WEBVTT - Carson Block: AI Could Trigger the Next Financial Crisis 0:00:02.720 --> 0:00:16.680 Bloomberg Audio Studios, Podcasts, Radio News. 0:00:19.000 --> 0:00:21.040 Welcome to Marin Talk to Money, the podcast in which 0:00:21.079 --> 0:00:23.840 people who know the market explain the market. I am 0:00:23.920 --> 0:00:26.560 Maren zum zep Web and this week I am speaking 0:00:26.600 --> 0:00:29.520 with well known short seller and Muddy Waters Capital founder 0:00:29.560 --> 0:00:31.400 and CEO Carson Block. 0:00:31.720 --> 0:00:34.080 Carson, Welcome to Marin Talks Money. 0:00:34.120 --> 0:00:35.360 Thank you, thanks for having me. 0:00:35.640 --> 0:00:38.920 You have been talking a lot recently about AI, and 0:00:38.960 --> 0:00:42.120 I wanted to just start on that straight away because 0:00:42.159 --> 0:00:45.040 I think the core of your view is that AI 0:00:45.360 --> 0:00:48.479 really is going to live up to everyone's expectations of 0:00:48.560 --> 0:00:52.080 it. It is going to be spectacularly successful. All the worries 0:00:52.120 --> 0:00:54.240 a lot of people have about the way that llms 0:00:54.240 --> 0:00:56.720 have hit his ceiling and development is over and we 0:00:56.760 --> 0:00:57.480 need a new path. 0:00:57.560 --> 0:00:59.320 All this kind of thing doesn't come into your view. 0:01:00.000 --> 0:01:03.120 I think that it's going to be very successful and 0:01:03.360 --> 0:01:06.360 there will be an employment apocalypse as a result. 0:01:06.959 --> 0:01:08.440 Am I over egging of you there? 0:01:08.920 --> 0:01:13.200 Let me just state that as background, I was a skeptic. 0:01:13.240 --> 0:01:17.120 So until February of this year, I refused to refer 0:01:17.360 --> 0:01:20.880 to any of the models that are out there that 0:01:20.920 --> 0:01:25.200 we're using as AI. They are large language models, and 0:01:25.280 --> 0:01:27.280 so I said, Okay, these things are not going to 0:01:27.360 --> 0:01:30.959 ever find the cure for cancer. They are not that intelligent. 0:01:31.800 --> 0:01:38.759 It's not artificial intelligence. They're lllms. Now, the current generation 0:01:39.319 --> 0:01:44.440 of LMS that was released in January early February of 0:01:44.440 --> 0:01:49.960 this year changed my mind to some extent. They're not 0:01:50.080 --> 0:01:54.320 going to independently develop the cure for cancers. They I 0:01:54.320 --> 0:01:57.360 mean I used to term them the greatest search engines 0:01:57.400 --> 0:02:03.120 ever developed, and that effectively they are. But the ability, 0:02:03.240 --> 0:02:08.120 their ability to synthesize that information and order that information, 0:02:10.280 --> 0:02:13.320 I mean there's real labor savings to be had right 0:02:13.360 --> 0:02:17.799 now in certain areas of knowledge industries I mean such 0:02:17.840 --> 0:02:20.760 as investment. I mean I immediately began using Claude to 0:02:20.840 --> 0:02:25.680 run Monte Carlo simulations that I previously would have needed 0:02:25.680 --> 0:02:29.320 somebody who was pretty young, recently graduated from a top 0:02:29.440 --> 0:02:33.880 university strong in math to run the thing that's important 0:02:34.840 --> 0:02:37.640 to really understanding or that change to change that changed 0:02:37.639 --> 0:02:41.880 my view as to what lllms are going to do 0:02:42.080 --> 0:02:46.519 to the labor market. Is that this generation was coded 0:02:46.600 --> 0:02:50.520 and tested in large part by its predecessor generation. It 0:02:50.600 --> 0:02:55.440 is coding and testing its successor generation. So you have 0:02:55.520 --> 0:03:00.240 a situation where the leading users of this generation of 0:03:00.280 --> 0:03:03.600 AI model, I mean they're in the technology industry. They 0:03:03.720 --> 0:03:11.160 are able to supplant by using AI or displace multiple 0:03:11.200 --> 0:03:14.919 members of a team. Now, vast majority of us are 0:03:14.919 --> 0:03:18.120 not that skilled yet with the current generation of model. 0:03:18.280 --> 0:03:22.240 But we extrapolate that we're going to be hitting a 0:03:22.360 --> 0:03:25.200 point in the not too distant future in which we're 0:03:25.240 --> 0:03:28.840 going to see exponential improvement in the capabilities of these models. 0:03:28.960 --> 0:03:32.680 Because they are coding and testing their successors and so 0:03:32.720 --> 0:03:35.320 on and so forth. Then you're going to see a 0:03:35.320 --> 0:03:39.360 lot more capability. You're going to see more comfort in 0:03:39.600 --> 0:03:43.600 understanding of how to use them. I think it's entirely 0:03:43.680 --> 0:03:50.120 conceivable that within a few years three five two, you 0:03:50.160 --> 0:03:54.160 could see within the US roughly fifteen percent of knowledge 0:03:54.160 --> 0:04:00.240 workers displaced from their jobs. And unlike the GFS, see 0:04:00.520 --> 0:04:03.840 it's not that you just need to grow your way 0:04:03.880 --> 0:04:07.280 out of this and reallocate human and financial capital. It's 0:04:07.640 --> 0:04:11.240 you know, those jobs are gone and the chairs, the 0:04:11.280 --> 0:04:14.440 number of chairs for humans is going to continue shrinking. 0:04:15.000 --> 0:04:21.239 So that's that's where I come out on on AI, 0:04:21.560 --> 0:04:24.480 l ll ms, you know, whatever you want to call them. 0:04:24.720 --> 0:04:27.880 Yeah, okay, so you lose a lot of jobs, but 0:04:28.480 --> 0:04:33.040 also the remaining jobs, the quality of them degrades or 0:04:33.200 --> 0:04:33.800 feels like. 0:04:33.800 --> 0:04:39.480 That's the end game, that these knowledge jobs disappear, and 0:04:39.640 --> 0:04:42.080 if people can get other jobs, they might not be 0:04:42.160 --> 0:04:44.360 of the same quality and certainly not at the same 0:04:44.400 --> 0:04:45.040 income level. 0:04:46.279 --> 0:04:48.000 I mean, a number of these jests, a number of 0:04:48.000 --> 0:04:52.680 these jobs will disappear. I'm a former attorney, I'm a 0:04:52.720 --> 0:04:58.119 professional litigant effectively now and my current you know current, 0:04:58.400 --> 0:05:01.520 you know, iteration of my career. And I got to 0:05:01.560 --> 0:05:07.240 tell you, like, at present, claud can handle a lot 0:05:07.320 --> 0:05:11.159 of tasks that I pay attorneys for. Now. We're all 0:05:11.160 --> 0:05:14.200 familiar with the hallucination it makes up, you know, A 0:05:14.360 --> 0:05:18.279 versus B. I'm not saying you remove humans entirely, but 0:05:18.520 --> 0:05:21.080 I mean, if i'm you know, when you go through discovery, 0:05:21.480 --> 0:05:24.559 you're paying some associate. I mean, depending on the law firm, 0:05:24.960 --> 0:05:28.480 you're paying some associate somewhere between. You know, for good 0:05:28.560 --> 0:05:31.640 law firm, junior associate five hundred dollars an hour to 0:05:31.960 --> 0:05:36.640 fifteen hundred dollars an hour to review email after email 0:05:36.880 --> 0:05:43.640 and text after text, and they make mistakes. Claude does 0:05:43.680 --> 0:05:46.720 it one hundred percent better. So now you have to 0:05:46.760 --> 0:05:48.920 ask you have to query it in the right way. 0:05:49.320 --> 0:05:52.039 But you know, when you're talking about I mean financial 0:05:52.080 --> 0:05:57.040 service related litigation, most of our outside attorneys, especially junior associates, 0:05:57.080 --> 0:06:00.400 don't understand what we do anyway. So I think you're 0:06:00.480 --> 0:06:04.600 much more effective having the client running these you know, 0:06:04.720 --> 0:06:09.600 running the discovery in claude and querying in querying claude 0:06:09.600 --> 0:06:13.400 at least to make your arguments and that's just saved. 0:06:14.120 --> 0:06:18.120 That will save innumerable billable hours. So that's a profession 0:06:18.800 --> 0:06:22.479 where that I think is going to shrink significantly in 0:06:22.560 --> 0:06:27.599 the coming several years. And that's a very highly paid profession. 0:06:28.000 --> 0:06:28.160 Yeah. 0:06:28.240 --> 0:06:31.159 The problem I suppose with that is that you you 0:06:31.320 --> 0:06:33.680 need to train the young people to have the more 0:06:33.720 --> 0:06:38.559 experienced people available to put in the correct queries. So 0:06:38.760 --> 0:06:43.040 once your once your pipeline of educating young people disappears 0:06:43.600 --> 0:06:46.159 or shrinks significantly, end up with a problem higher up 0:06:46.200 --> 0:06:47.640 the tree. 0:06:47.920 --> 0:06:52.440 Yeah, but that's I haven't seen businesses really think too 0:06:52.480 --> 0:06:55.240 long term about you know there, you know for a 0:06:55.320 --> 0:06:58.400 long time, right, I mean, everybody's focused on, you know, 0:06:58.440 --> 0:07:00.560 how much am I going to make this year? Maybe 0:07:00.560 --> 0:07:03.920 they're thinking next year. So and the problem, you know, look, 0:07:03.960 --> 0:07:07.280 one of the pushbacks to this idea that they're going 0:07:07.320 --> 0:07:11.000 to see the level of labor displacements that you know, 0:07:11.160 --> 0:07:12.960 I think you're going to see is that, oh, well, 0:07:13.040 --> 0:07:16.480 the technology the adoption curve. You know, technologies are never 0:07:16.560 --> 0:07:20.200 adopted at the rate at which they could theoretically be adopted. 0:07:20.240 --> 0:07:23.400 And yeah, that's true. But the thing here is that 0:07:23.680 --> 0:07:28.760 the cost savings for you know, for the the providers 0:07:28.760 --> 0:07:30.560 of these services, the cost savings are going to be 0:07:30.560 --> 0:07:33.880 so significant they're going to be able to massively underprice 0:07:33.920 --> 0:07:37.080 their competitors. So you know, it's going to be an 0:07:37.120 --> 0:07:42.240 existential problem for businesses that don't take advantage of this 0:07:42.320 --> 0:07:46.080 technology and reduce their cost and pricing structures accordingly. So 0:07:46.480 --> 0:07:49.600 that's why I think that the adoption curve argument here 0:07:50.360 --> 0:07:53.560 is not going to or counter argument is not going 0:07:53.600 --> 0:07:57.360 to win the day that there will actually be pretty 0:07:57.360 --> 0:07:59.560 steep uptake of these technologies. 0:08:00.600 --> 0:08:01.960 I mean, it hasn't happened yet. 0:08:02.160 --> 0:08:04.840 We haven't really seen this happening yet, you know, I mean, 0:08:05.200 --> 0:08:07.600 it's all still to come, right and that white collar workers, 0:08:07.640 --> 0:08:09.400 a number of white color workers in the US and 0:08:09.480 --> 0:08:12.800 in Europe is up since the first STEMP chat GPT 0:08:13.080 --> 0:08:15.080 was released and you. 0:08:15.040 --> 0:08:18.040 Know, there are more AI engineers than ever, etc. It's 0:08:18.680 --> 0:08:19.280 not yet a. 0:08:19.200 --> 0:08:22.280 Couple of things. Yeah, well, all right, So the current 0:08:22.320 --> 0:08:26.440 generation of these models, which was at least in January February, 0:08:26.800 --> 0:08:29.280 was the first This is the first generation where it's 0:08:29.600 --> 0:08:32.920 you know, I said, these are not idiotic. You know, 0:08:33.240 --> 0:08:37.440 they have flaws, they have limitations. But I mean previously, yeah, 0:08:37.520 --> 0:08:39.679 it was it was a running joke. But you know, 0:08:40.080 --> 0:08:43.400 what we extrapolated as humans is that the progress would 0:08:43.440 --> 0:08:45.920 be linear, and it's not linear. It's going to be 0:08:46.000 --> 0:08:51.640 exponential because of how they're able to develop their successors. Now, 0:08:52.480 --> 0:08:55.440 just a little pushback. I'm not normally a macro person. 0:08:55.559 --> 0:08:59.800 But one thing that you know, as far as the 0:09:00.080 --> 0:09:02.880 know the great job success in the US, a lot 0:09:02.880 --> 0:09:07.160 of that's been in healthcare. And you know, healthcare is 0:09:07.200 --> 0:09:12.400 a really interesting sector in the US because look, on 0:09:12.480 --> 0:09:15.520 one hand, okay, it's going to grow. And when you 0:09:15.559 --> 0:09:19.800 go back to two thousand, information technology and look the 0:09:19.840 --> 0:09:23.160 way that the you know, uh, you know BLS. Clayer, 0:09:23.320 --> 0:09:26.640 you know, classifies sectors as antiquated really should be updated. 0:09:26.679 --> 0:09:31.800 But information technology and telecommunications and healthcare are the only 0:09:31.840 --> 0:09:34.680 two sectors that have grown materially as a share of GDP, 0:09:36.480 --> 0:09:39.000 and healthcare has grown significantly more as a share of 0:09:39.040 --> 0:09:44.800 GDP than information technology. So, you know, is it driven 0:09:44.840 --> 0:09:49.560 by innovation some? Is it driven by demographics some, But 0:09:49.640 --> 0:09:52.520 a lot of it's driven really by the parasitic nature 0:09:53.080 --> 0:09:55.600 of this healthcare system, at least in the US that 0:09:56.679 --> 0:09:59.440 I mean, there's there's so much fraud, there's so much inefficiency. 0:09:59.559 --> 0:10:02.520 So look, if you say the US has a healthy economy, 0:10:02.520 --> 0:10:05.240 we have a healthy job market because we've added all 0:10:05.240 --> 0:10:09.200 these healthcare jobs, you know, I think that's a misinterpretation. 0:10:09.520 --> 0:10:13.880 So look, does that really go to whether AI or 0:10:14.000 --> 0:10:17.280 llms are going to make a meaningful dent in employment 0:10:17.320 --> 0:10:20.400 going forward? Now? Because as I said, we're really at 0:10:20.480 --> 0:10:24.040 that point now where you're starting to see how these 0:10:24.080 --> 0:10:28.920 can replace employees in certain situations, or at least prevent 0:10:28.960 --> 0:10:32.640 the hiring of employees, and the next generation will be 0:10:32.640 --> 0:10:33.880 even more powerful. 0:10:34.760 --> 0:10:37.320 Okay, well, well we'll wait and see what happened in 0:10:37.360 --> 0:10:39.880 five ten years if new jobs, new different types of 0:10:39.960 --> 0:10:42.319 jobs that created you know, this whole business of jobs 0:10:42.320 --> 0:10:45.280 we never thought would exist now ten years ago, job 0:10:45.320 --> 0:10:47.200 being ever thought existing this, So who knows what will 0:10:47.240 --> 0:10:49.920 happen in ten years. But let's take the base case, 0:10:49.960 --> 0:10:53.040 your base case here that fifteen percent of knowledge jobs disappear. Now, 0:10:53.080 --> 0:10:55.719 the interesting bit for the purposes of our listeners and 0:10:55.760 --> 0:10:56.640 we discuss. 0:10:56.240 --> 0:10:59.360 We are interesting podcasts, is what that does to the market. 0:10:59.440 --> 0:11:02.000 And I think that's where this gets interesting, isn't it. 0:11:03.000 --> 0:11:06.280 Yeah. And so again this has been a u turn 0:11:06.360 --> 0:11:09.040 for me because there was actually a podcast I did 0:11:09.040 --> 0:11:11.840 in January where I was talking about how we had 0:11:11.880 --> 0:11:16.440 developed this systematic momentum strategy and house at Muddy Waters, 0:11:16.520 --> 0:11:19.800 and you know, it's been compounding for us at north 0:11:19.840 --> 0:11:22.400 of seventy I mean even I think most recently about 0:11:22.520 --> 0:11:26.160 ninety percent per year. And I was completely sanguine about 0:11:26.160 --> 0:11:29.120 the outlook for the market and why. And you know, 0:11:29.160 --> 0:11:32.880 in the idea that momentum, as much as it shouldn't, 0:11:33.200 --> 0:11:35.040 you know, as much as the fact that the stock 0:11:35.080 --> 0:11:37.480 has gone up shouldn't determine that it'll continue to go on, 0:11:37.600 --> 0:11:38.120 but it does. 0:11:38.240 --> 0:11:40.000 I mean, it's a you know, momentum is one of 0:11:40.000 --> 0:11:41.480 the greatest strategies. 0:11:41.000 --> 0:11:44.560 There isn't, especially these days, and a lot of it 0:11:44.600 --> 0:11:47.880 relates to market structure. Yeah, And so this is where 0:11:48.480 --> 0:11:53.559 what I'm saying about job displacement due to AI becomes 0:11:53.600 --> 0:11:56.840 a problem. So in January, I'm completely sanguine. You know, 0:11:57.240 --> 0:12:00.280 question Carson, Well, when does you know when does it 0:12:00.280 --> 0:12:03.400 become a bad idea to invest in momentum within the 0:12:03.440 --> 0:12:08.920 S and P five hundred index when unemployment rises materially? Well, 0:12:09.000 --> 0:12:13.679 when do you see that happening? And not anytime soon? Well, okay, 0:12:14.160 --> 0:12:20.800 here's the issue. I now do see unemployment, particularly for 0:12:21.480 --> 0:12:25.760 high wage earners, rising significantly in the next several years. 0:12:26.440 --> 0:12:30.880 The problem that that creates for markets is that, especially 0:12:30.920 --> 0:12:33.240 with the S and P five hundred, so much of 0:12:33.240 --> 0:12:37.360 that is driven by flows, so especially for one K 0:12:37.559 --> 0:12:43.880 retirement plan contributions from these knowledge workers and that you know, 0:12:44.040 --> 0:12:45.240 and look when you look at the S and P 0:12:45.320 --> 0:12:47.800 five hundred index, I mean, it's it's kind of ridiculous 0:12:47.840 --> 0:12:49.880 that we use that as a proxy for the stock 0:12:49.960 --> 0:12:53.400 market because you know, generally, you know, two thirds of 0:12:53.400 --> 0:12:55.840 the stocks are underperforming the mean of the index, and 0:12:55.880 --> 0:12:58.040 the mean of the index is really driven by small 0:12:58.120 --> 0:13:03.040 number of companies that are out forming. But what has 0:13:03.120 --> 0:13:07.960 happened and so here I reference work that a friend 0:13:07.960 --> 0:13:10.680 of mine named Michael Green or Mike Green of Simplify 0:13:10.760 --> 0:13:14.160 asset management has done on passive investing, and he's been 0:13:14.200 --> 0:13:18.280 banging this strung since probably twenty nineteen or twenty twenty 0:13:18.280 --> 0:13:21.800 about how passive has warped markets. So it's created this 0:13:21.920 --> 0:13:27.120 virtuous cycle whereby money goes in from paychecks every month. 0:13:28.400 --> 0:13:32.120 And what happens is these index funds they buy stock 0:13:32.320 --> 0:13:35.680 at any price. They're completely priced and elastic. They remove 0:13:35.760 --> 0:13:38.760 supply of stock, and so for the largest names in 0:13:38.840 --> 0:13:41.520 the index or the indices, and if you look at 0:13:41.600 --> 0:13:45.120 Nasdaq one hundred, same thing. Those get the greatest share 0:13:45.160 --> 0:13:49.760 of every dollar that's allocated to the index, and so 0:13:49.800 --> 0:13:54.200 by removing supply, they create this situation in which the 0:13:54.280 --> 0:13:58.600 stocks that receive the largest allocations, the impact on their 0:13:58.600 --> 0:14:04.520 prices becomes parabolic versus linear, and that gets turbocharged with 0:14:04.600 --> 0:14:09.560 stock buybacks. So this has been great for the most part. 0:14:09.600 --> 0:14:11.480 I mean twenty twenty two, there was you know, there 0:14:11.559 --> 0:14:13.880 was like a little bump in the road, but we're 0:14:13.880 --> 0:14:17.559 well past that. So this has been fantastic for investors. 0:14:17.679 --> 0:14:23.280 But the thing is, if you have say fifteen percent 0:14:24.080 --> 0:14:27.880 of knowledge workers lose their jobs and they're not going 0:14:27.920 --> 0:14:31.200 to be able to replace that income they have credit 0:14:31.200 --> 0:14:34.520 card debt, they have mortgages, they have car loans, a 0:14:34.560 --> 0:14:40.440 student debt. This is going to massively impact the flows 0:14:41.160 --> 0:14:44.560 into the markets, and so probably the way it plays 0:14:44.560 --> 0:14:47.320 out on a micro level at first is a person 0:14:47.320 --> 0:14:51.160 who's lost his or her job will sell the taxable investments. 0:14:51.600 --> 0:14:54.520 A lot of the taxable investments are in the largest 0:14:54.600 --> 0:14:57.160 names in the S and P five hundred index, so 0:14:57.200 --> 0:15:00.320 you get selling there, but you've already had a situation 0:15:00.360 --> 0:15:02.960 where that laid off person is no longer making the 0:15:03.000 --> 0:15:05.480 contributions to the four one K plan. So you get 0:15:05.480 --> 0:15:10.520 to a point where those contributions go net zero, especially 0:15:10.560 --> 0:15:13.720 because from a demographic perspective, you do have people now 0:15:13.760 --> 0:15:18.400 who are also redeeming just because they've retired. And then 0:15:18.440 --> 0:15:20.760 you get to a point where on a net basis, 0:15:21.360 --> 0:15:25.360 the flows go negative as people have to sell you 0:15:25.600 --> 0:15:28.600 redeem their retirement accounts because they've been unable to replace 0:15:28.640 --> 0:15:34.040 their income and the issue. So, while passive has been 0:15:34.080 --> 0:15:38.800 this very virtuous cycle on the way up, it's built 0:15:38.880 --> 0:15:43.640 this significant fragility into the market, and so when those 0:15:43.680 --> 0:15:49.720 flows go in reverse, there's really not nearly enough active 0:15:49.760 --> 0:15:52.760 management out there to catch the following knives, especially because 0:15:53.120 --> 0:15:56.200 if you were looking at the valuations of some of 0:15:56.200 --> 0:15:59.720 the largest companies in the indices. In trying to look 0:15:59.760 --> 0:16:02.640 at them on a purely fundamental basis, it's hard to 0:16:02.760 --> 0:16:08.120 justify the evaluations in some cases. So that's where you get, 0:16:08.400 --> 0:16:14.120 in my view, the GFC type events. It has to 0:16:14.160 --> 0:16:17.040 do with the fragility that's been created in the equity 0:16:17.080 --> 0:16:20.280 markets through passive investing. So if you really want to 0:16:20.360 --> 0:16:24.720 understand my view, it's my own view on AI displacing 0:16:24.800 --> 0:16:28.600 labor layered on top of Mike Green's work on passive 0:16:28.640 --> 0:16:32.760 investing and in the vulnerabilities that creates in the equity 0:16:32.760 --> 0:16:34.160 markets and. 0:16:34.040 --> 0:16:37.440 That reduction in demand frequity, so that reduction and flow. 0:16:37.840 --> 0:16:41.280 It coincides with the period an unusual period over the 0:16:41.360 --> 0:16:44.160 last decade of an increase in the supply of equities 0:16:44.160 --> 0:16:47.200 as well, because we have these big IPOs coming through 0:16:47.280 --> 0:16:47.920 at the same time. 0:16:48.240 --> 0:16:50.080 It SpaceX and then we'll get open eye, we'll get 0:16:50.080 --> 0:16:50.840 down tropic. 0:16:50.640 --> 0:16:52.920 And of course on paper, we're thrilled by this right 0:16:52.960 --> 0:16:54.920 because it shows the stock market doing and the stock 0:16:54.960 --> 0:16:58.120 market is supposed to do, which is be a place 0:16:58.120 --> 0:17:00.840 where people putting new money into growth companies as opposed 0:17:00.840 --> 0:17:04.119 to an investor extracturn machine, which has effectively being for 0:17:04.160 --> 0:17:05.679 the last decade, the most of the money has been 0:17:05.680 --> 0:17:08.320 coming out in buybacks, dividends, etc. So you get at 0:17:08.359 --> 0:17:11.159 the same time as if you're right, of course, that 0:17:11.560 --> 0:17:15.000 the flows for the supply is coming up at the 0:17:15.000 --> 0:17:17.200 same time right now. 0:17:17.240 --> 0:17:20.159 The other question there, and I don't I don't have 0:17:20.200 --> 0:17:22.399 a view on this. You know, at least to the 0:17:22.440 --> 0:17:27.760 extent that it remains independent of labor displacement. But to 0:17:27.840 --> 0:17:31.879 what extent are the hyper scalers going to be going 0:17:31.960 --> 0:17:35.919 to continue to be able to issue credit to fund 0:17:36.160 --> 0:17:40.480 their AI buildouts because they need to issue that credit 0:17:40.760 --> 0:17:44.320 in order to maintain the stock buybacks. So if they can't, 0:17:44.520 --> 0:17:47.600 if they can't issue paper at yields they consider to 0:17:47.640 --> 0:17:52.520 be attractive, then they're going to have to divert resources 0:17:52.520 --> 0:17:56.240 that would be used for share buybacks to scaling out 0:17:56.600 --> 0:17:59.520 data centers, et cetera. So you could start to see, 0:18:00.000 --> 0:18:04.080 you know, to me, either scenario is equally probable going 0:18:04.200 --> 0:18:06.920 into when we have this, you know, when these streams 0:18:06.920 --> 0:18:09.040 start really showing up in the labor markets and the 0:18:09.119 --> 0:18:14.320 flows start to go toward net zero, can have this situation, 0:18:14.960 --> 0:18:17.119 you know, can like I said, equally probable to me 0:18:17.160 --> 0:18:21.080 at this point where companies are still engaging in share buybacks, 0:18:21.480 --> 0:18:25.840 so maybe it hits you know, this impact hits more suddenly, 0:18:26.359 --> 0:18:31.040 or companies have to taper the share buybacks because they're 0:18:31.040 --> 0:18:33.600 not able to fund in the credit markets, and so 0:18:33.680 --> 0:18:36.800 you start to see you start to see less upward 0:18:36.960 --> 0:18:40.000 pressure on stock prices as a result of that. Anyway, 0:18:40.200 --> 0:18:43.600 So either scenario, you know, at this point, I'm coin 0:18:43.600 --> 0:18:44.520 flip on those two. 0:18:45.680 --> 0:18:48.359 Yeah, and on the demographics, I mean, this was a 0:18:48.440 --> 0:18:52.119 scenario that everyone expected to happen anyway, just a little 0:18:52.119 --> 0:18:54.640 further out, as the baby boomers work their way through 0:18:54.680 --> 0:18:57.080 the system and start to sell out fund their retirements, 0:18:57.119 --> 0:19:00.440 we would have expected to see those flows reversed. Anyway, 0:19:00.720 --> 0:19:03.880 This just brings that forward by well, probably a decade. 0:19:05.080 --> 0:19:08.679 Yeah, yeah, I mean, look, I think you know, my 0:19:08.760 --> 0:19:12.560 impression of that question has been that there are people 0:19:12.560 --> 0:19:14.200 will argue, well, it's not going to be that bad 0:19:14.200 --> 0:19:18.480 because you're gonna have a great wealth transfer and you know, YadA, YadA, YadA, 0:19:18.520 --> 0:19:20.560 and you know money will be reallocated from credit to 0:19:20.600 --> 0:19:25.280 equities and younger. But you know, no view in a 0:19:25.320 --> 0:19:27.679 world that doesn't go through an AI disruption. No view 0:19:27.760 --> 0:19:30.760 on the timing of that. But but yeah, as I said, 0:19:30.800 --> 0:19:34.240 I think that we're you know, I think that question 0:19:34.359 --> 0:19:37.560 is not really going to be that relevant compared to 0:19:37.800 --> 0:19:39.480 what's okay. 0:19:39.600 --> 0:19:43.800 So let's say we have a GFC style disaster in 0:19:43.840 --> 0:19:48.280 the markets. It's quick, it's huge, very unpleasant. How do 0:19:48.359 --> 0:19:50.719 the play out after that? And we're very used to thinking, well, 0:19:50.720 --> 0:19:53.800 we don't really worry about this stuff over the long time. 0:19:53.800 --> 0:19:55.760 We just hang on because there'll be a government response. 0:19:55.760 --> 0:19:58.080 There'll be a fiscal response or a monetary response, so 0:19:58.200 --> 0:19:59.800 we'll make everything absolutely fine. 0:20:00.720 --> 0:20:03.439 Yeah, no, I'm look, there will be all of that. 0:20:03.520 --> 0:20:07.320 I mean. The one thing is the playbooks are now 0:20:07.680 --> 0:20:11.840 very well developed for you know, unlike the GFC, where 0:20:12.640 --> 0:20:15.040 a lot of the responses theoretical it took a long 0:20:15.080 --> 0:20:18.680 time to implement. Governments had to get authorities to do it. 0:20:18.960 --> 0:20:24.680 The authorities exist, they've been expanded during COVID. So from 0:20:24.720 --> 0:20:29.560 a monetary perspective that you know, I mean rates, you know, 0:20:29.600 --> 0:20:33.160 real rates will go zero or negative you know over well, 0:20:33.600 --> 0:20:35.640 the interesting thing those AI is going to be very 0:20:35.840 --> 0:20:39.600 is going to be deflationary. So maybe maybe it's a 0:20:39.680 --> 0:20:41.840 little bit stepping on a land mine if I talk 0:20:41.880 --> 0:20:46.240 about real rates, but nominal rates you know, will be 0:20:46.520 --> 0:20:48.760 I mean, you know, well below on the short you know, 0:20:48.880 --> 0:20:51.640 short end well below one hundred basis points, I mean, 0:20:51.680 --> 0:20:55.680 probably zero to ten BIPs, you know, very quickly. And 0:20:55.920 --> 0:20:59.639 there's going to be a lot of the going to 0:20:59.840 --> 0:21:02.360 Governments are going to have to buy. The central banks 0:21:02.400 --> 0:21:04.720 are going to have to buy you know, the entire 0:21:05.119 --> 0:21:08.360 curve basically to keep it really you know, to keep 0:21:08.440 --> 0:21:12.239 rates you know, long end rates low. Now there will 0:21:12.320 --> 0:21:14.520 be fiscal issues. I think fiscal is going to be 0:21:14.520 --> 0:21:18.359 a lot more interesting because the monetary playbook is well established. 0:21:20.000 --> 0:21:22.920 You know, the United States has the blessing of being 0:21:22.920 --> 0:21:27.160 able to borrow in the currency that it prints, and 0:21:27.240 --> 0:21:30.040 so I think, you know, q E forever is the 0:21:30.040 --> 0:21:35.960 future here effectively, but it'll be you know, one of 0:21:36.000 --> 0:21:38.840 the third order effects. And where I think it's also 0:21:38.960 --> 0:21:43.560 interesting to look at some trades setting up will be 0:21:43.720 --> 0:21:47.400 what happens in the US to state in local governments 0:21:48.840 --> 0:21:50.960 you know there, I mean, especially if you look at 0:21:51.320 --> 0:21:53.000 you know, the states that are going to have their 0:21:53.359 --> 0:21:57.200 tax bases the hardest hit are the ones that are 0:21:57.280 --> 0:22:03.199 the most prolific issuers of debt. So California, California is 0:22:03.280 --> 0:22:06.040 I mean, it's because so much of its tax base 0:22:06.960 --> 0:22:11.520 comes from income taxes, any individual income taxes. Anyway, this 0:22:11.600 --> 0:22:14.760 is going to be you know, like Armygeddon for California. 0:22:14.800 --> 0:22:18.760 From a fiscal perspective, so how quickly does the federal 0:22:18.760 --> 0:22:23.200 government step in in bailac California, Illinois, New York, New Jersey, Connecticut. 0:22:24.440 --> 0:22:26.800 Now it kind of depends on who. I mean a 0:22:26.840 --> 0:22:29.000 big part depends on who's in the White House and 0:22:29.080 --> 0:22:33.639 which party controls Congress. But you know, the from the 0:22:33.640 --> 0:22:37.240 fiscal perspective, you know, I think the US will be 0:22:37.280 --> 0:22:41.639 okay because of the potential for q E. You know, 0:22:41.760 --> 0:22:48.399 the Eurozone countries that's going to be very tough, you know, 0:22:48.520 --> 0:22:52.040 but but that's where but you know, I so so look, 0:22:52.080 --> 0:22:55.160 I think we go through, we go we know what's 0:22:55.160 --> 0:22:57.679 going to happen in terms of on the monetary side, 0:22:57.800 --> 0:23:00.800 and they will have to try to reinflate assts and 0:23:00.840 --> 0:23:03.640 that will eventually happen. Assets will be you know, reinflated. 0:23:05.359 --> 0:23:10.639 But this will this will reorder society significantly because I mean, 0:23:10.840 --> 0:23:13.919 the the politics, the amount of turbulence that we're going 0:23:14.000 --> 0:23:16.560 to have to go through, you know, to really figure 0:23:16.600 --> 0:23:21.480 out how we how we structure societies. It's going to 0:23:21.560 --> 0:23:24.240 be significant. And you know, look, I do think on 0:23:24.320 --> 0:23:26.280 the on the back end of this, on the back 0:23:26.400 --> 0:23:28.800 end of a lot of turbulence, and I don't you know, 0:23:29.119 --> 0:23:32.480 this is a number of years to get there. The 0:23:32.520 --> 0:23:36.000 good news is, you know, I think I think most 0:23:36.040 --> 0:23:39.320 people will be will live reasonably comfortable lives. Okay, I 0:23:39.359 --> 0:23:42.000 think you know, the you know, a lot of the 0:23:42.040 --> 0:23:44.680 private sector will be there to serve the government. Governments 0:23:44.680 --> 0:23:49.640 will be massive employers, massive consumers of highly inefficient products 0:23:49.640 --> 0:23:51.679 and services from the private sector. So I think a 0:23:51.680 --> 0:23:53.840 lot of your private sector ends up looking like Chinese 0:23:53.880 --> 0:23:57.280 state owned enterprises. Then there's another part of the private 0:23:57.320 --> 0:24:00.399 sector that focuses on the private sector. It's gonna be 0:24:00.440 --> 0:24:04.199