI 0:06:38.120 --> 0:06:40.920 bubble bursting, because it revealed how many people just don't 0:06:41.000 --> 0:06:45.479 understand stuff. The amount of bosses, influencers, movie stars, politicians, 0:06:45.480 --> 0:06:48.120 and other individuals of note that have fallen behind generative 0:06:48.160 --> 0:06:50.280 AI and claimed again and again that it can do 0:06:50.320 --> 0:06:52.680 things that it can't have effatuated myths that will lead 0:06:52.680 --> 0:06:55.000 to massive ciple harms both day and in the future, 0:06:55.160 --> 0:07:12.520 have revealed themselves to be fucking frauds. It's been obvious 0:07:12.560 --> 0:07:15.240 since twenty twenty three that generative AI flat out does 0:07:15.280 --> 0:07:17.640 not do what many people say it does or will do, 0:07:18.000 --> 0:07:21.000 and obvious since twenty twenty four that the economics didn't work. 0:07:21.200 --> 0:07:24.360 Yet still, every single goddamn day, we are surrounded by 0:07:24.360 --> 0:07:27.200 people screaming in our faces the AI is the future, 0:07:27.400 --> 0:07:29.920 that AI will replace us, that AI is the panacea 0:07:29.960 --> 0:07:32.320 to every fucking problem, even though it only seems to 0:07:32.320 --> 0:07:36.040 create new ones. I also believe that AI psychosis is 0:07:36.080 --> 0:07:38.280 a prevalent thing on a scale that few people want 0:07:38.320 --> 0:07:41.160 to discuss. There's something about the process of using a 0:07:41.200 --> 0:07:44.000 large language model that convinces people that AI is powerful 0:07:44.120 --> 0:07:46.240 because they managed to bunk it on the head enough 0:07:46.280 --> 0:07:50.240 times to make something useful come out seemingly intelligent. People 0:07:50.240 --> 0:07:52.880 will spend hours and hours making clawed code burb out 0:07:53.040 --> 0:07:56.880 quasi useful, quasi functional software, claiming that the productivity gains 0:07:56.880 --> 0:07:59.320 are obvious, even when if you sat down and worked 0:07:59.360 --> 0:08:01.040 out how long it took them and how much the 0:08:01.040 --> 0:08:03.080 API rates for the software would be. If they weren't 0:08:03.120 --> 0:08:05.400 being subsidized, they could probably have just learned to code 0:08:05.440 --> 0:08:07.440 in the first place, or paid someone else to do it. 0:08:07.760 --> 0:08:10.400 And no, please stop this thing of oh it's cheaper 0:08:10.400 --> 0:08:12.560 than hiring a developer. Did you go out and ask 0:08:12.560 --> 0:08:14.720 a developer? Did you actually get a quote? Do you 0:08:14.760 --> 0:08:16.400 actually know what you're talking about? Do you know a 0:08:16.440 --> 0:08:20.280 fucking thing about you're talking about? You see, I think 0:08:20.440 --> 0:08:23.680 large language models that are a global gas lighting experiment, 0:08:23.960 --> 0:08:26.880 where the test is to see whether inefficient and questionably 0:08:26.920 --> 0:08:29.880 functional software can convince you that it works because you 0:08:29.960 --> 0:08:33.240 managed to manipulate its handles with the right prompts, explaining 0:08:33.280 --> 0:08:36.560 away failures as you go, and getting increasingly agitated with 0:08:36.559 --> 0:08:38.640 those who don't agree with you that you're living in 0:08:38.679 --> 0:08:42.120 the future. This series forced me to live in the 0:08:42.160 --> 0:08:44.440 past for hours and hours and hours, and all it 0:08:44.520 --> 0:08:47.160 taught me is that people are desperate to compare this 0:08:47.240 --> 0:08:49.640 era to another as a means of explaining away the 0:08:49.720 --> 0:08:52.320 bad habits and specious ideas of a tech industry run 0:08:52.360 --> 0:08:56.600 by fucking influences and management consoles. By comparison, the dot 0:08:56.640 --> 0:09:00.000 com bubble is relatively sensible, built on good ideas execut 0:09:00.200 --> 0:09:02.680 by greedy people rushing to do things based on ideas 0:09:02.679 --> 0:09:06.439 they didn't understand, and analyzes that were utter bullshit. The 0:09:06.559 --> 0:09:09.400 AI bubble is happening in a remarkable era of digital 0:09:09.440 --> 0:09:12.840 information and connectivity that allows us to process and analyze 0:09:12.840 --> 0:09:16.120 details at scale and with speed, And everybody who inflated 0:09:16.160 --> 0:09:18.680 this era should, and, by my fucking sword, will be 0:09:18.760 --> 0:09:22.760 made to experience inextricable, permanent shame for the horrors that 0:09:22.800 --> 0:09:25.680 follow as a result. And believe me, even if I'm 0:09:25.800 --> 0:09:28.400 half right, things will likely be so much worse than 0:09:28.440 --> 0:09:30.840 when the dot com bubble burst. I don't see a 0:09:30.840 --> 0:09:33.480 scenario where anything survives the collapse of core we've let 0:09:33.520 --> 0:09:35.760 alone the collapse of open AI. I don't see a 0:09:35.800 --> 0:09:38.400 scenario where any more than a few AI starts survive, 0:09:38.679 --> 0:09:40.720 if any and if that happens, there won't be any 0:09:40.720 --> 0:09:44.120 customers for those GPUs, not that anywhere near enough existed 0:09:44.120 --> 0:09:46.440 to fill even a tiny percentile of the demand that 0:09:46.480 --> 0:09:50.040 currently exists, or for the capacity we've already built. On 0:09:50.080 --> 0:09:52.680 some level. I hope I'm wrong. I just don't know 0:09:52.720 --> 0:09:56.199 how anymore. I really don't thank you so much for listening. 0:09:56.280 --> 0:09:59.319 It's been in intense January with this, with CES, with 0:09:59.360 --> 0:10:02.760 the inchittaphone ential crisis. I'm probably going to take the 0:10:02.760 --> 0:10:04.920 week off next week and run a call zone media 0:10:05.000 --> 0:10:07.480 rewind to give my aching brain and lungs a break. 0:10:07.800 --> 0:10:11.120 Thank you so much for listening. Who knows, though, some 0:10:11.160 --> 0:10:14.200 bullshit will probably happen because it's earning season when this runs, 0:10:14.559 --> 0:10:16.720 some crazy shit will happen, or maybe it'll be normal. 0:10:16.800 --> 0:10:19.480 I could just rest, and if that happens, I'll probably 0:10:19.520 --> 0:10:23.079 end up pinking and squawking into the micros. I always though, regardless, 0:10:23.120 --> 0:10:25.360 you'll hear my voice one way or another. Thank you 0:10:25.440 --> 0:10:37.600 so much everyone, This has been bubble week. Thank you 0:10:37.640 --> 0:10:39.000 for listening to Better Offline. 0:10:39.120 --> 0:10:41.559 The editor and composer of the Better Offline theme song 0:10:41.640 --> 0:10:44.240 is Matasowski. You can check out more of his music 0:10:44.280 --> 0:10:47.960 and audio projects at Mattasowski dot com, m A T 0:10:47.960 --> 0:10:52.400 T O S O W s ki dot com. You 0:10:52.440 --> 0:10:54.920 can email me at easy at Better Offline dot com 0:10:55.040 --> 0:10:57.319 or visit Better Offline dot com to find more podcast 0:10:57.400 --> 0:11:00.720 links and of course, my newsletter. I also really recommend 0:11:00.760 --> 0:11:02.679 you go to chat dot Where's Youread dot at to 0:11:02.760 --> 0:11:05.160 visit the discord, and go to our slash. 0:11:04.840 --> 0:11:08.000 Better Offline to check out I'll Reddit. Thank you so 0:11:08.080 --> 0:11:11.520 much for listening. Better Offline is a production of cool 0:11:11.600 --> 0:11:12.160 Zone Media. 0:11:12.280 --> 0:11:15.120 For more from cool Zone Media, visit our website cool 0:11:15.200 --> 0:11:18.520 Zonemedia dot com or check us out on the iHeartRadio app, 0:11:18.600 --> 0:11:21.280 Apple Podcasts, or wherever you get your podcasts.