The Capital Cycle Podcast - AI Disruptees

Episode Date: May 29, 2026

Opportunities among companies deemed AI losers. Edward Chancellor talks to Charles Carter, a European Equities Fund Manager. For more information, or to access select articles from Marathon’s Global Investment Review publications which accompany this podcast series, please visit www.thecapitalcycle.co.uk Hosted on Acast. See acast.com/privacy for more information.

And welcome to another episode of the Capital Cycle podcast. This is Edward Chancellor, and I have with me Charles Carter, who's a European portfolio manager of Marathon. Welcome, Charles. Great to see you, Eddie. So Charles, we're going to talk about the threat or the potential threat to certain businesses by the arrival of AI and whether you believe in some cases the market is overdoing things. And in general, in the latest global investment review piece you've written, you point out that in very speculative financial markets, and let no one doubt we are in a very speculative financial market, players often make us beeline for technology stocks and other sectors are left for dead.

There is a sort of crowding out effect from technology manias. And I wrote on a similar theme in a column quite really, recently, which I pointed out, that the stock market has a pretty abysmal track record when it comes to pricing winners and losers during earlier tech bubbles. Yes. The game, if you like, have spot the winners and losers was a very popular one in that period. You mentioned that the TMT boom of the late 1990s, when the new economy winners were sort of lauded over the old economy losers. I remember an advertisement that appeared in August 2000 from

Robertson Stevens, which was an investment bank that specialized in TMT stocks. And it ran this full-page advertisement in the Wall Street Journal and the Financial Times with just the words, Thanks, old economy, we'll take it from here. And of course, that was at the point of maximum hubris and I think the company folded within a couple of years after that but it did illustrate I think quite well just how the market can get things wrong when it is in this speculative frame one could also point out that that was virtually to the day or month that old economy stocks started putting in a fantastic performance while the so-called new economy stocks tanked give us a bit more on the market's ability

to spot losers or poor ability to spot losers. There was a piece recently in the economist, I think, called Technomyopia, where they looked at 80 cases going back over 20 years through the data where sectors declined by more than 20% relative to the market over a three-month period. And they looked at that to see to what extent the market got it right and whether or not the sectors continued to go down or indeed recovered, what they found was that the market was right approximately half the time, which of course is not actually much better than just a simple coin toss. So it's not necessarily very good at predicting these secular decline problems, although obviously there will be cases and Kodak and Polaroid, one goes back through the past,

there will be secular losers. But the market's just not very good at spotting them actually. when you look at the ones that it misses. So you think history could be repeating itself? There's a chance, I believe, that there are companies now which have been beaten up for reasons to do with AI, which may indeed recover strongly because they have assets and management teams that can adapt to the new scenario. We discussed in the February podcast where the AI was eating software. What about Marathon's exposure to the idea of Sathpocalypse? We haven't had much exposure in our European portfolios to that theme. The big company in Europe is SAP, which has been suffering.

And the jury is still out as to whether or not they can really leverage their asset in terms of the data that they have on companies and their position as a system of record in the corporate ecosystem, whether they can leverage that to, at least. actually be a winner in the AI world. So there's a big debate about that. But where we have suffered, in particular, has been through the ownership of some more data-oriented companies from companies that are involved in online classified advertising, where historically, these businesses have had strong moats around them as a result of these unique data assets, proprietary data sets, and then, in some cases, network effects. So could you give us some examples? examples from the portfolio. So a couple of must-have information companies that we've got in the European portfolio would be RELX, the old Redel-Sphere business and Experian. Tell us a bit about those

businesses. Experian's main business is in the world of credit. So they manage a number of credit bureaus across the world where they get lots of data from banks who give their data for free, so-called contributory data. And then they mix that with data from lots of other sources to provide their financial service customers' insights into the credit worthiness of potential customers. Instantly, Experian, in the late 1990s was owned by Great Universal Stores, GUS, correct? That's correct. And that was one of the beaten down old economy stocks of the period that came roaring back after 2000, I remember. In fact, I think the credit business that originated in Gus was a result of the business

that they had in home shopping, where they would basically lend credit. to their customers to buy their clothes or other items. And then Relix, that comprises its main business now being a risk business, which has been put together from acquisitions over the last 20 years or so. They're involved in gathering data on particularly auto insurance risks, which they then sell to insurance companies. They're also involved in that risk area in things like cybersecurity, identification, software, and various other things.

And then they have the old Elsevier science business, which comprises a number of leading journals, which academics want to be published in. Free content. Free content, which they then sell to various parties, both on the academic side, but also into research and other areas. And then the final piece is their legal business, which houses the Lexus NECD,

Nexus company, which is one of the two big data holders for case law, particularly in the US and in the UK, competing with Thompson Westlaw. We'll get on to the Lexis business in a second, but your argument, and I've seen this elsewhere, is that in the age of AI, proprietary data could actually become more valuable rather than less valuable. Yes. You've seen that a little bit so far with Relix's legal business where it has been adding AI tools such as Protaget to the LexisNexis database. And that's been enabling them to see greater usage of the database. And that allows them to increase pricing. So actually the revenues and profits in that area have actually been increasing in the age of AI so far. The question I think a lot of people will have is whether or not

not the advent of a lot of new entrants into this field, such as Harvey, Ligora, and Latterley with Claude making a big push in this area, how that's going to create more competition in the sort of analytics side of the business. But I believe that the core data business is very hard to replicate. And there are these thorny problems that people, on the whole, try to ignore nowadays, which is the question of AI hallucinations. And a few weeks ago, the top U.S. corporate law firm Sullivan and Cromwell was forced to apologize to a federal bankruptcy judge after using generative AI in its court filings that contained

hallucinations and fabricated dozens of case citations. So that is a potential problem of using AI. for legal purposes? Yeah, on the surface, legal is ripe for the taking in that so much of the data is publicly available. So cases are published. The question is trying to understand which is the case that prevails in Anglo-Saxon law. You need to do a lot of work to make sure that the database, the data set that you have, is very accurate. This comes to the point of the sort of probabilistic way which AI works when scraping through a lot of data using the deep learning techniques

compared to what Reid-Elfphere is doing, which is adding a lot of value to the data in a way that creates a more deterministic type of analysis and it's therefore less prone to error. And what about the experience business with its data analytics? There, I think the data, as I say, is contribute to data. so it's not actually available on the internet. So it's much harder to replicate. To scrape and steel.

To scrape and steel. And it's also numerical. I mean, LLMs are very good at language, obviously, and may, of course, get much better when it comes to data. But it is numerical data, and it tends to be passed through to the consumer on a kind of machine-to-machine basis. So you don't have this risk that you have fewer analysts

working at customers who may be losing their jobs, And the old per seat model that you have in some cases in the services industry is undermined. And you also write about the impact of AI on some online classified companies that you own in your portfolio. Yes. So companies like Rights Move in the UK, which is the country's leading real estate portal, similar business in Germany, Scout 24 that we've acquired a stake in recently. and then Auto-Trader, which is the leading portal in the second-hand car business. These are companies which have created these three-sided marketplaces. So they are the places where customers go because they believe that's where they'll find the inventory. And it's where the estate agents or the car dealers will want to place their inventory

because they know that's where they're going to see the maximum number of customers. So there's been this network effect which has created monopolies in many cases in these vertical markets. And the market is thinking that these online classified businesses are about to go the way of yellow pages. Is that correct? Yeah. So the idea would be pure disant mediation that you would, instead of going to the right move

or the auto-trader website, you'd go straight to your agentic AI interface. It knows you very well. It can scrape all of the internet, including the online portals. So in theory, it could have a richer data set, and it can then create options for you. You can ask questions, then narrow down your search. And so why would you need to ever go to one of the vertical sites? That's the hard risk on disintermediation. The kind of perhaps softer risk is that they stand at the kind of top of the funnel, as it's known in the industry.

It's the place that you go to the outset. and it kind of undermines the value of what the verticals are doing and they become more kind of dumb pipe. And then I think there are some other risks around how it forces some of these very profitable businesses to have to invest more either in their brands or in their own AI tools to defend their moat. I was talking to the investor and author Sandy Nann

the other day, read this great book, Engines that Move Markets, about the impact of new technology. on markets. And we talked about AI and one of the points he made, and I think you hold this view, is that as things currently stand, the large language models and the businesses that are trying to create around it currently don't have the domain expertise that companies like Right Move have developed over the years. And that domain expertise is a source of competitive advantage you think might endure? There's a lot of data in both of these examples of the

verticals, both houses and in cars, where there's more proprietary data than meets the eye. Details about the car's history, its exact specification. It's much harder to scrape all of that information as opposed to knowing the specific identifying number of the car and then all the history of the car, crashes and so forth. So they can build up a lot of expertise around both houses and cars. And then, of course, there's the other side of what they do, which is the value that they bring to their real estate agent and car dealer customers. So, you know, for a real estate agent, if they know that Eddie Chancellor is looking at a new country property, they can identify that he may be looking to sell his existing

property. That's a lead that they can pass on and sell effectively to an estate agent. So there's value in the system from what they do. And similarly, for something like an auto trader, they are doing a lot of work, helping with the workflow and inventory management and various other things to help their car dealers become more productive. And of course, they can use AI tools. It's the chief executive of Wrightman who said it's not rocket science. They can use a lot of these AI tools to add to their vertical expertise to fend off the competition from the more horizontally oriented AI companies. And for the AI companies, the challenge is then one of whether it's really possible to compete with every startup in the world, which has,

as you say, this vertical expertise. Although you don't write about travel portals, I have a friend Jonathan Tepper who wrote about a stock in bookings holdings, which owns a number of travel websites. He said that they have so many relationships with so many different hotels and other holiday businesses, that there is no way that the likes of Open AI can actually get into their business. And Open AI has actually stopped trying to do that, is now sending them traffic. So we will discover whether these businesses go the way of yellow pages. But Yellow Pages, they were tremendously profitable businesses, if you remember,

But they were very simple and it was very easy for the likes of first Yahoo and then Google to eat their lunch and then the entire business. We've had companies in the portfolio where we've taken a good hard look at the impact of AI and decided that they are at risk. So we've exited a number of positions in the field of staffing and recruitment as a result of that as well as certain media assets that look vulnerable in the age of. of AI. Interesting. So you write that those who are standing back waiting for a catalyst to change, waiting for the picture to become more clear on the impact of AI, they're likely to miss out given the way stock markets anticipate things and rallies come very quickly. Yes, there's a sense that one hears a lot about how, because we don't know whether AI or how it's going to develop and to what extent it can become an even bigger threat in the future.

There's no point in holding these stocks while the jury is effectively out deliberating their existential crisis, as it were. My point there would be, well, if you stand around waiting, the chances are when the catalyst comes, you'll miss it or that the market will anticipate it. For us as long-term investors with relatively diversified portfolios, we can make these investments across a lot of different sectors, not knowing exactly when they will bear fruit, but knowing in aggregate or hoping in aggregate that we'll have a better than 50% chance of picking the ones that do actually do well and recover from the problems in the environment at the moment.

And the general picture at the moment is that the market has gone out and shot first and is preparing to ask questions later. Well, thank you very much for that, Charles, and look forward to seeing you again soon. It's always a pleasure. Thank you very much, Eddie. Thank you for your time today. I hope you will listen to the next edition of the capital cycle. This communication is provided for information purposes only. Please refer to Marathon's website and the global investment reviews for further information, including important disclosures.