1 00:00:02,800 --> 00:00:08,080 Speaker 1: Ze Media, Hello, and welcome to this week's Better Offline Monologue. 2 00:00:08,119 --> 00:00:18,880 Speaker 1: I'm your host ed Ze tron. That's right, it's your 3 00:00:18,960 --> 00:00:21,239 Speaker 1: second damn monologue this week. But next week we're going 4 00:00:21,280 --> 00:00:24,360 Speaker 1: to have an awesome guest economist Paul Kajowski, and no 5 00:00:24,440 --> 00:00:26,160 Speaker 1: doubt some sort of us will break that will make 6 00:00:26,200 --> 00:00:29,160 Speaker 1: next week's monologue even spicier than this one. And man 7 00:00:29,240 --> 00:00:32,479 Speaker 1: is this spicy anyway. Earlier this week, I put out 8 00:00:32,479 --> 00:00:36,120 Speaker 1: a free newsletter about the SUBPRIMEI crisis. My ongoing theory 9 00:00:36,120 --> 00:00:38,720 Speaker 1: that as AI companies strive to try and make their 10 00:00:38,800 --> 00:00:41,320 Speaker 1: rotten economics work, they'll left to start cranking up the 11 00:00:41,320 --> 00:00:44,280 Speaker 1: price and making rate limits worse and generally trying to 12 00:00:44,280 --> 00:00:47,440 Speaker 1: move things around to make these products anything close to profitable. 13 00:00:47,800 --> 00:00:49,160 Speaker 1: But they won't even get close. 14 00:00:49,960 --> 00:00:50,920 Speaker 2: Now take a. 15 00:00:50,920 --> 00:00:53,159 Speaker 1: Little history trip. That's how I'm going to That's what 16 00:00:53,200 --> 00:00:56,440 Speaker 1: I'm calling it. When the subprime crisis happened, the subcrime 17 00:00:56,560 --> 00:00:59,480 Speaker 1: mortgage crisis, of course, millions of people built their lives 18 00:00:59,480 --> 00:01:02,040 Speaker 1: around the idea that easy money would always be available 19 00:01:02,280 --> 00:01:04,319 Speaker 1: and that anyone could get a mortgage, and that housing 20 00:01:04,360 --> 00:01:08,119 Speaker 1: would only ever increase in value. In reality, the value 21 00:01:08,120 --> 00:01:11,160 Speaker 1: of housing was massively overinflated by the lacks standards of 22 00:01:11,200 --> 00:01:14,119 Speaker 1: a mortgage industry incentivized the sign as many people as 23 00:01:14,120 --> 00:01:17,680 Speaker 1: possible thanks to a lack of regulation and easily available funding, 24 00:01:18,959 --> 00:01:21,360 Speaker 1: the value of housing and indeed the larger housing and 25 00:01:21,440 --> 00:01:25,440 Speaker 1: construction boom was of mirage. In reality, housing wasn't worth 26 00:01:25,440 --> 00:01:27,319 Speaker 1: anywhere near what it was being sold for, and the 27 00:01:27,319 --> 00:01:31,440 Speaker 1: massive demand for housing was only possible with unlimited resources 28 00:01:31,760 --> 00:01:34,960 Speaker 1: and lacks well underwriting. So that which is the process 29 00:01:35,000 --> 00:01:37,520 Speaker 1: of evaluating whether someone can get a house. You might 30 00:01:37,600 --> 00:01:39,800 Speaker 1: know that, but I've been encouraged by Robert and Sourphy 31 00:01:39,800 --> 00:01:42,920 Speaker 1: to be obvious with things. Those buying houses they couldn't 32 00:01:42,920 --> 00:01:46,119 Speaker 1: afford with adjustable rate mortgages either didn't understand the terms 33 00:01:46,200 --> 00:01:48,760 Speaker 1: or believed members of the media and government officials that 34 00:01:48,840 --> 00:01:51,760 Speaker 1: suggested housing prices would never decrease and that one could 35 00:01:51,840 --> 00:01:54,920 Speaker 1: easily refinance the mortgage in question. You know, kind of 36 00:01:54,960 --> 00:01:58,400 Speaker 1: like saying things like, you know, AI is always getting 37 00:01:58,400 --> 00:02:01,080 Speaker 1: more efficient, the values always going up, and venture capital 38 00:02:01,080 --> 00:02:02,200 Speaker 1: will always invest in that. 39 00:02:02,240 --> 00:02:03,840 Speaker 2: This is the new hypergrowth era. 40 00:02:04,400 --> 00:02:08,640 Speaker 1: Similarly, AI startups products are all subsidized by venture capital 41 00:02:08,760 --> 00:02:11,720 Speaker 1: and must in literally every case allow users to burn 42 00:02:11,800 --> 00:02:14,600 Speaker 1: tokens far and excess of their subscription fees. A business 43 00:02:14,600 --> 00:02:17,680 Speaker 1: that only works, and I put that in quotation marks 44 00:02:17,720 --> 00:02:20,480 Speaker 1: as long as venture capital continues to fund it. And 45 00:02:20,520 --> 00:02:22,720 Speaker 1: when I say that, I'm being quite literal. If you 46 00:02:22,720 --> 00:02:25,400 Speaker 1: go and use Perplexity, you're comfortably able to even on 47 00:02:25,440 --> 00:02:27,720 Speaker 1: a twenty dollars plan, burn thirty forty to fifty one 48 00:02:27,760 --> 00:02:31,000 Speaker 1: hundred dollars worth of tokens within account a month. Anthropic 49 00:02:31,080 --> 00:02:34,040 Speaker 1: allowed you, until very recently, to burn anywhere between eight 50 00:02:34,080 --> 00:02:37,400 Speaker 1: to thirteen and a half dollars per dollar of subscription revenue. 51 00:02:37,639 --> 00:02:40,440 Speaker 1: While from the outside these may seem like these are 52 00:02:40,560 --> 00:02:44,520 Speaker 1: functional businesses with paying users. Without the hype cycle justifying 53 00:02:44,520 --> 00:02:47,640 Speaker 1: the endless capital, these businesses wouldn't be possible, let alone 54 00:02:47,720 --> 00:02:50,280 Speaker 1: viable in any way, shape or form, And indeed they 55 00:02:50,320 --> 00:02:53,919 Speaker 1: wouldn't have any customers. My evidence being if they could 56 00:02:53,960 --> 00:02:57,080 Speaker 1: get customers by charging their actual rates and offering a 57 00:02:57,520 --> 00:03:00,720 Speaker 1: non subsidized product, they'd have them, and they would have 58 00:03:00,760 --> 00:03:03,239 Speaker 1: had them from the beginning. Let me give you an example. 59 00:03:03,680 --> 00:03:06,200 Speaker 1: Harvey is an AI tool for lawyers that just raised 60 00:03:06,200 --> 00:03:09,280 Speaker 1: two hundred million dollars at an astonishing eleven billion dollar 61 00:03:09,400 --> 00:03:13,760 Speaker 1: valuation or while having an equally astonishingly small one hundred 62 00:03:13,800 --> 00:03:17,079 Speaker 1: ninety million dollars in AR or fifteen point eight million 63 00:03:17,080 --> 00:03:19,880 Speaker 1: dollars a month. It raised another one hundred and sixty 64 00:03:19,919 --> 00:03:22,720 Speaker 1: million dollars in December twenty twenty five, after raising three 65 00:03:22,800 --> 00:03:25,440 Speaker 1: hundred million dollars in June twenty twenty five. After raising 66 00:03:25,440 --> 00:03:28,320 Speaker 1: three hundred million dollars in February twenty twenty five, where's 67 00:03:28,320 --> 00:03:30,960 Speaker 1: the fucking money, Harvey? Were you putting it? 68 00:03:31,120 --> 00:03:31,520 Speaker 2: Harvey? 69 00:03:31,639 --> 00:03:34,519 Speaker 1: Mister Harvey has been very unfair to the venture capitalists. 70 00:03:34,760 --> 00:03:37,800 Speaker 1: Actually they're fueling I can't even say that. Remove even 71 00:03:37,880 --> 00:03:40,600 Speaker 1: one of those venture capital rounds and Harvey coughs up 72 00:03:40,640 --> 00:03:44,480 Speaker 1: blood and dies. Much like subprime loans allowed borrowers to 73 00:03:44,480 --> 00:03:47,520 Speaker 1: get mortgages they had no hope of paying, hype cycles 74 00:03:47,560 --> 00:03:50,400 Speaker 1: create the illusion of viable businesses that cannot and will 75 00:03:50,440 --> 00:03:53,880 Speaker 1: never survive without the subsidies. The same does for companies 76 00:03:53,920 --> 00:03:57,000 Speaker 1: like open Ai and Anthropic, both of whom created priority 77 00:03:57,080 --> 00:04:00,080 Speaker 1: processing tiers for their enterprise customers in the middle of 78 00:04:00,120 --> 00:04:02,840 Speaker 1: twenty twenty five, and the latter of which, just as 79 00:04:02,840 --> 00:04:05,680 Speaker 1: I discussed in my last monologue, added peak rate limits 80 00:04:05,680 --> 00:04:08,320 Speaker 1: from five am to eleven pm Pacific time, you know, 81 00:04:08,480 --> 00:04:12,080 Speaker 1: just the entire day, and that was after they created 82 00:04:12,080 --> 00:04:15,760 Speaker 1: weekly limits late last year. The customers are the subprime 83 00:04:15,800 --> 00:04:19,159 Speaker 1: borrowers too. They built workflows around using these products that 84 00:04:19,200 --> 00:04:21,120 Speaker 1: may or may not be possible with new rate limits. 85 00:04:21,160 --> 00:04:21,760 Speaker 2: Think about it. 86 00:04:21,920 --> 00:04:24,600 Speaker 1: If your whole business, the whole reason you use this 87 00:04:24,960 --> 00:04:28,400 Speaker 1: software subscription is to do tasks, and suddenly the amount 88 00:04:28,440 --> 00:04:30,960 Speaker 1: of tasks you can do is limited. Is this really 89 00:04:31,320 --> 00:04:33,680 Speaker 1: is anything tenable anymore? Especially if you're one of those 90 00:04:33,920 --> 00:04:36,520 Speaker 1: people who can't actually code and using this to vibe code. 91 00:04:36,520 --> 00:04:37,719 Speaker 2: I'm not really sure this works. 92 00:04:38,000 --> 00:04:41,080 Speaker 1: But in the case of the enterprise customers using Priority processing, 93 00:04:41,200 --> 00:04:44,240 Speaker 1: their costs massively spiked, which is why Cursor and Replet 94 00:04:44,240 --> 00:04:47,560 Speaker 1: and several other AI startups suddenly made their products worse 95 00:04:47,560 --> 00:04:49,720 Speaker 1: in the middle of twenty twenty five, adding their own 96 00:04:49,760 --> 00:04:53,599 Speaker 1: rate limits and changing their pricing. In reality, none of 97 00:04:53,600 --> 00:04:57,840 Speaker 1: this ever made sense. None of this was actually possible 98 00:04:58,080 --> 00:05:02,440 Speaker 1: outside of endless resources. Now traveling back in time, by 99 00:05:02,440 --> 00:05:05,920 Speaker 1: November twenty nine, twenty three percent of US consumer mortgages 100 00:05:05,960 --> 00:05:08,480 Speaker 1: were underwater, meaning that they were worth less than their loans. 101 00:05:08,800 --> 00:05:11,400 Speaker 1: And I think we're eventually going to see that specifically 102 00:05:11,480 --> 00:05:14,400 Speaker 1: with venture capital valuations. By the way, I think there's 103 00:05:14,400 --> 00:05:17,520 Speaker 1: going to be a point when it's like ninety percent 104 00:05:17,560 --> 00:05:21,159 Speaker 1: of AI startup valuations are going to be way lower 105 00:05:21,440 --> 00:05:25,119 Speaker 1: or nil actually, and I truly think the subprime AI 106 00:05:25,200 --> 00:05:27,120 Speaker 1: crisis will be much much worse for the value is 107 00:05:27,160 --> 00:05:29,400 Speaker 1: AI companies made up more than fifty percent of venture 108 00:05:29,400 --> 00:05:32,600 Speaker 1: capital investments in twenty twenty five. I actually don't know 109 00:05:32,640 --> 00:05:34,560 Speaker 1: how any of them make it. This is something that 110 00:05:34,600 --> 00:05:39,680 Speaker 1: I think about a lot. My thesis basically says, these 111 00:05:39,720 --> 00:05:43,120 Speaker 1: companies are all dying. I don't see how it works out. 112 00:05:43,320 --> 00:05:45,880 Speaker 1: They're not getting acquired. They can't go public because they 113 00:05:45,880 --> 00:05:48,120 Speaker 1: have the worst economics of all time. In the moment 114 00:05:48,120 --> 00:05:51,120 Speaker 1: you show the markets that it looks real bad. Minimax 115 00:05:51,520 --> 00:05:54,200 Speaker 1: AI company in China that went public, I think it's 116 00:05:54,240 --> 00:05:57,159 Speaker 1: fifty something billion dollars of revenue and like two hundred 117 00:05:57,160 --> 00:05:59,839 Speaker 1: and something million and losses a little bit like that 118 00:06:00,000 --> 00:06:03,440 Speaker 1: make the real money. Nevertheless, I really don't know how 119 00:06:03,440 --> 00:06:06,279 Speaker 1: that works. I truly don't. I didn't think about it. 120 00:06:06,320 --> 00:06:07,920 Speaker 1: I'm like, is there a way they could get acquired? 121 00:06:08,000 --> 00:06:09,800 Speaker 1: Is there something else they could do? There really isn't 122 00:06:09,839 --> 00:06:11,400 Speaker 1: all they can do is check up their prices and 123 00:06:11,440 --> 00:06:14,360 Speaker 1: hope that people don't leave. And I think that's what's 124 00:06:14,360 --> 00:06:16,800 Speaker 1: gonna happen, if they even bother, and if they can 125 00:06:16,839 --> 00:06:20,680 Speaker 1: even get that far. Now, I know what you're all thinking, 126 00:06:20,760 --> 00:06:22,440 Speaker 1: and I hear this a lot. I'm kind of died 127 00:06:22,480 --> 00:06:24,279 Speaker 1: of hearing it, but I'm honest and I get why 128 00:06:24,320 --> 00:06:28,520 Speaker 1: people do it. But despite the subprime comparison, this is 129 00:06:28,640 --> 00:06:31,320 Speaker 1: not a too big to fail situation. And in this 130 00:06:31,360 --> 00:06:33,520 Speaker 1: week's premium newsletter, I'm going to dig into that question 131 00:06:33,520 --> 00:06:36,480 Speaker 1: in depth. Please do subscribe. If the money goes directly 132 00:06:36,520 --> 00:06:38,880 Speaker 1: to me, it's a main source of income now that 133 00:06:39,320 --> 00:06:39,960 Speaker 1: it's very. 134 00:06:39,839 --> 00:06:40,440 Speaker 2: Important to me. 135 00:06:40,560 --> 00:06:42,600 Speaker 1: But nevertheless, I'm going to give you a sizeable preview 136 00:06:42,640 --> 00:06:56,320 Speaker 1: here because I want this information out there. So too 137 00:06:56,360 --> 00:06:59,440 Speaker 1: Big to Fail refers to the Troubled Asset Relief Program 138 00:06:59,520 --> 00:07:02,680 Speaker 1: TOP over four hundred billion dollars and specifically the bailouts 139 00:07:02,720 --> 00:07:05,719 Speaker 1: of AIG in the surrounding finance industry, mostly focused on 140 00:07:05,720 --> 00:07:08,640 Speaker 1: the commercial paper loan industry that kept major banks and 141 00:07:08,920 --> 00:07:12,440 Speaker 1: financial institutions going as well as fucking ge capital. Oh 142 00:07:12,480 --> 00:07:16,800 Speaker 1: and also the PDCF and ts LF finance systems that 143 00:07:16,880 --> 00:07:18,520 Speaker 1: kept them going to I'll get to that. 144 00:07:18,480 --> 00:07:21,360 Speaker 2: In a bit, but the bailout was in the trillions. 145 00:07:21,440 --> 00:07:24,200 Speaker 1: Like I don't think people realize how and why it happened, 146 00:07:24,240 --> 00:07:26,840 Speaker 1: so I'm going to tell you. Anyway, AIG was too 147 00:07:26,840 --> 00:07:28,880 Speaker 1: big to fail because it had twenty billion dollars in 148 00:07:28,920 --> 00:07:32,320 Speaker 1: outstanding commercial paper and several life insurance subsidiaries that were 149 00:07:32,320 --> 00:07:35,400 Speaker 1: billions of dollars in the hole thanks to AIG literally 150 00:07:35,480 --> 00:07:38,760 Speaker 1: gambling with the money, not actually at casinos, on cds's 151 00:07:38,760 --> 00:07:42,320 Speaker 1: CDOs and the like. The knock on effects of AIG's 152 00:07:42,360 --> 00:07:44,920 Speaker 1: collapse would have been catastrophic for the money market funds 153 00:07:44,920 --> 00:07:47,840 Speaker 1: that held commercial paper from multiple companies. And commercial paper 154 00:07:47,880 --> 00:07:51,080 Speaker 1: is a short term loan type thing, anywhere from one 155 00:07:51,120 --> 00:07:53,680 Speaker 1: to four days to like I think two hundred and 156 00:07:53,800 --> 00:07:57,119 Speaker 1: something days, but most of it was either very short 157 00:07:57,200 --> 00:08:01,200 Speaker 1: term and in AIG's case, was also without any collateral 158 00:08:01,400 --> 00:08:03,280 Speaker 1: in any way. In any case, that used to be 159 00:08:03,320 --> 00:08:06,560 Speaker 1: the primary way that banks and finance institutions g capital 160 00:08:06,640 --> 00:08:09,440 Speaker 1: actually used to be able to fund their businesses. They 161 00:08:09,440 --> 00:08:12,480 Speaker 1: would go to the market and go, hey, well, why 162 00:08:12,520 --> 00:08:15,000 Speaker 1: you lend us some money just for like a few days, 163 00:08:15,040 --> 00:08:16,480 Speaker 1: and we'll get right back to you, and for the 164 00:08:16,480 --> 00:08:19,160 Speaker 1: most part. It worked right up until it didn't. I 165 00:08:19,200 --> 00:08:22,520 Speaker 1: also want to be clear about what the bailout actually 166 00:08:22,800 --> 00:08:26,080 Speaker 1: bailed out, because I think people here too big to 167 00:08:26,120 --> 00:08:28,160 Speaker 1: fail and they're like, right, they bought the houses that 168 00:08:28,200 --> 00:08:30,760 Speaker 1: were being foreclosed. No, no, no, that's not what they did. 169 00:08:31,000 --> 00:08:33,120 Speaker 1: It was like forty something billion dollars that was meant 170 00:08:33,160 --> 00:08:35,720 Speaker 1: to go to stop foreclosures. No one really knows what 171 00:08:35,800 --> 00:08:39,000 Speaker 1: happened to it, which is cool. No, the vast majority 172 00:08:39,000 --> 00:08:43,199 Speaker 1: of the bailouts were for financial instruments, not houses, not apartments, 173 00:08:43,280 --> 00:08:46,360 Speaker 1: not helping regular people, but making sure that the toxic 174 00:08:46,400 --> 00:08:50,440 Speaker 1: financial products associated with the finance industry were absorbed and 175 00:08:50,480 --> 00:08:53,040 Speaker 1: bought off and that liquidity remained in the market via 176 00:08:53,120 --> 00:08:56,920 Speaker 1: the primary dealer credit and term securities lending facilities that 177 00:08:57,000 --> 00:08:59,680 Speaker 1: provided as much as one hundred billion dollars to banks 178 00:08:59,679 --> 00:09:03,040 Speaker 1: and financial institutions a day. Now. 179 00:09:03,080 --> 00:09:03,800 Speaker 2: These still exist. 180 00:09:03,920 --> 00:09:06,440 Speaker 1: The repo facilities still exists, but are not used at 181 00:09:06,440 --> 00:09:08,760 Speaker 1: the same scale. But this, I want you to think 182 00:09:08,760 --> 00:09:11,440 Speaker 1: about this, like every day, fifty one hundred billion dollars 183 00:09:11,520 --> 00:09:13,760 Speaker 1: was just used to keep the to give them short 184 00:09:13,840 --> 00:09:16,720 Speaker 1: term funding to just get through the day. People should 185 00:09:16,720 --> 00:09:19,640 Speaker 1: have been in fucking prison to be clear, people should 186 00:09:19,640 --> 00:09:22,640 Speaker 1: have gone to jail for this, but this was necessary 187 00:09:22,840 --> 00:09:27,040 Speaker 1: to stop the financial system crashing to actually falling apart. 188 00:09:27,200 --> 00:09:30,400 Speaker 1: Lending insurance would not have got funded in the same way. 189 00:09:30,440 --> 00:09:33,400 Speaker 1: There was basically a credit freeze anyway, but I'm talking 190 00:09:34,240 --> 00:09:39,200 Speaker 1: nothing would have happened. And even during the Great Financial Crisis, 191 00:09:39,320 --> 00:09:41,920 Speaker 1: mortgages still were getting written, loans were still getting rerint. 192 00:09:42,280 --> 00:09:46,439 Speaker 1: Things still happened, and the financial system is a confidence game, 193 00:09:46,480 --> 00:09:50,000 Speaker 1: and the PDCF and TSLF existed to stop everything from 194 00:09:50,040 --> 00:09:52,800 Speaker 1: actually going to zero. And like I said, those few 195 00:09:52,840 --> 00:09:56,080 Speaker 1: loans were still being written and funded, and they needed 196 00:09:56,120 --> 00:09:59,800 Speaker 1: to make sure that insurance and borrowing they still actually 197 00:09:59,840 --> 00:10:02,199 Speaker 1: had happened because the world runs on debt and insurance. 198 00:10:02,960 --> 00:10:06,040 Speaker 1: And oh my god, this really reading about the finance 199 00:10:06,160 --> 00:10:08,480 Speaker 1: the Great Financial Crisis really did blackpilm me all over 200 00:10:08,520 --> 00:10:10,320 Speaker 1: again though, because you read what these fuck nuts did. 201 00:10:10,600 --> 00:10:12,839 Speaker 1: They were just they were like, yeah, we're gonna hedge 202 00:10:12,840 --> 00:10:16,520 Speaker 1: our bets on the on the mortgages by buying credit 203 00:10:16,559 --> 00:10:19,760 Speaker 1: default swaps that were not that we assume that aig 204 00:10:19,880 --> 00:10:23,200 Speaker 1: will be able to pay. Don't need to check. Well 205 00:10:23,240 --> 00:10:25,920 Speaker 1: they should have checked. They can fucking bet anyway. Some 206 00:10:26,120 --> 00:10:28,840 Speaker 1: estimate the true size of the bailout when you account 207 00:10:28,840 --> 00:10:31,040 Speaker 1: for that liquidity, to be over ten trillion dollars. Because 208 00:10:31,080 --> 00:10:35,360 Speaker 1: the underdiscussed part of the AIG bailout was that commercial paper, 209 00:10:35,679 --> 00:10:38,000 Speaker 1: the commercial paper lending industry that funded most of the 210 00:10:38,000 --> 00:10:40,600 Speaker 1: finance industry kind of broke. 211 00:10:40,960 --> 00:10:41,600 Speaker 2: It took years. 212 00:10:41,640 --> 00:10:45,439 Speaker 1: The finance industry actually had to undwind their dependence on it. 213 00:10:45,520 --> 00:10:49,240 Speaker 1: Colgate used to use commercial paper. It's fucking weird. In 214 00:10:49,280 --> 00:10:52,600 Speaker 1: any case, there really aren't any comparables to the AI bubble. 215 00:10:52,760 --> 00:10:55,440 Speaker 1: While open AI and Anthropic might be very prominent and 216 00:10:55,559 --> 00:10:59,000 Speaker 1: very annoying, their actual economic existence is relatively small. And 217 00:10:59,080 --> 00:11:01,840 Speaker 1: the overall AI in industry barely had sixty five billion 218 00:11:01,880 --> 00:11:04,000 Speaker 1: dollars in revenue last year, with much of that flowing 219 00:11:04,000 --> 00:11:06,840 Speaker 1: between a few counterparties and funded by venture capital dollars. 220 00:11:07,240 --> 00:11:09,640 Speaker 1: And also we don't know if all of that was cash. 221 00:11:09,679 --> 00:11:12,360 Speaker 1: By the way, a chunk of that is the inference 222 00:11:12,400 --> 00:11:15,320 Speaker 1: spend with Microsoft. That if that still tokens. If you 223 00:11:15,800 --> 00:11:18,120 Speaker 1: work at Microsoft and you want to talk to me 224 00:11:18,160 --> 00:11:21,840 Speaker 1: about the finances or any of these companies, please please 225 00:11:21,960 --> 00:11:23,600 Speaker 1: hit me up. I would love to hear from you. 226 00:11:23,640 --> 00:11:28,559 Speaker 1: Easy at Better Offline dot Com. Nevertheless, had AIG defaulted, 227 00:11:28,720 --> 00:11:30,960 Speaker 1: it would have left multiple banks without the funds to 228 00:11:31,000 --> 00:11:34,559 Speaker 1: continue functioning and killed its insurance companies in the process, 229 00:11:34,600 --> 00:11:37,840 Speaker 1: sending them into government receivership that wouldn't have guaranteed the 230 00:11:37,840 --> 00:11:41,480 Speaker 1: policies had the same value even have been fulfilled. Even 231 00:11:41,559 --> 00:11:43,680 Speaker 1: with that bailout, the government had to plug a hole 232 00:11:43,679 --> 00:11:46,360 Speaker 1: in the side of the finance industry for several years, 233 00:11:46,360 --> 00:11:48,600 Speaker 1: and they ended up bring it back during COVID as well. 234 00:11:49,440 --> 00:11:51,880 Speaker 1: While the collapse of open AI and Anthropic might tank 235 00:11:51,920 --> 00:11:54,920 Speaker 1: parts of the market, too big to fail is a 236 00:11:55,040 --> 00:11:57,440 Speaker 1: term that refers to something that has systemic risk to 237 00:11:57,480 --> 00:12:01,120 Speaker 1: the economy, and these are two very different things. Because 238 00:12:01,120 --> 00:12:03,680 Speaker 1: the market still shat its pants during the great financial 239 00:12:03,720 --> 00:12:07,160 Speaker 1: crisis that happened. The government wasn't stopping the markets from 240 00:12:07,160 --> 00:12:11,079 Speaker 1: collapsing by funding the markets. They were keeping the banking 241 00:12:11,120 --> 00:12:14,920 Speaker 1: system alive, so the markets just functioned at all. And 242 00:12:15,000 --> 00:12:16,959 Speaker 1: I really need you to know that difference, because it's 243 00:12:17,000 --> 00:12:19,319 Speaker 1: not the same thing. The collapse of AIG would have 244 00:12:19,400 --> 00:12:21,840 Speaker 1: quickly ripped through the funding mechanisms of most of the 245 00:12:21,960 --> 00:12:25,720 Speaker 1: US finance industry, and it's collapse still materially harmed one 246 00:12:25,760 --> 00:12:29,240 Speaker 1: of the main funding mechanisms of the US economy. There 247 00:12:29,280 --> 00:12:32,760 Speaker 1: is no such comparison with AI. AI is not producing 248 00:12:32,800 --> 00:12:36,080 Speaker 1: productivity benefits at scale. It is not replacing jobs at scale, 249 00:12:36,200 --> 00:12:38,080 Speaker 1: not that I want that to happen, of course, It's 250 00:12:38,080 --> 00:12:40,520 Speaker 1: not load bearing in an economic sense, in that the 251 00:12:40,559 --> 00:12:43,120 Speaker 1: flow of money in the US economy would collapse if 252 00:12:43,160 --> 00:12:46,760 Speaker 1: open AI or Anthropic did, and a bailout in general 253 00:12:46,800 --> 00:12:49,199 Speaker 1: exists to solve a problem rather than pass the bug. 254 00:12:49,360 --> 00:12:51,560 Speaker 1: And I also want to be clear if that if 255 00:12:51,640 --> 00:12:55,240 Speaker 1: they were truly too big to fail, Ah, no, and no, 256 00:12:55,320 --> 00:12:58,920 Speaker 1: wouldn't it wouldn't it be that when Anthropic goes down 257 00:12:59,040 --> 00:13:04,120 Speaker 1: or chat GPT goes down, the economy would stop, because 258 00:13:04,160 --> 00:13:06,120 Speaker 1: that's what would have happened if AIG died. 259 00:13:07,120 --> 00:13:10,440 Speaker 2: Goddamn have I nearly said AGI like five times? Anyway. 260 00:13:10,960 --> 00:13:12,800 Speaker 1: I also want to address that Fanny May and Freddie 261 00:13:12,840 --> 00:13:15,880 Speaker 1: Mack massive mortgage companies that were absorbed by the US government. 262 00:13:16,120 --> 00:13:18,880 Speaker 1: They're not a comparison as the deaths of though those 263 00:13:18,920 --> 00:13:21,960 Speaker 1: companies would have actually destroyed the US housing market like 264 00:13:22,280 --> 00:13:24,040 Speaker 1: they were like sixty five billion dollars in the whole. 265 00:13:24,080 --> 00:13:25,600 Speaker 2: It was very very very. 266 00:13:25,440 --> 00:13:28,480 Speaker 1: Bad, Like they I need you like The Big Short 267 00:13:28,520 --> 00:13:30,920 Speaker 1: is a great movie, it does not do justice to 268 00:13:30,960 --> 00:13:33,400 Speaker 1: how fucked things would have been. I also want to 269 00:13:33,400 --> 00:13:36,600 Speaker 1: be clear that none of these companies should have been 270 00:13:36,640 --> 00:13:38,800 Speaker 1: allowed to function in the same way, like any of 271 00:13:38,840 --> 00:13:41,520 Speaker 1: the banks involved in this should have like none of 272 00:13:41,559 --> 00:13:43,480 Speaker 1: the executives in questions should be allowed to work in 273 00:13:43,520 --> 00:13:45,920 Speaker 1: finance anymore. I think that it was genuinely evil and 274 00:13:45,960 --> 00:13:50,080 Speaker 1: the whole situation was caused by an industry wide just 275 00:13:50,760 --> 00:13:51,800 Speaker 1: gambling industry. 276 00:13:51,840 --> 00:13:52,520 Speaker 2: That's what it was. 277 00:13:52,600 --> 00:13:56,520 Speaker 1: It would It was truly horrifying, and everyone involved should 278 00:13:56,559 --> 00:13:58,079 Speaker 1: be in jail or jobless. 279 00:13:58,160 --> 00:14:00,320 Speaker 2: I think they're fucking awful. Nevertheles less. 280 00:14:00,320 --> 00:14:02,640 Speaker 1: It was necessary to do because of the fuck with 281 00:14:03,360 --> 00:14:06,480 Speaker 1: nature of the whole thing, because of the stupid gambling 282 00:14:06,520 --> 00:14:08,840 Speaker 1: they did, because of the low reserves they had, and 283 00:14:08,880 --> 00:14:12,600 Speaker 1: how brittle everything was, which is their fault. That's very 284 00:14:12,640 --> 00:14:16,920 Speaker 1: different to the bailout not being necessary. Though, now getting 285 00:14:16,960 --> 00:14:19,800 Speaker 1: back to AI. Without open AI and anthropic the AI 286 00:14:19,880 --> 00:14:22,600 Speaker 1: industry would disappear, as with the demand for AI compume. 287 00:14:22,840 --> 00:14:25,040 Speaker 1: It would lead to billions of dollars of loans going unpaid, 288 00:14:25,080 --> 00:14:27,000 Speaker 1: and the value of venture capital would probably be cut 289 00:14:27,040 --> 00:14:29,160 Speaker 1: in half or as much as eighty or ninety percent 290 00:14:29,200 --> 00:14:30,560 Speaker 1: has happened with the dot com bubble. 291 00:14:31,000 --> 00:14:32,600 Speaker 2: It'd be horrible for the markets core. 292 00:14:32,600 --> 00:14:35,760 Speaker 1: We would die, iron would die, and heaviest would die. 293 00:14:36,200 --> 00:14:38,640 Speaker 1: I think the open A and anthropics death would end 294 00:14:38,720 --> 00:14:41,240 Speaker 1: up ripping through the entirety of silicon value. I think 295 00:14:41,480 --> 00:14:44,120 Speaker 1: that it would be a payback for years of focusing 296 00:14:44,160 --> 00:14:47,640 Speaker 1: on growth rather than creating anything. On the banking side, 297 00:14:47,760 --> 00:14:50,160 Speaker 1: I do think we are somewhat safer. I think while 298 00:14:50,160 --> 00:14:53,120 Speaker 1: there aren't heavy reserve requirements, there are actual regulations around 299 00:14:53,320 --> 00:14:57,480 Speaker 1: high risk speculation. And the scale of speculation here is 300 00:14:57,560 --> 00:15:00,000 Speaker 1: ten tiny compared to I think it's like five trillion 301 00:15:00,080 --> 00:15:03,040 Speaker 1: and synthetic CEOs or something of that nature. There aren't 302 00:15:03,080 --> 00:15:07,400 Speaker 1: trillions of dollars of speculation here, billions, hundreds of billions. Sure, 303 00:15:07,560 --> 00:15:11,040 Speaker 1: And I'm fairly sure, fairly confident that a lot of 304 00:15:11,080 --> 00:15:13,640 Speaker 1: these banks and private equity firms and private credit firms 305 00:15:13,720 --> 00:15:16,840 Speaker 1: have some degree of reserves built in for these loans 306 00:15:16,920 --> 00:15:20,080 Speaker 1: going tits up. In the case of AI Data centers, 307 00:15:21,320 --> 00:15:24,680 Speaker 1: their debts, these AI companies debts wouldn't create a calamity 308 00:15:24,680 --> 00:15:26,960 Speaker 1: that would stop loans in general, or insurance in general 309 00:15:27,000 --> 00:15:29,760 Speaker 1: from being collateralized or any other basic functions of the 310 00:15:29,840 --> 00:15:33,000 Speaker 1: US economy. I'll also add the Nvidia and the rest 311 00:15:33,000 --> 00:15:34,560 Speaker 1: of the Magnificent Seven are going to die as a 312 00:15:34,600 --> 00:15:36,760 Speaker 1: result of the AI bubble bursting. They're not going to 313 00:15:36,840 --> 00:15:41,000 Speaker 1: need bailouts. I think oracles bailout chances are there, But 314 00:15:41,120 --> 00:15:44,000 Speaker 1: even then I'm kind of hesitant to say so, because 315 00:15:45,480 --> 00:15:47,440 Speaker 1: I don't know what they'd be better. Maybe bailing out 316 00:15:47,480 --> 00:15:50,440 Speaker 1: the loans they create a little Larry Ellison's shaped top 317 00:15:50,720 --> 00:15:53,000 Speaker 1: just for that fuck nut, But I actually don't know 318 00:15:53,040 --> 00:15:56,560 Speaker 1: if that would happen. Top was deeply unprop it was 319 00:15:56,600 --> 00:15:59,840 Speaker 1: deeply unpopular, deeply deeply unpopular. And Trump's already having enough. 320 00:16:00,440 --> 00:16:03,239 Speaker 1: I'm sure he's doing that. We're touching the stove competition. 321 00:16:03,800 --> 00:16:06,160 Speaker 1: But at the same time, I just don't see it 322 00:16:06,200 --> 00:16:08,520 Speaker 1: happening in the same way. And even then, I still 323 00:16:08,560 --> 00:16:11,320 Speaker 1: think Oracle would die and get absorbed into another company, 324 00:16:11,320 --> 00:16:14,200 Speaker 1: probably Microsoft. Wouldn't that be funny, man? The anti trust 325 00:16:14,200 --> 00:16:18,320 Speaker 1: would be crazy. That being set, I think that the 326 00:16:18,480 --> 00:16:21,359 Speaker 1: AI bubble will eventually be seen as a smaller precursor, 327 00:16:21,440 --> 00:16:23,160 Speaker 1: much like the dot com bubble was to the Great 328 00:16:23,160 --> 00:16:26,760 Speaker 1: Financial Crisis, to a larger financial crisis created by private 329 00:16:26,760 --> 00:16:28,760 Speaker 1: credit and private equity, and there are hundreds of billions 330 00:16:28,760 --> 00:16:32,640 Speaker 1: of dodgy investments in software and random companies. I am 331 00:16:32,720 --> 00:16:35,000 Speaker 1: still working out how bad this would be, and I 332 00:16:35,000 --> 00:16:36,800 Speaker 1: don't want to be an alarmist. I actually don't think 333 00:16:36,800 --> 00:16:38,640 Speaker 1: it will be as bad as the Great Financial Crisis, 334 00:16:38,640 --> 00:16:40,160 Speaker 1: but that is a question I'm going to try and 335 00:16:40,240 --> 00:16:43,320 Speaker 1: answer in the future. An open AI and anthropic aren't 336 00:16:43,320 --> 00:16:45,800 Speaker 1: getting bailed out. It's not happening. I don't care if 337 00:16:45,800 --> 00:16:48,920 Speaker 1: you think, oh, the Department of Events uses them, They're 338 00:16:48,960 --> 00:16:50,440 Speaker 1: not going to bail them out for that reason, They're 339 00:16:50,440 --> 00:16:52,600 Speaker 1: not going to put them into conservativeship. It's not going 340 00:16:52,680 --> 00:16:55,240 Speaker 1: to happen. It's not It would be deeply unpopular, and 341 00:16:55,360 --> 00:16:58,680 Speaker 1: also to what end a festering hole in the side 342 00:16:58,680 --> 00:17:03,480 Speaker 1: of the government. You already have Congress anyway. AI data 343 00:17:03,520 --> 00:17:06,600 Speaker 1: centers aren't getting bailed out either. I really need to 344 00:17:06,600 --> 00:17:08,919 Speaker 1: be clear about that. Houses were not bailed out for 345 00:17:08,960 --> 00:17:13,000 Speaker 1: closures will not bailed out. Perhaps the ABSs might have 346 00:17:13,119 --> 00:17:16,000 Speaker 1: some hope the asset back securities, but even then I 347 00:17:16,000 --> 00:17:18,720 Speaker 1: don't think the scale of those is anywhere close to 348 00:17:18,760 --> 00:17:21,800 Speaker 1: the Great Financial Crisis. And to be clear, the collapse 349 00:17:21,800 --> 00:17:25,080 Speaker 1: of the AI industry will fucking hurt. It will brutalize 350 00:17:25,080 --> 00:17:27,800 Speaker 1: the banks in private credit firms involved. It will fuck 351 00:17:27,920 --> 00:17:31,280 Speaker 1: venture capital for years, if not over a decade. Earning 352 00:17:31,320 --> 00:17:34,080 Speaker 1: seasons are going to look like the dog from John 353 00:17:34,119 --> 00:17:36,440 Speaker 1: Carpenter's The Thing. It's going to be horrible for them 354 00:17:36,440 --> 00:17:38,880 Speaker 1: and horrible for people invested in the market. It's not 355 00:17:38,960 --> 00:17:41,480 Speaker 1: going to be nice. I am not saying that it's 356 00:17:41,520 --> 00:17:45,399 Speaker 1: not going to hurt, but that is meaningfully different to 357 00:17:45,480 --> 00:17:47,840 Speaker 1: anything being too big to fail. And I'm tired of 358 00:17:47,880 --> 00:17:50,960 Speaker 1: people saying things without actually bothering to understand what they mean.