WEBVTT - Rob Arnott on AI Mania: Lessons From the Dot-Com Era 0:00:02.720 --> 0:00:16.759 Bloomberg Audio Studios, Podcasts, Radio News. 0:00:18.200 --> 0:00:20.960 Welcome to Meren Talk's Money, the podcast in which people 0:00:20.960 --> 0:00:23.880 who know the markets explain the markets. I am Meren 0:00:23.960 --> 0:00:26.720 Sumset Web. Now. This week we welcome back to the show. 0:00:26.800 --> 0:00:29.840 Rob Arnott. Rob is founder and chairman of the board 0:00:29.880 --> 0:00:34.240 of Research Affiliates are research intensive asset management firm. Now, Rob, 0:00:34.479 --> 0:00:36.160 I've asked you back on for a couple of reasons. 0:00:36.159 --> 0:00:39.400 First because we always have really interesting chats, and I 0:00:39.440 --> 0:00:41.960 know that the listeners super enjoyed the show that we 0:00:42.000 --> 0:00:45.559 did last time. But also because the question that we 0:00:45.640 --> 0:00:49.479 get asked the most on the show out and about 0:00:49.760 --> 0:00:54.120 everywhere we go is is there an AI bubble? And 0:00:54.400 --> 0:00:57.760 if there is spoiler alert, almost everybody thinks there is. 0:00:57.960 --> 0:01:00.360 What do I do? What do I do now? You 0:01:00.400 --> 0:01:02.800 wrote back in March, you wrote a great report about 0:01:02.800 --> 0:01:05.119 this about you know, whether this is a bubble or not, 0:01:05.160 --> 0:01:06.959 comparing it to the dot com bubble, looking at in 0:01:06.959 --> 0:01:09.960 a wider context, etc. There was quite a few months back, 0:01:10.040 --> 0:01:11.360 and that was at the point when it looked like 0:01:11.440 --> 0:01:14.920 things were cracking slightly. Right. Since then, if it is 0:01:14.959 --> 0:01:19.000 a bubble, it's an even bigger bubble. And one thing 0:01:19.120 --> 0:01:21.960 that we do know is that whether there is a 0:01:22.000 --> 0:01:25.880 bubble in AI or not, there is definitely a bubble 0:01:25.920 --> 0:01:28.760 in people asking us questions about whether there is a 0:01:28.760 --> 0:01:33.160 bubble in AI, and in politicians and eccentral bankers and 0:01:33.360 --> 0:01:37.479 supernational organizations looking at you the IMF suggesting that there 0:01:37.560 --> 0:01:41.560 might be a disorderly correction as we you'phemistically like to 0:01:41.560 --> 0:01:44.200 call these things at a later date. So all this 0:01:44.360 --> 0:01:47.400 is going on, which is why I thought you Rob 0:01:47.600 --> 0:01:51.720 of a perfect guest. So Rob, welcome to Meren Talks Money, 0:01:52.040 --> 0:01:58.320 Thank you onwards. So let's start then with that report 0:01:58.360 --> 0:02:01.160 that you wrote in March, and am I okay to 0:02:01.160 --> 0:02:02.520 put a link to that in the show note so 0:02:02.560 --> 0:02:03.800 everyone can look at themselves later. 0:02:04.000 --> 0:02:07.320 Yeah, of course, absolutely, please please propelling. 0:02:07.440 --> 0:02:09.320 Okay, So there will be a link to this report 0:02:09.520 --> 0:02:12.239 in the show notes so everyone can read it in FULD. 0:02:12.520 --> 0:02:16.040 But let's start with talking about that. Basically, what you're 0:02:16.040 --> 0:02:17.720 doing is you're comparing it to the dot com bubble, 0:02:17.720 --> 0:02:20.600 which really was a proper bubble and obvious bubble from 0:02:20.639 --> 0:02:23.000 so long in advance and back then we looked at 0:02:23.000 --> 0:02:25.400 that bubble, all of us and said, well that's a bubble. 0:02:25.560 --> 0:02:27.960 But we couldn't tell you when it was going to burst. 0:02:27.960 --> 0:02:30.480 So one of the things you do is you compare 0:02:30.919 --> 0:02:32.840 the AI bubble, which going to call it a bubble 0:02:32.880 --> 0:02:35.960 for shorthand here the AI bubble with that bubble. So 0:02:36.040 --> 0:02:38.840 let's talk first about what's similar, what is the same 0:02:38.960 --> 0:02:40.280 between these two situations. 0:02:40.919 --> 0:02:44.519 You've got froth evaluations. You've got to spread and valuation 0:02:44.720 --> 0:02:49.120 between growth stocks and value stocks. That's become to borrow 0:02:49.160 --> 0:02:54.720 the British term ginormous uh, and it's uh. You've got 0:02:54.760 --> 0:02:58.720 to spread and valuation between large cap and small cap stocks. 0:02:58.760 --> 0:03:03.920 That's at w reck extremes. Now. One of the things 0:03:03.960 --> 0:03:07.840 that people say now that they also said in the 0:03:07.880 --> 0:03:11.560 dot com bubble is this isn't a bubble, this is real. 0:03:12.760 --> 0:03:16.560 Back then, they said the Internet is going to change everything. 0:03:16.639 --> 0:03:18.840 It'll change how we buy and sell goods and services. 0:03:18.880 --> 0:03:21.280 It'll change how we get our news, how we do 0:03:21.360 --> 0:03:26.919 our research. It'll change how we interact and network with friends, 0:03:26.919 --> 0:03:31.040 with family, with clients. It's going to change everything. And 0:03:31.080 --> 0:03:34.240 these are the dominant players and they're going to build 0:03:34.240 --> 0:03:37.600 our future. Most of that was true. 0:03:37.960 --> 0:03:40.280 All of the first bet was true, right, all of 0:03:40.320 --> 0:03:40.920 the first bet. 0:03:41.280 --> 0:03:47.279 Yes, Now where the narrative failed is that newcomers came in, competed, 0:03:47.960 --> 0:03:53.080 and the leading players weren't all leading players ten years later. 0:03:53.720 --> 0:03:56.880 In fact, the ten most valuable tech stocks in the 0:03:56.920 --> 0:04:00.520 world coming into the year two thousand, z ro out 0:04:00.560 --> 0:04:02.520 of ten beat the S and P over the next 0:04:02.560 --> 0:04:06.280 fifteen years. You had to wait eighteen years for one 0:04:07.040 --> 0:04:10.560 Microsoft to actually edge past the S and P. Eighteen 0:04:10.640 --> 0:04:13.320 years is a long time to wait. The other things 0:04:13.360 --> 0:04:18.280 that happened were the adoption of the Internet happens slower 0:04:18.320 --> 0:04:22.599 than the advocates expected. The adoption of AI is happening 0:04:22.680 --> 0:04:27.159 slower than the advocates of this revolution suggest. So the 0:04:27.240 --> 0:04:32.119 narrative AI will change everything, My goodness, it absolutely will. 0:04:32.760 --> 0:04:38.080 It's breathtaking. Anybody who's used chat GPT five, it's astonishing 0:04:38.480 --> 0:04:43.160 how intelligent it is. We wrote an article fundamental growth 0:04:43.200 --> 0:04:47.920 that's on the Internet and is likely to appear at 0:04:47.960 --> 0:04:52.640 a journal early in the new year, and we gave 0:04:52.680 --> 0:04:55.760 it to Chat GPT five point zero and said, tell 0:04:55.839 --> 0:04:58.480 us what's good and bad about this paper. Tell us 0:04:58.640 --> 0:05:04.520 where there's repetition we can trim, tell us whether there 0:05:04.520 --> 0:05:08.080 are references or citations that we missed. It gave a 0:05:08.120 --> 0:05:10.160 two page summary of the paper that was better than 0:05:10.200 --> 0:05:12.680 anything I or Cam Harvey or the other authors could 0:05:12.720 --> 0:05:17.520 possibly have written. Beautiful, wonderful synopsis. Then it gave us 0:05:17.560 --> 0:05:21.280 a rundown of ways to improve the paper. We took 0:05:21.440 --> 0:05:24.000 most of its suggestions. Then it gave us a list 0:05:24.080 --> 0:05:27.880 of ten citations. We missed four of them. We looked 0:05:27.880 --> 0:05:30.240 at them and we thought, wow, why didn't we think 0:05:30.240 --> 0:05:32.920 of that? Yes, put it in. Four of them were 0:05:32.960 --> 0:05:35.839 more borderline. We ignored them. Two of them didn't exist. 0:05:36.080 --> 0:05:39.120 So it still hallucinates. It creates stuff out of thin air, 0:05:39.440 --> 0:05:42.320 which is something that you've got to be very careful about. 0:05:42.360 --> 0:05:47.400 But it's breathtaking. The leaders of the AI revolution our 0:05:47.480 --> 0:05:54.320 leaders today. Will they be disrupted? Disruptors get disrupted. Think 0:05:54.320 --> 0:05:56.680 of the most valuable tech stocks in the world in 0:05:56.720 --> 0:06:01.680 two thousand, Microsoft, It went through it off decade, and 0:06:01.720 --> 0:06:05.559 then it regained its footing and became the world leader 0:06:05.600 --> 0:06:10.880 that it is today, Cisco. Cisco's chairman in March of 0:06:10.920 --> 0:06:12.920 two thousand said, I don't see why we can't grow 0:06:12.960 --> 0:06:14.839 forty percent a year. As far as I can see, 0:06:15.360 --> 0:06:17.120 forty percent a year means you're going to get be 0:06:17.160 --> 0:06:21.360 six times as large in five years. How big are 0:06:21.360 --> 0:06:23.960 they now? They're six times as large. It took twenty 0:06:23.960 --> 0:06:24.680 five years. 0:06:24.440 --> 0:06:24.880 Not five. 0:06:25.080 --> 0:06:26.000 Yeah, it's a yeah. 0:06:26.160 --> 0:06:28.880 This is exactly the kind of thing that we hear 0:06:29.000 --> 0:06:32.040 from Nvidia, which by the way, as we're speaking, has 0:06:32.080 --> 0:06:34.560 been hitting new highs. Exactly the kind of thing we 0:06:34.680 --> 0:06:38.640 hear from them, that this growth can go on indefinitely. 0:06:38.440 --> 0:06:43.839 If you can find power stations to power. I read 0:06:43.880 --> 0:06:47.479 somewhere that's something like five percent of the US power 0:06:47.520 --> 0:06:50.320 supply is now used by AI, and they're wanting it 0:06:50.360 --> 0:06:54.880 to double every year for the next three years. Well, 0:06:54.880 --> 0:06:56.560 that would take it to forty percent. 0:06:56.960 --> 0:06:57.159 Yeah. 0:06:57.400 --> 0:07:02.320 There's a wonderful paper by Aschenbrenner, one of the co 0:07:02.400 --> 0:07:07.280 founders of Open Ai, called Situational Awareness, that talks about 0:07:07.400 --> 0:07:12.120 the future of AI. And it's wonderful because it's so 0:07:12.160 --> 0:07:15.000 insightful on how many ways it can change the world. 0:07:15.560 --> 0:07:19.040 It's off target in a couple of ways. He says, 0:07:19.080 --> 0:07:22.480 we need to increase the power generation capability of the 0:07:22.600 --> 0:07:26.520 US grid by forty percent by twenty twenty seven. Well, 0:07:27.400 --> 0:07:30.240 maybe we should, but it's not going to happen. 0:07:31.440 --> 0:07:34.400 And even of course, if you manage to increase the 0:07:35.120 --> 0:07:38.840 power going into the grid. Making the grid resilient enough 0:07:38.880 --> 0:07:42.080 to carry that huge increase in power would be verging 0:07:42.080 --> 0:07:44.280 on impossible, and a timeframe like that, right, So there 0:07:44.320 --> 0:07:49.280 are power constraints on AI, there are regulatory constraints on AI, 0:07:49.480 --> 0:07:53.520 maybe ethical constraints on AI. And then I guess it's 0:07:53.560 --> 0:07:56.960 also the case if we stick with this comparison to 0:07:58.080 --> 0:08:01.040 the dot com bubble and possibly even too maybe even 0:08:01.080 --> 0:08:03.560 the smart the smartphone boom or whatever you like to 0:08:03.560 --> 0:08:05.760 call it, in from two thousand and seven on, it's 0:08:05.800 --> 0:08:09.840 around that Chinese competition, which didn't exist in quite the 0:08:09.840 --> 0:08:12.960 same way in the same volume. Then we've moved, we've 0:08:13.000 --> 0:08:15.320 moved from the similarities to the differences too quickly. That 0:08:15.400 --> 0:08:16.840 was what I meant to do, but we've done it. 0:08:17.160 --> 0:08:21.560 So there's more competition, global competition that suggests that maybe 0:08:21.720 --> 0:08:24.000 first mover advantage isn't quite what it might have been, 0:08:24.040 --> 0:08:26.480 and maybe there's a first mover disadvantage or there certainly 0:08:26.520 --> 0:08:28.200 there certainly was in the dot com bubble. 0:08:28.360 --> 0:08:32.880 Yeah, well, rounding out the top five back in two thousand, 0:08:33.080 --> 0:08:37.040 you have Intel, who was that disrupted? Good lord, yes, 0:08:38.640 --> 0:08:45.840 you've got Nokia. It still makes phones. It was fifth 0:08:45.880 --> 0:08:48.720 most valuable company on the planet. Back in two thousand, 0:08:49.200 --> 0:08:53.600 you've got loosened. Loosen doesn't exist anymore. I mean, it's 0:08:53.720 --> 0:08:58.080 it's a graveyard. And the other part that's just fascinating 0:08:58.160 --> 0:09:01.680 is March of two thousand, when the bubble burst. Roll o' 0:09:01.679 --> 0:09:05.199 clock forward two years, the Nasdaq was down fifty percent 0:09:05.240 --> 0:09:07.880 on its way to an eighty percent drop Russell two 0:09:07.960 --> 0:09:12.280 thousand value. Remember, value was trading cheap. Small cap was 0:09:12.280 --> 0:09:16.880 trading cheap again, same thing today, Russell two thousand value 0:09:16.920 --> 0:09:20.080 was up fifty three percent in those two years. The 0:09:20.200 --> 0:09:23.920 bull market of the nineteen nineties for most stocks didn't 0:09:24.080 --> 0:09:28.000 end until March of two thousand and two, and that 0:09:28.160 --> 0:09:30.560 was two years into the bursting of the bubble. The 0:09:30.559 --> 0:09:34.440 big cap indexes were dragged down, but the median stock 0:09:34.559 --> 0:09:37.800 was up for those next two years. The same thing 0:09:37.920 --> 0:09:41.679 could play out today. I'm not predicting that, but what 0:09:41.720 --> 0:09:46.000 I am saying is that decoupling of frothy large cap 0:09:46.240 --> 0:09:55.840 popular names from very successful mainstream businesses in valuation can 0:09:57.080 --> 0:10:00.199 mean revert and I think likely, Well. 0:10:00.320 --> 0:10:03.719 Well, you say can mean revert historically always have mean 0:10:03.720 --> 0:10:06.800 reversal correct, which doesn't mean that it happens this time 0:10:06.880 --> 0:10:09.120 and doesn't tell us when it might happen. But there 0:10:09.160 --> 0:10:13.560 has never previously been a point when this kind of 0:10:13.640 --> 0:10:17.640 disparity has not reverted to a historical mean of some kind. 0:10:17.720 --> 0:10:21.160 Yeah, here's a fun factoid. Members of the S and 0:10:21.200 --> 0:10:24.920 P five hundred or of the Russell one thousand are 0:10:25.160 --> 0:10:28.480 at valuation multiples that are two and a half times 0:10:28.520 --> 0:10:32.160 as large as non members. Now people will say, yeah, 0:10:32.160 --> 0:10:35.800 that's because these are wonderful companies and these aren't pardon me, 0:10:35.960 --> 0:10:38.400 over the last thirty years, roll o' clock back thirty 0:10:38.480 --> 0:10:40.680 years they were at parody. They were about the same 0:10:40.760 --> 0:10:43.880 valuation multiples. So they blew out one hundred and fifty 0:10:43.880 --> 0:10:48.680 percentage point gap opened up. These guys. The non members 0:10:49.040 --> 0:10:52.280 have had earnings and dividend growth one percent a year 0:10:52.480 --> 0:10:57.960 faster than the members. Oh so they're better companies on 0:10:58.160 --> 0:11:02.840 average that are valued at forty percent the valuation multiples 0:11:02.840 --> 0:11:07.199 of the index members. We have a short paper membership 0:11:07.240 --> 0:11:12.440 has its privileges that's in draft form that examines this difference, 0:11:12.480 --> 0:11:14.920 and it just points out the obvious that if you're 0:11:14.960 --> 0:11:18.240 a member of the index, membership has its privileges, you 0:11:18.320 --> 0:11:22.640 are entitled one to a higher valuation multiple. Because you're 0:11:22.679 --> 0:11:27.040 a member two, you're entitled to steady inflows of capital 0:11:27.480 --> 0:11:31.079 propping the price up and holding it higher. And both 0:11:31.120 --> 0:11:35.280 of those facts are realities of today's market. Indexing has 0:11:35.840 --> 0:11:41.599 really swamped the markets and has compromised the price discovery 0:11:41.640 --> 0:11:44.120 mechanism of figuring out what companies are actually worth. 0:11:44.679 --> 0:11:46.880 Yeah, the point being that, because of the rise of 0:11:46.920 --> 0:11:50.600 passive investing, companies automatically get flows every time anyone buys 0:11:50.600 --> 0:11:54.720 an index, so the differentiation between between the valuation of 0:11:54.760 --> 0:11:56.440 stocks becomes most least. 0:11:57.120 --> 0:11:58.000 That's exactly right. 0:11:58.760 --> 0:12:00.960 Okay, But here we are, and I think you know, 0:12:01.000 --> 0:12:04.720 we're already talking ourselves into agreeing that this is a 0:12:04.760 --> 0:12:07.760 bubble and that it shows all the characteristics of a bubble. 0:12:07.960 --> 0:12:10.360 And then we have the few things we haven't discussed, 0:12:10.400 --> 0:12:13.719 like the insane levels of concentration in the US market 0:12:13.920 --> 0:12:17.120 any global we've been at and global markets, so we've 0:12:17.160 --> 0:12:20.400 been at concentration levels like this before. Have we been 0:12:20.440 --> 0:12:23.080 quite the highest? I think we have been near these levels, 0:12:23.080 --> 0:12:26.120 but it's always a massive danger sign, not even not 0:12:26.200 --> 0:12:26.640 even near. 0:12:26.720 --> 0:12:33.000 No, this is the highest concentration in global and particularly 0:12:33.160 --> 0:12:36.640 US because it's US concentration that leads to the global concentration. 0:12:36.920 --> 0:12:38.440 I mean, the US is two thirds of the world 0:12:38.559 --> 0:12:42.360 stock market. It's three fourths of the developed world stock market. 0:12:42.760 --> 0:12:46.600 So it's the magnificent seven thirty percent of the US 0:12:46.679 --> 0:12:49.720 stock market. They're twenty percent of the world stock market, 0:12:49.920 --> 0:12:54.240 these seven companies. And yes they're magnificent. Yes they have 0:12:54.320 --> 0:12:58.040 product that everybody wants. Yes they have product that everyone needs. 0:12:58.160 --> 0:13:02.679 Yes they're reshaping the future of the world. But this 0:13:02.800 --> 0:13:05.920 is a competitive landscape. Look at Intel. I mean, even 0:13:05.920 --> 0:13:10.199 the Chips Act couldn't save Intel, and a fifty billion 0:13:10.200 --> 0:13:13.440 dollars infusion from the US government could couldn't help it 0:13:14.120 --> 0:13:17.320 defend itself against the rise of Nvidia, because in Nvidia 0:13:18.040 --> 0:13:24.120 disrupted them. And disruption is normal. Disruption happens regularly in 0:13:24.160 --> 0:13:28.920 a healthy capital economy. Creative destruction is a real thing. 0:13:29.120 --> 0:13:33.800 If you are behind the curve, you're going to lose 0:13:34.000 --> 0:13:38.400 business and lose market share and be pushed aside. And 0:13:38.800 --> 0:13:42.800 that's actually a good thing. It means that the better 0:13:42.840 --> 0:13:47.000 ideas get a chance to gain traction and grow. And 0:13:47.640 --> 0:13:53.480 these seven companies are all disruptors and are all very 0:13:53.600 --> 0:13:57.040 powerful in their industry, but none of them is immune 0:13:57.080 --> 0:14:01.679 to competition, and Vidia has already a number of organizations 0:14:01.760 --> 0:14:05.559 lapping at their heels with new super chips that are 0:14:05.559 --> 0:14:08.360 in some ways better than in videos, and Video got 0:14:08.360 --> 0:14:12.840 a head start which was tremendous. The story behind that's 0:14:13.080 --> 0:14:18.160 also fun. Jensen Wang said, these super chips that we're building, 0:14:18.600 --> 0:14:20.480 let's save some real estate on them and just not 0:14:20.560 --> 0:14:23.760 fill it with circuitry because the next round of circuitry 0:14:23.800 --> 0:14:26.160 is going to be better and we won't have to 0:14:26.200 --> 0:14:31.000 redesign the whole chip. Who builds a chip with dead 0:14:31.000 --> 0:14:33.920 space in it? It was a stroke of genius. It 0:14:34.000 --> 0:14:38.880 allowed them to plunk in the new, better stuff faster 0:14:39.360 --> 0:14:41.520 a year or two faster than they would have otherwise. 0:14:41.680 --> 0:14:44.840 And boy, what a stroke of genius. So these are 0:14:44.880 --> 0:14:49.440 brilliant companies. That was his disruption of the whole silicon industry. 0:14:49.520 --> 0:14:51.480 I suppose one thing we should just say and side 0:14:51.520 --> 0:14:55.200 is that it's not the concentration in itself that is 0:14:55.360 --> 0:14:57.800 necessarily a problem. In the UK, stop market is also 0:14:57.920 --> 0:15:00.760 very concentrated, doing a very large percentage of it is 0:15:00.800 --> 0:15:03.440 just our biggest stocks. But part of the problem with 0:15:03.640 --> 0:15:06.640 this concentration is the correlation. So they're all in the 0:15:06.680 --> 0:15:09.320 same industry, all in the same business and if you 0:15:09.360 --> 0:15:12.120 look at the UK market, for example, while we have concentration, 0:15:12.560 --> 0:15:15.320 the big companies are generally and in different businesses. So 0:15:15.440 --> 0:15:18.720 it's both things, the concentration and the correlation between them, right, 0:15:19.120 --> 0:15:20.720 and the business connections between them. 0:15:20.760 --> 0:15:24.840 Of course, here's a couple of fun factoids. The market 0:15:24.920 --> 0:15:29.520 value of the Magnificent Seven was twenty trillion at the 0:15:29.600 --> 0:15:33.320 end of September. That's bigger than all stock markets in 0:15:33.360 --> 0:15:41.320 Europe combined. That's bigger than all of the Chinese stock market. Basically, 0:15:41.440 --> 0:15:44.720 the market's value for a company is supposed to be 0:15:44.800 --> 0:15:48.960 the market's best gas at future profit distributions back to 0:15:49.000 --> 0:15:53.320 the shareholders discounted back to today, so the net present 0:15:53.440 --> 0:15:57.160 value of future profits. What the market is saying is 0:15:57.200 --> 0:16:00.480 that the future profits of these seven companies will exceed 0:16:00.800 --> 0:16:05.080 all profits generated by all of Europe, bigger than all 0:16:05.120 --> 0:16:09.440 profits generated by all of China. Is that possible? Yes? 0:16:10.720 --> 0:16:15.320 Is that likely? I don't think so. These companies compete 0:16:15.320 --> 0:16:19.120 against one another, and Vidia's fifty percent profit margins are 0:16:19.560 --> 0:16:21.880 not going to continue more than two or three more 0:16:21.960 --> 0:16:26.960 years because competition is coming into play. If they're roll o' 0:16:27.000 --> 0:16:30.360 clock forward ten yures, if they are forty percent of 0:16:30.400 --> 0:16:33.880 the super chip market instead of ninety, if their profit 0:16:33.960 --> 0:16:37.920 margins are still a robust thirty percent not fifty. If 0:16:38.560 --> 0:16:44.840 the pricing power means that revenues per chip are down 0:16:45.040 --> 0:16:47.640 by two thirds over the next ten years, and that's 0:16:47.680 --> 0:16:51.600 probably conservative. Plug those into a spreadsheet and what you 0:16:51.640 --> 0:16:54.960 find is okay, goodness, Their two and thirty five profits 0:16:54.960 --> 0:16:57.360 are going to be no better than their twenty twenty 0:16:57.360 --> 0:17:00.160 five profits, and yet their priced is if they're going 0:17:00.200 --> 0:17:05.320 to see stupendous growth. Pallunteers another fabulous example. How many 0:17:05.400 --> 0:17:10.240 companies get to be priced at triple digit multiples of sales, 0:17:10.560 --> 0:17:15.360 not of profits. It peaked at one hundred, one hundred 0:17:15.840 --> 0:17:19.960 eighty times sales. So with sales last year of two 0:17:20.040 --> 0:17:26.120 to three billion, they were they're worth about four hundred billion. Now, 0:17:26.240 --> 0:17:29.119 how do you justify four hundred billion market cap when 0:17:29.160 --> 0:17:31.240 your sales are two to three billion. The only way 0:17:31.280 --> 0:17:32.919 to do it is to have those sales grow to 0:17:33.000 --> 0:17:38.000 twenty billion, forty billion, one hundred billion and profits to 0:17:38.160 --> 0:17:44.159 match and do that fast enough. Is that possible? Yeah? 0:17:44.880 --> 0:17:47.640 Is it likely? I don't think so. So I see 0:17:47.680 --> 0:17:52.240 bubbles galore. I also see some very profitable, highly successful 0:17:52.240 --> 0:17:57.920 companies priced it premium multiples, but not nosebleed levels. And 0:17:59.080 --> 0:18:02.040 so I think there's a lot of stocks in this 0:18:02.160 --> 0:18:08.919 bubble that are not nonsensically priced. And yeah, that's a 0:18:08.920 --> 0:18:12.240 good thing. That's a healthy side of the market. 0:18:12.840 --> 0:18:15.239 Okay, I get this my feeling that you don't think 0:18:15.280 --> 0:18:18.240 it's quite ready to burst yet. We're not quite at 0:18:19.240 --> 0:18:20.440 you know, Cisco levels. 0:18:20.840 --> 0:18:24.160 Well, some of them are at Cisco level. Cisco crested 0:18:24.240 --> 0:18:28.320 it in early two thousand, one hundred and eighty times 0:18:28.320 --> 0:18:31.480 earnings and pallunteers way past that. 0:18:34.119 --> 0:18:36.200 But I just want to read that one of the 0:18:36.200 --> 0:18:38.640 sentence that you write in this report, which I think 0:18:38.720 --> 0:18:40.960 kind of sums up everything you're saying, which is a 0:18:41.080 --> 0:18:45.119 very simple, fast developing sectors writing waves of innovation and 0:18:45.200 --> 0:18:48.479 optimism are inherently volatile, which is what you mean when 0:18:48.520 --> 0:18:51.040 you talk about the regular disruption of these things. And 0:18:51.040 --> 0:18:53.840 one of the extraordinary things about a bubble is that 0:18:54.920 --> 0:18:58.359 inside the stories that people tell themselves inside a bubble, 0:18:58.359 --> 0:19:01.920 that the general narrative that there is no volatility, there 0:19:01.960 --> 0:19:06.320 isn't competition, there isn't the disruptors remain the disruptors. They 0:19:06.359 --> 0:19:09.600 do not become the disrupted. That's how the story works exactly. 0:19:09.640 --> 0:19:12.200 And the thing that tends to break these bubbles is either, 0:19:12.480 --> 0:19:13.879 you know, well, it's either it's a change in the 0:19:13.920 --> 0:19:16.280 cost of capital, it's a change in the wider environment, 0:19:16.840 --> 0:19:19.840 or it's the realization that the narrative doesn't include the 0:19:19.880 --> 0:19:20.960 next wave of disruption. 0:19:21.400 --> 0:19:25.880 Yeah, and people talk about the jobs lost. Every technological 0:19:26.000 --> 0:19:30.840 revolution since the industrial revolution has destroyed millions of jobs. 0:19:31.560 --> 0:19:37.399 The thing is that painful to those whose jobs or 0:19:37.400 --> 0:19:43.640 whose families members' jobs get destroyed. Yes, do people care 0:19:43.680 --> 0:19:46.120 about those job losses a generation later? 0:19:46.760 --> 0:19:46.840 No. 0:19:47.720 --> 0:19:51.120 The thing that is different is that the computers will 0:19:51.160 --> 0:19:54.160 be more intelligent than we are in the not too 0:19:54.200 --> 0:19:58.880 distant future. In some ways, they already are. My CEO, 0:19:59.400 --> 0:20:03.600 Chris Bright likes to play with AI every weekend for 0:20:04.000 --> 0:20:07.359 three to five hours, testing its capabilities, and he had 0:20:07.359 --> 0:20:12.639 a conversation with GPT five about how AI fits in 0:20:12.680 --> 0:20:17.160 the realm of human evolution, and it gave a discourse, 0:20:17.600 --> 0:20:21.560 short discourse that was so insightful you'd think it was 0:20:21.560 --> 0:20:27.199 from a double PhD in anthropology and philosophy. And it 0:20:27.480 --> 0:20:32.280 closed with the observation, we don't know yet whether human 0:20:32.359 --> 0:20:38.440 intelligence and machine intelligence will merge, or whether human intelligence 0:20:38.720 --> 0:20:42.160 will at some stage be a fond and distant memory 0:20:42.800 --> 0:20:48.280 for AI. And then it closes with the sentence the 0:20:48.359 --> 0:20:51.399 wonderful thing in this revolution is that we get to choose. 0:20:52.400 --> 0:20:57.520 And I thought, who the heck is we? 0:20:57.520 --> 0:21:00.480 Well, that's an interesting question, isn't it who we? In 0:21:00.560 --> 0:21:02.840 this context? I don't think it's me. It might be you, 0:21:02.920 --> 0:21:04.119 but I'm not feeling like it's me. 0:21:05.119 --> 0:21:11.359 But anyway, existential threat, Yeah, do I worry about that? 0:21:11.800 --> 0:21:14.760 I don't because I try to worry about things I 0:21:14.800 --> 0:21:17.919 can have any influence over. This revolution is coming and 0:21:17.920 --> 0:21:21.439 it's big. But who's going to be the disruptor in 0:21:21.480 --> 0:21:24.239 the coming decade? Who will be the market leaders ten 0:21:24.320 --> 0:21:28.439 years from now? Some of market leaders today will be 0:21:28.520 --> 0:21:31.960 the market leaders in ten years. Some won't. They're all 0:21:32.440 --> 0:21:39.919 chrisd as if they will be the market leaders. 0:21:47.920 --> 0:21:50.359 So I think where we are now is that we 0:21:50.400 --> 0:21:53.679 are agreeing that we are at the very least in 0:21:54.080 --> 0:21:59.680 bubble territory, that valuations are in bubble territory, that there 0:21:59.720 --> 0:22:03.680 is not enough understanding of the next round of disruption. 0:22:03.800 --> 0:22:06.480 One thing we haven't mentioned, actually is the extent of 0:22:06.520 --> 0:22:09.040 the capex and the extent which AI capex is crowding 0:22:09.080 --> 0:22:12.760 out everything else in the US, and when stuff gets very, 0:22:12.880 --> 0:22:16.760 very capex heavy, of course, if you do get disrupted, 0:22:16.800 --> 0:22:18.560 you're in a whole lot of trouble, right. I mean, 0:22:18.760 --> 0:22:22.200 a lot of the dot com bubble with very capex light, 0:22:22.600 --> 0:22:25.520 and same with smartphones, platforms, et cetera. Capex light. But 0:22:25.560 --> 0:22:26.680 this is pretty capex heavy. 0:22:26.840 --> 0:22:31.080 I'm glad you mentioned that. One of the fascinating facts 0:22:31.280 --> 0:22:38.159 about growth stocks is that R and D is positively 0:22:38.240 --> 0:22:43.160 and reasonably strongly correlated with future growth for companies. I mean, 0:22:43.200 --> 0:22:45.480 it's not perfect. You can waste money on R and 0:22:45.560 --> 0:22:47.680 D just like you can waste money on anything else. 0:22:48.600 --> 0:22:53.920 Capex spending is negatively correlated with future growth. Really surprising. 0:22:53.920 --> 0:22:56.280 We don't tell the AI found that. 0:22:56.560 --> 0:23:02.040 Well, here's an example. In a healthy capitalist system, happy 0:23:02.040 --> 0:23:06.760 customers are the essential component of success. If you're going 0:23:06.800 --> 0:23:09.880 to succeed, you must have happy customers who value your 0:23:09.880 --> 0:23:12.840 product more highly than what you charge for it. In 0:23:13.000 --> 0:23:16.879 Vidia has happy customers. They line up for the privilege 0:23:16.880 --> 0:23:20.920 of getting on a waiting list to buy massive quantities 0:23:20.920 --> 0:23:24.639 of super chips. In Vidia's customers have yet to figure 0:23:24.680 --> 0:23:28.439 out how to monetize that capex. They're having trouble figuring 0:23:28.520 --> 0:23:32.600 out how to earn money on AI And so is 0:23:32.640 --> 0:23:36.960 that going to happen? Yes, it will, But is it 0:23:37.040 --> 0:23:40.080 happening fast enough to support the prices of these stocks? 0:23:41.080 --> 0:23:45.040 Probably not, And therein lies one of the key challenges. 0:23:45.320 --> 0:23:51.400 Okay, So add that infa mix, and we could say, 0:23:51.560 --> 0:23:53.159 when you wrote this back in March, you said, and 0:23:53.200 --> 0:23:56.080 maybe we're in nineteen ninety seven, maybe we're in the 0:23:56.160 --> 0:23:59.120 late nineteen nineties, or maybe we're right on the edge 0:23:59.160 --> 0:24:01.920 of that nasty that beginning of two thousand. We can't 0:24:01.920 --> 0:24:04.399 be sure. It sounds like you would put yourself closer 0:24:04.400 --> 0:24:06.080 to the late nineties at the moment. 0:24:06.320 --> 0:24:09.760 No, No, I would. I would say that we're more 0:24:09.880 --> 0:24:11.720 likely to be near the cusp than not. 0:24:12.560 --> 0:24:16.480 Now, okay, we say, let's put ourselves at November ninety nine. 0:24:16.600 --> 0:24:18.920 You want to do that, Sure we're in November ninety nine. 0:24:18.960 --> 0:24:22.520 If we were in November ninety nine, what would be done. 0:24:22.520 --> 0:24:24.119 The first thing we'd do is we'd sell our Cisco 0:24:24.160 --> 0:24:26.520 because we've already made pushing three thousand percent on that 0:24:26.640 --> 0:24:28.280 right then, what would we do? 0:24:28.600 --> 0:24:34.440 The trick about bubbles is never short them because they 0:24:34.440 --> 0:24:37.359 can go longer, last longer, and go further than you 0:24:37.400 --> 0:24:38.199 can imagine. 0:24:38.280 --> 0:24:40.120 That's what a lot of spread patter is found out 0:24:40.119 --> 0:24:41.840 in the UK in nineteen ninety nine. 0:24:42.040 --> 0:24:46.280 And so what you do is you underweight, you fade 0:24:46.359 --> 0:24:50.119