WEBVTT - The Biggest AI Investment Mistakes Investors Could Make 0:00:02.720 --> 0:00:17.079 Bloomberg Audio Studios, Podcasts, Radio News. 0:00:18.360 --> 0:00:20.280 Welcome to Marin Talks Money. In the podcast in which 0:00:20.280 --> 0:00:22.960 people who know the markets explain the markets. I am 0:00:23.040 --> 0:00:25.520 Maren Somerset Web and this week I am speaking with 0:00:25.560 --> 0:00:29.600 Sarah Catterer. Sarah is the CEO of Causeway Capital Management 0:00:29.640 --> 0:00:32.280 affirm and she co founded in two thousand and one. 0:00:32.600 --> 0:00:35.320 Today we speak about how AI is going to impact 0:00:35.320 --> 0:00:38.320 and is impacting the fund management industry. We talk about 0:00:38.400 --> 0:00:40.800 bargains in the UK and whether anyone is ever going 0:00:40.840 --> 0:00:43.800 to pick them up, what's going on in biotech and pharma, 0:00:44.080 --> 0:00:48.400 and why you may need more medical devices than you think. Sarah, 0:00:48.520 --> 0:00:50.120 Welcome to Marin Talks Money. 0:00:50.240 --> 0:00:53.160 Thank you for inviting me right now. Why don't we start? 0:00:53.240 --> 0:00:56.000 Can I just ask you to tell me what it 0:00:56.040 --> 0:00:58.640 is the Causeway actually does. You've been on the podcast before. 0:00:58.720 --> 0:01:01.640 We've got piles and you listen, it'll rundown of what 0:01:01.720 --> 0:01:02.640 is your business actually do. 0:01:03.640 --> 0:01:08.600 Cosway is a fund manager. We manage approximately eighty billion 0:01:08.720 --> 0:01:13.560 dollars of client assets and our clients are largely institutional 0:01:13.600 --> 0:01:17.479 and our business is global equity. So we have both 0:01:17.520 --> 0:01:20.880 fundamental research and quantitative research, and we have strategies that 0:01:21.200 --> 0:01:26.479 deploy primarily fundamental with quantitative risk control and then primarily 0:01:26.560 --> 0:01:30.960 quantitative with fundamental risk control. So that's what distinguishes us. 0:01:31.000 --> 0:01:35.480 It's the it's the convergence of both fundamental quant. 0:01:35.640 --> 0:01:38.039 Yeah, the combination of the two, which is not something 0:01:38.080 --> 0:01:41.440 you find that often. Okay, So following directly on from that, 0:01:41.600 --> 0:01:43.760 I wanted to ask you about AI because I know 0:01:43.800 --> 0:01:45.440 you've been thinking about a lot. I know you've been 0:01:45.480 --> 0:01:48.000 talking about it, but the key question is how are 0:01:48.000 --> 0:01:50.040 you using it in your own business. You know, hearing 0:01:50.120 --> 0:01:52.440 a lot about the businesses that are being disrupted by 0:01:52.480 --> 0:01:55.160 AI and how they're being disrupted, and I'm beginning to 0:01:55.200 --> 0:01:57.000 hear a lot from people saying, well, what do we 0:01:57.000 --> 0:01:58.920 need fund managers for because AI can do this? 0:01:59.120 --> 0:02:05.400 Yes, if only I know. Well, I believe that the 0:02:05.440 --> 0:02:09.120 initial applications of AI for us at Causeway has been 0:02:09.120 --> 0:02:11.720 that we've had a similar experience to other enterprises and 0:02:11.760 --> 0:02:17.040 that we've got tremendous efficiency gains from coding so that 0:02:17.240 --> 0:02:20.960 our software engineers could do so much more in the 0:02:21.000 --> 0:02:24.360 same amount of time. And I've noticed this based on 0:02:25.320 --> 0:02:29.240 our head of investment, technology has been able to do 0:02:29.320 --> 0:02:31.920 much more for our quant effort and the people who 0:02:31.960 --> 0:02:36.519 work for him that the quant software infrastructure has become 0:02:36.919 --> 0:02:40.359 much more developed more quickly. That means that our quant 0:02:40.400 --> 0:02:45.840 researchers can test more ideas more and do so not 0:02:46.000 --> 0:02:49.680 just more quickly with more accuracy. The head of our 0:02:49.720 --> 0:02:54.680 digital services just our whole it effort. He's been able 0:02:54.720 --> 0:02:58.840 to help us produce tax optimizer. He did in about 0:02:58.880 --> 0:03:02.440 two weeks and normally would have taken months. This was all, 0:03:03.480 --> 0:03:08.200 I guess in part. Vibe coded using AI models to 0:03:08.480 --> 0:03:13.520 accelerate the software development. But the output has been really good, 0:03:13.600 --> 0:03:17.600 and the interaction with our team to make sure that 0:03:17.639 --> 0:03:19.519 it does what it's supposed to do for our mutual 0:03:19.520 --> 0:03:22.480 funds has been great. But I'm going to just step 0:03:22.520 --> 0:03:27.839 back and say that from a coding perspective fantastic. From 0:03:27.880 --> 0:03:32.800 an investment perspective, the answer is mixed, and it's that 0:03:33.480 --> 0:03:39.440 AI has been made us, each of us as research 0:03:39.440 --> 0:03:42.600 analysts able to read everything. We have an assistant now 0:03:42.800 --> 0:03:48.360 our AI agents who read everything, every company transcript, earning transcript, 0:03:48.400 --> 0:03:55.240 every government agency filing, every every trade journal, every bit 0:03:55.280 --> 0:03:56.280 of news. 0:03:56.240 --> 0:03:59.400 Every actually the annual report. Someone's reading the report. 0:03:59.480 --> 0:04:02.520 Finally, it reads them all it can do all of that, 0:04:02.560 --> 0:04:06.320 but what it can't do is make decisions with the information. 0:04:07.160 --> 0:04:12.160 I don't know how AI will replicate the fundamental portfolio 0:04:12.160 --> 0:04:15.440 manager who relies on his or her experience, has a 0:04:15.520 --> 0:04:19.039 level of intuition that's probably thirty to forty percent of 0:04:19.040 --> 0:04:23.520 the investment decision comes from something that you can't even quantify. 0:04:23.600 --> 0:04:28.800 It's this instinct honed after decades of doing this, and 0:04:28.880 --> 0:04:35.440 from a quanm perspective, with AI, the ability to find 0:04:35.520 --> 0:04:39.760 signals that aren't actually explainable to clients as quite a temptation. 0:04:39.880 --> 0:04:44.719 We'd be really careful not to go and use the tool, 0:04:45.560 --> 0:04:51.440 i'd say ineffectively. So in short, AI is an accelerant 0:04:51.440 --> 0:04:56.600 to productivity improvement, but it hasn't yet developed where it 0:04:56.600 --> 0:05:00.200 can replace that human judgment. And I think that's true 0:05:00.240 --> 0:05:02.839 across many of the companies that we research. They have 0:05:02.960 --> 0:05:07.480 some of the similar experiences. So it doesn't confine to finding. 0:05:07.680 --> 0:05:10.760 When we're talking about it reading everything, every research report, 0:05:10.880 --> 0:05:14.640 every every company announcement, every everything, and that'll be done 0:05:14.720 --> 0:05:18.040 and summarized and presumably produced for enos. That's I mean, 0:05:18.040 --> 0:05:20.000 if you're looking at thousands of companies, which I know 0:05:20.080 --> 0:05:23.320 you are, that's a massive time saving. But back to 0:05:23.360 --> 0:05:26.960 the intuition bit. It is the reading of those reports, 0:05:27.040 --> 0:05:30.279 the reading of the company announcements, followed by the reading 0:05:30.360 --> 0:05:32.440 of the transcripts of earning calls, et cetera, et cetera, 0:05:32.680 --> 0:05:35.160 where a good analysts might pick up the bits and 0:05:35.200 --> 0:05:38.720 bobs and make the connections that create the intuition. So 0:05:38.760 --> 0:05:41.800 without the without the hard graft bit, yes, can you 0:05:41.839 --> 0:05:42.760 have the bay. 0:05:42.880 --> 0:05:45.200 I can make some connections. You can find in the 0:05:45.240 --> 0:05:48.520 footnotes where there might be counting anomalies. What it can't 0:05:48.520 --> 0:05:50.440 do is what we're doing. It can't have a meeting 0:05:50.480 --> 0:05:53.839 with management and observe the body language. You can't say, 0:05:54.200 --> 0:05:56.160 you know, this doesn't feel right. I don't even know 0:05:56.200 --> 0:05:58.120 how to put it into words. But I think they're 0:05:58.160 --> 0:06:01.400 becoming the measure a little more caution, and I'm going 0:06:01.440 --> 0:06:05.280 to take that back and assimilate that information. We don't 0:06:05.320 --> 0:06:08.440 record meetings we I think management would find that a 0:06:08.440 --> 0:06:12.440 little off putting and probably we wouldn't get any meetings. 0:06:12.839 --> 0:06:16.960 But we're taking notes. But what we're it's what we're 0:06:17.080 --> 0:06:21.000 hearing and the way we're observing how the message is 0:06:21.000 --> 0:06:25.719 being delivered. That's so incredibly important. And this is not. 0:06:27.720 --> 0:06:31.839 Someday when robots are doing the job and it's all visual, 0:06:32.440 --> 0:06:33.960 I think the other side of the table will be 0:06:34.000 --> 0:06:38.080 a robot too. It's going to be a pretty grim world. 0:06:38.320 --> 0:06:41.640 How And I don't normally get into young people early on. 0:06:41.680 --> 0:06:43.480 We normally save that for the end. But the way 0:06:43.520 --> 0:06:45.760 you're talking about the fund management business now the fundamental 0:06:45.800 --> 0:06:47.960 part of it. Yes, what is the young person who 0:06:47.960 --> 0:06:50.080 wants to be in fund management these days do because 0:06:50.120 --> 0:06:52.760 all that grunt work that was the training ground. 0:06:52.800 --> 0:07:00.000 Yeah, yes, the grunt work initially is probably a good idea. 0:07:00.160 --> 0:07:04.000 We encourage analysts to do everything from scratch to understand 0:07:04.040 --> 0:07:07.679 what the meaning is of all these documents, where where 0:07:07.720 --> 0:07:11.000 to go? When you in a prompt, one has to 0:07:11.040 --> 0:07:13.720 ask the AI to search certain areas, or we have 0:07:14.640 --> 0:07:21.000 agent builders that link into systems. It links into databases 0:07:21.040 --> 0:07:23.040 and you have to know which database is to plumb. 0:07:24.360 --> 0:07:27.120 But once an analysis has done this a few times, 0:07:28.000 --> 0:07:30.920 they don't necessarily want to do the gruntwork because it 0:07:30.920 --> 0:07:35.040 displaces in a zero sum day any something more higher 0:07:35.120 --> 0:07:39.560 level thinking oriented. So I'm okay with the AI doing 0:07:39.600 --> 0:07:42.920 the grunt work. It's just that if people stop thinking, 0:07:42.960 --> 0:07:46.840 that's when we're in trouble. It's it's using that time 0:07:47.000 --> 0:07:50.040 freedom where you're not entering the entering in the quarter's 0:07:50.120 --> 0:07:52.680 earnings into your spreadsheet, but the AI is doing it 0:07:52.760 --> 0:07:57.200 your clawed ad in or whatever the tool is. What 0:07:57.960 --> 0:08:00.800 how else is the analyst using that freedom time? Is 0:08:01.840 --> 0:08:04.960 he or she thinking about more about the assumptions, really 0:08:05.080 --> 0:08:09.080 testing the model, going back to the notes and thinking 0:08:09.120 --> 0:08:11.400 again about where they might have been wrong, taking a 0:08:11.400 --> 0:08:14.880 step back and thinking more about structural change and how 0:08:15.000 --> 0:08:18.640 this company what come out of left field being one's 0:08:18.720 --> 0:08:23.120 own devil's advocate, and the AI can help with that too, interestingly, 0:08:23.960 --> 0:08:26.480 But that's the higher level thinking that we want and 0:08:26.560 --> 0:08:27.320 we are promoting. 0:08:27.840 --> 0:08:32.720 Okay, I'm convinced we still need you. Thanks. When you 0:08:32.760 --> 0:08:35.160 look at how working in your own company, and then 0:08:35.160 --> 0:08:38.040 you look at the businesses in which you invest, what 0:08:38.080 --> 0:08:40.240 are you looking for in the companies you're investing in? 0:08:41.000 --> 0:08:44.960 Well, I'm glad you asked, because we for every sector globally, 0:08:44.960 --> 0:08:47.400 have done a deep dive into how we believe they 0:08:47.400 --> 0:08:49.959 will be impacted by AI. This is quite an endeavor. 0:08:50.000 --> 0:08:52.560 We did this fundamentally in our quant colleagues observed the 0:08:52.600 --> 0:08:56.360 process and the output. But what we found is for 0:08:56.559 --> 0:08:59.640 many there are some structural changes that are going to 0:08:59.640 --> 0:09:03.120 be quite significant, and then others there's a lot of upside. 0:09:03.520 --> 0:09:06.480 Let's go to healthcare to start. We have a number 0:09:06.480 --> 0:09:09.079 of pharmaceutical companies in the portfolio and now we're adding 0:09:09.080 --> 0:09:12.800 more medical technology because they've been left out in the 0:09:12.840 --> 0:09:17.800 cold in this huge AI driven rally. But their ability 0:09:17.800 --> 0:09:21.439 to innovate, their ability to in the case of pharmaceuticals, 0:09:21.559 --> 0:09:25.520 discovered new molecules advance them through the clinical trial process 0:09:25.600 --> 0:09:30.640 more quickly, efficiency, lower cost is that could be given 0:09:30.720 --> 0:09:35.560 that patent lifes remain the same. That means more years 0:09:35.720 --> 0:09:39.440 out in the marketplace and less in development. That's a 0:09:39.480 --> 0:09:44.320 profit booster. It's more hard pressed to see how there'll 0:09:44.360 --> 0:09:49.320 be some sort of mysterious AI entity that will develop drugs. 0:09:49.840 --> 0:09:54.160 Pharmaceutical business is very laborious and involves a lot of 0:09:54.200 --> 0:09:59.040 interaction with the health authorities and clinical trials are carefully 0:09:59.080 --> 0:10:02.480 scripted and then being able to target patients with AI 0:10:02.600 --> 0:10:05.840 in advance of a doing trials so they are they 0:10:05.920 --> 0:10:10.600 are the right individuals for that particular drug. That's another 0:10:10.840 --> 0:10:14.760 way to accelerate the process and get output that can 0:10:14.800 --> 0:10:18.080 be get that drug into the market more quickly. So 0:10:18.240 --> 0:10:20.840 for healthcare, we think this is going to be fantastic. 0:10:22.320 --> 0:10:25.559 Healthcare companies they gather lots and lots of data and 0:10:25.600 --> 0:10:27.720 their ability to use it in ways they never could 0:10:27.760 --> 0:10:31.640 before with AI is not that dissimilar from other industries. 0:10:31.640 --> 0:10:35.319 It's just they have so much intellectual property and their 0:10:35.360 --> 0:10:39.480 ability to innovate now is improved. And innovation equals pricing power, 0:10:40.160 --> 0:10:44.200 and pricing power is margin, and margin is profitability, and 0:10:44.200 --> 0:10:46.880 profitability leads to cash flow, and cash flow gets returned 0:10:46.920 --> 0:10:47.640 to shareholders. 0:10:48.800 --> 0:10:50.959 Okay, that makes sense. How does that then interact? I 0:10:50.960 --> 0:10:53.000 mean one of the one of things about healthcare sector 0:10:53.080 --> 0:10:56.360 is how regulated. It hugely regulated across the board, regulated 0:10:56.360 --> 0:10:58.880 in different ways in every country, but the layers of 0:10:58.960 --> 0:11:01.000 regulation massive. 0:11:01.160 --> 0:11:05.520 Is that a barrier, It's part of the moat, and 0:11:05.679 --> 0:11:10.359 regulation and compliance and on the need for audible outcomes 0:11:10.800 --> 0:11:14.280 are one of the reasons why some industries they will 0:11:14.640 --> 0:11:16.920 they just won't have this sort of AI competition. 0:11:17.000 --> 0:11:20.840 The market might be inferring they're going to incur. So 0:11:20.920 --> 0:11:23.160 a mote is a plus and a minus. A mote 0:11:23.200 --> 0:11:27.840 can protect the business from competition, but it can also 0:11:28.080 --> 0:11:33.280 mean that there will be it conhibit growth. You need competition, yes, 0:11:33.559 --> 0:11:39.040 yes you do. But the working their way for the 0:11:39.240 --> 0:11:43.720 regulatory maze is what healthcare companies do very well, and 0:11:43.760 --> 0:11:46.440 the large ones have whole teams for this, and the 0:11:46.480 --> 0:11:49.000 teams may be able to harness AI to be more 0:11:49.040 --> 0:11:51.880 efficient doing that, which could be also a cost savings 0:11:51.880 --> 0:11:53.120 as well as time saving. 0:11:53.880 --> 0:11:56.240 Interesting, and we've all talked on this podcast about how 0:11:56.240 --> 0:11:59.760 big companies love regulation because nothing, nothing holds competition back 0:11:59.760 --> 0:12:02.920 more and a good bit of regulation. And one of 0:12:03.000 --> 0:12:04.800 the thought was about AI was that it might be 0:12:05.080 --> 0:12:09.720 one way for new entrance to circumvent We'll find easier 0:12:09.720 --> 0:12:12.080 ways to deal with large parts of regulation. But it 0:12:12.120 --> 0:12:14.600 sounds like your case is that that is not. 0:12:16.040 --> 0:12:18.920 Going to happen. It depends in something that gets put 0:12:18.960 --> 0:12:22.880 into the human body. The regulations there for a reason. Yeah. Yeah, 0:12:23.160 --> 0:12:27.600 it's a little different. Say financial services, okay, that necessarily 0:12:27.960 --> 0:12:29.720 got to create a health outcome. 0:12:30.440 --> 0:12:33.200 Okay, So what other sectors will have a regulatory remote 0:12:33.280 --> 0:12:36.679 that will limit the use of AI elsewhere? 0:12:37.320 --> 0:12:42.000 Well, iound I've been talking about software as a service. 0:12:42.080 --> 0:12:46.440 This sounds a little silly, but it's tangential. The SaaS 0:12:46.480 --> 0:12:49.200 companies have all sold off this idea that you're a 0:12:49.240 --> 0:12:53.000 software business, you had a seventy plus percent gross margin 0:12:53.120 --> 0:12:56.480 or higher. Now your business is losing its mode and 0:12:57.160 --> 0:13:00.520 the software will be vibe coded. Anybody can now code, 0:13:01.040 --> 0:13:05.120 but code isn't the mote. And these software companies that 0:13:05.200 --> 0:13:11.160 provide like SAP or Oracle very comprehensive for Fortune for 0:13:11.200 --> 0:13:14.920 five hundred global companies, and SAP in particular this mission 0:13:14.960 --> 0:13:19.199 critical software where they're dealing with systems of ledger in 0:13:19.320 --> 0:13:23.959 businesses that are heavily regulated. That means that data is crucial. 0:13:24.600 --> 0:13:28.680 There can't be errors in the outcome. So software needs 0:13:28.679 --> 0:13:34.000 to be deterministic. Yet generative AI is non deterministic. We 0:13:34.080 --> 0:13:37.400 can be a useful tool for users in the enterprise 0:13:37.440 --> 0:13:40.079 to get access to their data and see it in 0:13:40.160 --> 0:13:44.400 different ways. But it can't run the business. It can't 0:13:44.440 --> 0:13:47.000 run HR, can't run account it can't run supply. 0:13:46.760 --> 0:13:48.840 Chain, can't run relations. 0:13:49.000 --> 0:13:51.920 Yes, so I think about that as the market has 0:13:52.040 --> 0:13:58.400 just decided these companies are all to be incurring great competition, 0:13:59.160 --> 0:14:04.679 and we would disagree. It's the it's regulation and the 0:14:04.720 --> 0:14:07.880 need for the businesses out everything they do to be 0:14:08.080 --> 0:14:13.440 fully audible and transparent, auditable that that make it will 0:14:13.440 --> 0:14:16.559 make it quite hard for new AI native AI startups 0:14:16.600 --> 0:14:18.319 to undermine them. 0:14:18.559 --> 0:14:21.440 Okay, so you quite heavily invested in the likes of SAPA. 0:14:22.200 --> 0:14:29.840 We are. We are fundamentally the good management, incredibly sticky 0:14:29.920 --> 0:14:33.360 business and all that migration to cloud that's coming up, 0:14:33.760 --> 0:14:36.000 that's one of the absolute So what we know, there's 0:14:36.000 --> 0:14:38.440 a lot we don't know. What we do know is 0:14:38.480 --> 0:14:41.760 that for enterprises to take advantage of AI, they need 0:14:41.800 --> 0:14:44.080 to move their data to the cloud, they need to 0:14:44.120 --> 0:14:46.480 have it accessible for the AI, and they need to 0:14:46.480 --> 0:14:49.600 be able to access that tremendous compute on an as 0:14:49.640 --> 0:14:50.480 needed basis. 0:14:51.160 --> 0:14:53.280 And are you on the other side of this, are 0:14:53.320 --> 0:14:55.240 you and any of the hyperscalars that they don't fit 0:14:55.320 --> 0:14:57.920 into your your fundamental matrix. 0:14:58.320 --> 0:15:01.640 Some of them do, some of them do. For example, 0:15:02.080 --> 0:15:08.040 the two great digital advertising leviathans, Meta and Alphabet generate 0:15:08.720 --> 0:15:11.560 enough free cash flow to where they can fund their 0:15:12.200 --> 0:15:15.920 foundational model ambitions and so they can fund their own 0:15:16.000 --> 0:15:19.200 cap X, which is very different from some of these 0:15:20.160 --> 0:15:26.560 massive new AI companies where it's all about finding more 0:15:26.600 --> 0:15:28.160 investors who believe in the dream. 0:15:28.840 --> 0:15:31.480 They're going to be voyant with the opena ipo. 0:15:32.240 --> 0:15:34.600 I cannot say. 0:15:34.800 --> 0:15:38.320 I'm kind of guessing where else are you finding value 0:15:38.320 --> 0:15:38.880 in the market? 0:15:39.600 --> 0:15:43.560 Well, I did mention in passing medical technology. I just 0:15:43.560 --> 0:15:46.520 think it's fascinating that you could find companies like Boston 0:15:46.880 --> 0:15:52.960 Scientific and heart valves and very complex, important cardiovascular devices 0:15:53.040 --> 0:15:56.560 that you cannot get anywhere else. They used to trade 0:15:56.800 --> 0:15:59.760 twenty five or thirty times earnings and now they're multiples. 0:15:59.760 --> 0:16:05.560 Are half that. Uh, Companies like Striker or zimmer BioMed. 0:16:05.680 --> 0:16:12.560 They're fantastic medical devices as long as human bodies break 0:16:12.600 --> 0:16:17.640 down are going to be needed. And the very research 0:16:17.680 --> 0:16:22.080 intensive business. And to the degree they keep innovating and 0:16:22.120 --> 0:16:26.040 they're not in commodity products, they should do quite well. 0:16:26.120 --> 0:16:29.320 Part of what's weight on them has been concerns about 0:16:30.200 --> 0:16:34.880 the g LP one trends and consumer and individuals will 0:16:34.920 --> 0:16:40.200 no longer need new body parts, and we think that's unlikely. 0:16:40.240 --> 0:16:44.720 In fact, the longer they live, the more part replacements 0:16:44.720 --> 0:16:45.240 they'll need. 0:16:45.400 --> 0:16:48.080 Okay, so people will get thin, they'll still need the 0:16:48.120 --> 0:16:50.480 replacement polsitives. I quite need them until later. 0:16:50.960 --> 0:16:51.880 Yes, and they won't. 0:16:51.920 --> 0:16:52.280 They won't. 0:16:52.320 --> 0:16:57.720 They won't all get thin, sadly, wun't they? No, Some 0:16:57.760 --> 0:17:00.680 people will neither want an injection or where they wanted 0:17:00.760 --> 0:17:02.920 they will take a pill, they won't, they won't do it, 0:17:03.160 --> 0:17:08.200 and compliance will well, as always with drugs, they're not necessary. 0:17:08.240 --> 0:17:10.040 People aren't necessarily willing to comply. 0:17:10.320 --> 0:17:13.119 Sure, it has been fascinating seeing the obesity rate in 0:17:13.160 --> 0:17:15.440 the US drop. It has for the first time. 0:17:15.720 --> 0:17:20.800 Yes, it has change, and that should lead to better 0:17:20.840 --> 0:17:25.240 healthcare outcomes and lower cost healthcare. 0:17:25.240 --> 0:17:29.040 Two new hips each whenevery one's ninety, yes one at sixty, 0:17:29.359 --> 0:17:32.200 I'll take two, please, and a couple of knees. Or 0:17:56.359 --> 0:17:59.680 when I look at your at your your main vehicles 0:17:59.760 --> 0:18:02.679 ugly global, global, international, two different words for it. But 0:18:04.000 --> 0:18:07.240 you also have some emergen markets exposure. The Global fundkind 0:18:07.240 --> 0:18:09.440 of twenty percent in emerging markets, and there's also one 0:18:10.040 --> 0:18:13.240 Emerging Markets fund, right what's interesting there, And we've talked 0:18:13.240 --> 0:18:15.840 about this quite a lot recently about how the difficulty 0:18:15.840 --> 0:18:18.240 of emerging markets being now so defined by career in 0:18:18.240 --> 0:18:21.879 Taiwan and that market concentration making a nonsense of the 0:18:21.920 --> 0:18:24.840 idea that you can get diversified exposure within index has 0:18:24.840 --> 0:18:26.560 been one of the things we've been I'm afraid going 0:18:26.560 --> 0:18:28.920 on about for rather too long. But then the question 0:18:29.000 --> 0:18:32.320 is if you're not going to have index index style 0:18:32.359 --> 0:18:35.040 exposure to those two countries. Of course, to China, what 0:18:35.080 --> 0:18:35.639 are you going to have? 0:18:36.320 --> 0:18:40.960 Well, I would argue vociferously against indexing emerging markets, and 0:18:41.040 --> 0:18:44.160 I don't just give recent examples. Maybe twenty twenty two 0:18:44.160 --> 0:18:47.440 isn't that recent, but Russia, like, why would you even 0:18:47.480 --> 0:18:52.240 wanted an index weight in that market? We were underweight 0:18:52.280 --> 0:18:55.560 and very underweight, and that was extremely helpful. Sometimes emerging 0:18:55.600 --> 0:19:01.200 markets go to zero, and that one did. That's why 0:19:01.200 --> 0:19:04.280 they're emerging. They have much higher levels of risk because 0:19:04.280 --> 0:19:10.119 they're growth markets, and they interestingly, they're not all demographically 0:19:10.160 --> 0:19:15.000 in that growth phase. China, we didn't. It's been the 0:19:15.119 --> 0:19:17.520 largest of the emerging markets, and now it's number three 0:19:17.680 --> 0:19:21.439 behind South Korea and Taiwan, which is also fascinating, but 0:19:21.520 --> 0:19:22.560 it's within China, and that. 0:19:22.680 --> 0:19:24.560 All three of them have shrinking populations. 0:19:25.440 --> 0:19:29.119 They do, but at least South Korea and Taiwan have 0:19:30.000 --> 0:19:37.480 semiconductor giants in memory, and Taiwan has TSMC in chip production, 0:19:37.760 --> 0:19:40.560 and there are no substitutes at present, there won't be 0:19:40.600 --> 0:19:43.600 for a couple of years. No matter how many fabs 0:19:43.600 --> 0:19:48.960 are built or semiconductor facilities, it's going to take some 0:19:49.080 --> 0:19:52.240 time for supply to catch up with the demand. If 0:19:52.600 --> 0:19:56.680 demand may sag, and there may be technological reasons why 0:19:57.480 --> 0:20:01.080 we don't, you know, memories and quite hot as hotly 0:20:01.160 --> 0:20:05.119 demand as it is now. But that's very clear, and 0:20:05.160 --> 0:20:08.400 I can see those markets of those are highly highly concentrated, 0:20:08.720 --> 0:20:12.719 and the stocks they're those big three stocks are very 0:20:12.800 --> 0:20:15.840 large in the benchmark. It's beginning to look a little 0:20:15.840 --> 0:20:20.800 bit more like the US, oddly than anywhere else. What's 0:20:20.840 --> 0:20:23.760 interesting in emerging markets. I think in China there are 0:20:23.760 --> 0:20:25.520 all kinds of companies that have been left behind. We 0:20:25.520 --> 0:20:29.440 were talking about US hyperscalers. You could have the Chinese version, 0:20:29.520 --> 0:20:34.160 effectively the operating system of China through we chatter wishin 0:20:34.280 --> 0:20:39.440 which is Tencent. They're the largest publisher of music. 0:20:39.200 --> 0:20:39.800 Video or. 0:20:41.400 --> 0:20:43.800 Video games globally, and they have a music business, they 0:20:43.840 --> 0:20:48.160 have a messaging business, they have the most amazing flywheels. 0:20:48.200 --> 0:20:52.919 So they're massively cash generative like US hyper scalers, and 0:20:53.080 --> 0:20:58.400 they're developing through HY three their own foundational models. They 0:20:58.440 --> 0:21:00.680 got they were a little behind in a but it's 0:21:00.960 --> 0:21:03.399 when you have that kind of that that sort of 0:21:03.440 --> 0:21:06.320 financial resource, their net cash, it's not too hard to 0:21:06.359 --> 0:21:09.240 catch up, and they will. They'll find a way to 0:21:09.320 --> 0:21:15.240 get AI infused throughout their consumer products, that all the 0:21:15.280 --> 0:21:19.600 social media they do, and they'll keep their customers and 0:21:19.680 --> 0:21:24.920 that what is it one point three billion monthly average users? 0:21:25.160 --> 0:21:31.520 Like they're just this doc just keeps falling and we 0:21:31.680 --> 0:21:34.520 just look at it and we wonder, why, what an 0:21:34.560 --> 0:21:38.480 amazing company. It shouldn't trade at mid teens multiples of earnings. 0:21:38.480 --> 0:21:41.280 It should be much And part of this is just China. 0:21:41.440 --> 0:21:44.120 No one's interested right now. It's so there's this love 0:21:44.160 --> 0:21:46.800 hate relationship with China. Only if three or four years 0:21:46.800 --> 0:21:50.280 ago it was uninvestible, that's probably the time we all 0:21:50.280 --> 0:21:52.600 should have bought a lot of it. But now all 0:21:52.640 --> 0:21:56.879 eyes are in one area and not there. And especially 0:21:57.040 --> 0:22:04.360 for these Chinese companies that aren't necessarily at the considered 0:22:04.400 --> 0:22:08.280 at the absolute forefront of AI, they've been discarded. And 0:22:08.920 --> 0:22:11.080 this is where taking a two year view is really 0:22:11.200 --> 0:22:13.520 useful as opposed to just thinking out the next three 0:22:13.520 --> 0:22:16.080 to six months, because it gives us a chance to 0:22:16.160 --> 0:22:19.920 accumulate what maybe one of I think the I think 0:22:19.920 --> 0:22:21.760 my colleagues would agree with me, probably one of the 0:22:21.760 --> 0:22:26.880 best companies in China overall for a very attractive valuation. 0:22:27.320 --> 0:22:29.840 Yeah, and the lack of interest in it is just fashioned, 0:22:30.200 --> 0:22:33.399 just people coming slowly back to being in the Chinese market. Again, 0:22:33.640 --> 0:22:35.840 the fact that it's impossible to draw your eyes away 0:22:35.840 --> 0:22:39.200 from the stars and the etector in the US. And 0:22:39.400 --> 0:22:41.119 as they said, these few as you market. 0:22:41.200 --> 0:22:45.840 Yes, I think so. I think so that there's a 0:22:45.920 --> 0:22:48.560 certain amount of short term ism baked into that. It 0:22:48.680 --> 0:22:50.639 may or I mean, we could be we cause we 0:22:50.680 --> 0:22:53.640 could be wrong and ten Cent doesn't develop out their 0:22:53.680 --> 0:22:57.160 AI features fast enough and and part of their business 0:22:57.200 --> 0:23:02.119 gets competed away. But it's very unlikely that's going to happen, 0:23:02.160 --> 0:23:04.960 given their track record, given what they've already built, and 0:23:05.000 --> 0:23:09.400 given the marketplaces where they dominate. They you know, they've 0:23:09.440 --> 0:23:12.919 got infrastructure as a service, their platforms a service, and 0:23:13.000 --> 0:23:16.199 software as a service, and they will they have this 0:23:16.720 --> 0:23:19.720 cloud business and that gives them a lot of compute 0:23:19.800 --> 0:23:23.800 and a lot of flexibility. And again that balance sheet, 0:23:23.960 --> 0:23:26.399 it's like they have all the raw materials and they 0:23:26.440 --> 0:23:29.400 have the history of doing it, and that usually convinces 0:23:29.480 --> 0:23:33.479 us as fundamental investors. And given how low the valuation is, 0:23:34.240 --> 0:23:38.360 this isn't the risk is very modest. Versus I mean 0:23:38.400 --> 0:23:44.600 you mentioned semiconductors we started. Given where they trade, they're 0:23:44.840 --> 0:23:46.040 very risky right now. 0:23:46.119 --> 0:23:49.000 Well, because they're cychnical businesses. Right, this is exactly the 0:23:49.040 --> 0:23:52.040 wrong point of the cycle historically, if you were looking 0:23:52.080 --> 0:23:54.600 at this is a normal cycle, exactly the. 0:23:54.640 --> 0:23:57.199 Wrong cycle exactly. You don't believe in a cycle if 0:23:57.240 --> 0:24:00.480 you think that they the demand goes on prepared actually 0:24:00.560 --> 0:24:03.920 like this, and the supply will be will not meet 0:24:03.960 --> 0:24:08.520 demand indefinitely. But that's I think physically impossible. 0:24:08.800 --> 0:24:10.640 Yeah, normally at this point you would stop to see 0:24:10.640 --> 0:24:12.639 a shot rise and supply and the fool of and 0:24:12.680 --> 0:24:13.880 demand at the same time. 0:24:13.920 --> 0:24:17.920 Or all the market needs to perceive is that their 0:24:18.119 --> 0:24:21.840 supply is planned. That's all. That's all it'll take to 0:24:21.880 --> 0:24:23.199 tip them back the other direction. 0:24:23.600 --> 0:24:25.320 Well, the always us to say, I remember someone telling 0:24:25.400 --> 0:24:28.480 me when I was a very very junior broker that 0:24:28.640 --> 0:24:31.720 the chip makers you seld them when they were very cheap, 0:24:31.760 --> 0:24:33.440 and you bought them when they were very expensive, because 0:24:33.480 --> 0:24:34.560 that's how the cycle looks. 0:24:35.240 --> 0:24:39.359