Picks, Shovels, and Street Sweepers: Investing in the AI Boom and Beyond

The world is captivated by the artificial intelligence boom, with hundreds of billions being poured into infrastructure. But is it a sustainable revolution or a speculative bubble? In this episode, Brunner’s Julian Bishop and James Ashworth provide a sober analysis based on their research trips to the tech hub of San Francisco and the industrial centre of Tokyo.

They draw parallels to past technological booms, questioning whether today’s massive capital expenditure can generate a sufficient return. Discover their balanced strategy: investing in the essential ‘picks and shovels’ of the AI gold rush while also finding value in resilient, tangible businesses far from the hype. This episode offers an insight into navigating one of the most important – and hyped – investment themes of our time, exploring how to uncover opportunities, while remaining alert to the risks that could be present.

This is a marketing communication. Please refer to the key information document or KID before making any final investment decisions. Investing involves risk. The value of an investment and the income from it may fall as well as rise and investors might not get back the full amount invested. Past performance does not predict future returns. The mention of any particular security or strategy should not be considered as a recommendation. For further information on the Brunner Trust please go to www.brunner.co.uk.

JB: Hello and welcome to the podcast from the Brunner Investment Trust. My name's Julian Bishop. I'm the current lead manager and I'm joined today by James Ashworth.

JA: Good morning, good afternoon.

JB: So, James, we've just been on the road, haven't we?

JA: Yeah, it's back-to-school season, just for us now, early October, September's a month full of conferences and events, and as you say, we've both been on the road. Maybe we'll kick off with you telling us a little about what you've been up to and your trip to Asia last month.

JB: Yeah, so I went to Tokyo. I hadn't been for a while, not since COVID actually, so it was a good opportunity to catch up with management from a bunch of Japanese companies. By the way, I would fully recommend a trip to Tokyo at the moment. It's a very long flight, it's 14 hours because you can't go over Russia. But once you're there, it's wonderful, and at the moment it's very cheap. I'd say taxis, restaurants, hotels, all about half the price of Europe, which is pretty remarkable, given that it used to have this reputation for being very expensive. So that's my travel advice, but when we were there, we saw a bunch of companies, one of Brunner's Holdings is a Japanese company called Itochu. Itochu is a trading company. So really, it's a portfolio of businesses. It's one of the few Japanese businesses, in fact, one of the few businesses outside the United States, that Warren Buffett's Berkshire Hathaway owns, and it's a portfolio of businesses. So, for example, they own Kwik Fit in the UK, so if you ever get your tyres changed at Kwik Fit…

JA: You can’t get quicker than a Kwik Fit fitter!

JB: That’s what they say!

JA: Other tyre changers are available.

JB: Yeah, exactly. Other tyre changers are available. But they also have a stake, for example, in BHP Billiton's Australian iron ore business, which is one of the great metals resources in the world, very low-cost iron ore reserve in Western Australia and the Pilbara region, so they own parts of that. If you've been to Japan, you'll have seen these ubiquitous convenience stores.

JA: Yes, C stores are everywhere in Japan, aren't they?

JB: Literally every street corner will be a convenience store, and they own one of the most prominent chains called Family Mart, and they own a big food distributor, and they own lots and lots of other businesses.

JA: A real sort of ragbag collection. It's not really a ragbag, that's probably an unfair way of saying it, but it's a very diversified selection of assets.

JB: Exactly, and you know, trades low double digit, multiple, lots and lots of cash generations. So that's a great holding for us, so it's good to see them. Plus, a few other companies that we're thinking about. So, typically, when we want to invest in a new company, we would generally prefer to meet them. So, we saw a couple of companies that were on our to-do list. We've often said that we turn over a lot of stones, we're always kissing frogs, looking for new ideas, and this is a great way to do this and the Japanese market, it's interesting, I mean, it's not a particularly expensive market, sort of lacks the glamour of the United States. Japan's obviously got some demographic problems, etc., but it has some world-class businesses, particularly in engineering and certain elements of technology, so that was really, really interesting. And then James, so you've been to the States.

JA: Yes, so I went the other way. You went east and I went west.

JB: Yes. So, tell us about that.

JA: I spent a couple of weeks in San Francisco, meeting all sorts of companies, but obviously with a very tech-biased sort of schedule. Clearly, the West Coast is known for those incredible high-tech businesses that we all know. I spent a lot of time meeting software companies, tech companies, and we had lots of discussion around some of the disruptive changes that are happening in the world, particularly AI, which was the theme of the conferences that I was at and clearly this is a very exciting new technology. Silicon Valley is renowned for creating new forms of technology and AI is the wave at the moment. So, huge amount of focus on AI, but also got to see some other companies that we have invested in or that we're looking at as well. So, like you, it was a really good research trip. And I think one of the key takeaways from that was everyone's very, very excited about AI and how this is going to be a revolutionary technology. We've seen things like ChatGPT really ramp up incredibly quickly over the last couple of years. I think they've talked about having 700 million average users, and this is a product that is less than 3 years old. So, it's really had really rapid adoption by users and there's lots of questions about the impact it's going to have on society more broadly and the roles. What jobs people will do in a world where AI is capable of doing lots of different tasks. And it's an interesting thing for us as investors as well, because it's become a big focus for the market. Clearly, the hyperscaler companies, they're called hyperscale, so we're thinking Amazon, Google, Microsoft, Meta, which used to be called Facebook, spending a huge amount of money on AI infrastructure today. If you look at the largest 5 spenders, they're forecast to spend $370 billion just this year. That's up nearly fourfold since just 2020. It's a huge amount of money and it's expected to carry on rising, and it's obviously been a key factor for our fund and the performance as well, so, maybe that's a good maybe segue into how we're thinking about AI and maybe Julian, you want to explain a little bit about our views on AI. How we're reacting to this and the investments that we're making in this space.

JB: Yes, it's a difficult one. AI has a huge amount of focus in the investment world. Huge impact on markets. So, I think it's fair to say if you look in the States, in particular, but if you look at semiconductors globally, we've seen some huge moves in very, very large businesses as people have seen the outcome from this extraordinary amount of expenditure on AI. I think that's an exciting thing, but it's also slightly concerning. We live in the financial world, so we're used to talking about very big numbers, right? But the fact that a handful of companies are spending almost $400 billion on infrastructure for artificial intelligence. And let's be clear about what that infrastructure is, that's largely semiconductors, a lot of them from Nvidia, so graphics processing units that are used in parallel computing, plus all the electrical gubbins that goes into these and into these huge data centres, and when I say huge, they're extraordinary. One thing I would encourage listeners to do is go and type in, Facebook data centre in Google and there's a picture of these new data centres, and one of them, in particular, overlaid on Manhattan.

JA: Yes, that's an incredible picture.

JB: And it's half the size of Manhattan. So, the footprints of these are extraordinary. And these chips, they cost tens of thousands of dollars - I think it's about $10,000 each - they consume a vast amount of electricity, so literally as much as a kettle. So, the electrical requirements of these sites are absolutely huge, you're talking about the electrical requirements of a single data centre being the same as a large town, for example. So, this is where the AI spend is going, and what I find interesting is that this spend is very, very different from the tech companies of the past, because the actual amount of money it costs to build Apple, to build Google, to build Facebook was not actually that much asset-like businesses have pretty low capital requirements. The bulk of the value came from their network effects, their brands, their intellectual property, their programming prowess, etc. etc. Whereas here, we have a technological revolution that is proving very, very expensive. And then the concerning thing is that whilst there's all this spend happening, the actual revenues from AI, if you ignore the hype for one second, if you look at the actual revenues, the sales of AI companies so far are very, very small. So, why don’t you move on to that?

JA: So, I think that's an interesting point, you do need to separate here, the amount of investment that's being spent on the infrastructure, which is building the data centres, we're talking, as we said, nearly $400 billion. It's such a large amount of money actually, that you can see it in the US GDP numbers. There was more than a percentage point of GDP growth in the first half of this year comes just from building data centres.

JB: I think I saw somewhere that actually if you excluded the impact of data centre construction, the US is almost in a recession.

JA: Yes.

JB: It was very, very low GDP growth, excluding that.

JA: Yes, it's absolutely huge, and I think that's one of the things that's a bit strange here with AI, it's actually, at the moment, it's not AI changing the world. At the moment, what is really happening is it's the building of the infrastructure that is changing the world, that's what's driving Nvidia, that's what's driving the GDP numbers.

JB: Yeah.

JA: So, to a degree, we haven't really seen the impact yet of AI revenues. We haven't seen it really have a hugely disruptive impact on most jobs, maybe that's to come, or maybe it isn't, and maybe that's a good segue. You know, we're talking about $400 billion a year of CapEx. How did they justify this? How does the maths work? Is there going to be a good return on this investment, do you think?

JB: Well, that's the sort of key question that everyone's struggling with. So the AI sector has some very, very articulate acolytes, you know, you have Jensen Huang at Nvidia.

JA: Nvidia’s CEO.

JB: You've got Sam Altman.

JA: At OpenAI.

JB: Yeah, and it is a pretty incredible technology, and anyone who's sort of messed around with ChatGPT tends to be pretty impressed by its output. Now what the usefulness of it so far is slightly open to debate. There are clearly some areas where it will be useful. We've seen it already in things like coding, so just increasing the efficiency of computer coding, which theoretically makes it easier to write software, you know. So, that's one sort of obvious use case, but beyond that, so far the real world use cases - beyond novelty in helping people with their homework, etc. - have actually been shown to be quite limited. So, at the moment you have this incredible CapEx binge. You have this incredible technology. You can understand why people are investing in it, but so far, the bulk of that spend, I would say, is speculative. And anyone who's sort of looked into financial market history or indeed economic history, would have seen that these waves have happened again and again in history. So, I mean, most recently, obviously there was a dotcom bubble.

JA: Yes, where we had a huge investment in CapEx and building capacity for internet usage. I think Global Crossing at one point was spending, you know, billions and billions of dollars a year and basically building out…

JB: The physical.

JA: Yeah, the physical internet and you know, we know how that story ended.

JB: Yeah.

JA: So, you know, WorldCom, you know, these businesses didn't survive that era.

JB: Yeah.

JA: So, that's the dotcom example. Do we have other examples?

JB: Go back in time, you can see there's clear elements of this in, for example, the railways, so there was a huge sort of stock market bubble around railways when they were first built.

JA: We're way back in the 1700’s here, I think, aren't we?

JB: I think sort of 1850’s, wasn’t it?

JA: Was it 1850’s? OK.

JB: Or maybe a bit later even, but it was these American railroads being built out across the west and so on and so forth.

JA: And in the UK as well, I think they had a similar railways and buses in the 1850s, I think.

JB: Yeah, and they were obviously very capital intensive again. Obviously very revolutionary, but the people who actually built it didn't really do that well. Ultimately, there was obviously a big sort of bust after the boom. And so, yeah, anyone who sees this amount of speculative capital expenditure going on, obviously - and knows about economic history - will see a red flag there, I think.

JA: I think that's the interesting point there, that we do have these revolutionary technologies repeatedly, actually, over the last sort of 300 years.

JB: Since the industrial revolution, yeah.

JA: And in many cases, actually, the beneficiaries are sometimes the beneficiary's basically the consumer, and very few firms make lots of money. I mean, aircraft is a great example, and Warren Buffett gave a stat a few years ago that the cumulative profits of all airlines since the Wright brothers' first flight to Kitty Hawk had been 0, you know, airlines in aggregate had made no money.

JB: Wow, ok.

JA: But it's obviously fantastic for, you know, humans to be able to jet round to San Francisco or to Tokyo or to go and see the world, but really it wasn't profitable for firms.

JB: Yeah.

JA: And we've had similar things with the dotcom era, right. The beneficiaries of all this CapEx that was spent building out the internet, the profits didn't accrue to Global Crossing and to WorldCom, they accrued to Facebook and to Google and to Netflix.

JB: Most of whom weren't formed until several years later, so I think, you know, Facebook was founded in 2004 or something like that.

JA: Google was 1998.

JB: But that had only just got going and then the iPhone, of course, was 2007?

JA: 2007, yeah.

JA: So, you're right, it took several years for the business models to emerge that really took advantage of that technology and the really big profits didn't emerge until well after the technology was invented, and it's possible we see something like that here. So, we did some maths, didn’t we.

JA: I was going to say, what do the numbers look like? I mean, there's clearly, we don't really know what the… how much investment's going to happen or how much revenue's going to happen.

JB: No.

JA: We do know what the investment is today, which is roughly, you know, close to $400 billion for some of the hyperscalers per year, and that's expected to grow. I mean, maybe just maybe just sort of sketch out how much money's likely to be spent on building AI and how much revenue do we need to earn to justify that level of investment?

JB: Well, so this is the maths that we did. So, if consensus forecast for 2026 CapEx on AI is about $450 billion and then it's expected to grow from there, as you said. So, if you just add up that sort of total by 2031, there is an expectation that over 3 trillion dollars.

JA: Trillion with a T.

JB: Would have been spent on AI CapEx. And 2030's not that far away, you know, this is 5 years away. And you need to see a return on that CapEx. These businesses need to see profits from that CapEx, that's how you judge how good a business is. And so, we did some maths, and we think that by 2031, you need to see AI revenues, so this is the revenues of people like ChatGPT of - and this is a very, very wide range, because obviously this is subject to a bunch of assumptions, but between $2 and 4 trillion dollars.

JA: Yeah.

JB: A year.

JA: Which is a big number, really big number.

JB: So, where do you think OpenAI's revenues are today?

JA: I'm not sure I'm able to make a good guess here because I'm looking at the script. I've got a number down here.

JB: Ah, OK.

JA: But the numbers today are very small, you know, it's a handful of billions, I think, isn't it?

JB: It is. So, I think they reported that their first half sales were about 4 billion - this is OpenAI - and at the moment they have run rate, their annual run rate is about $13 billion. So, remember, we need to get to 2 to 4 trillion by 2030.

JA: Right, so it's a huge increase from the current level.

JB: It’s an absolutely vast increase from this current run rate. Well, no, let's not be too much of a doubting Thomas. I'm sure there will be growth in revenues from people subscribing to AI tools and businesses using them, and so on. But $2 to $4 trillion is a vast amount of money, and the stats that I find interesting and makes me very worried about whether this level of CapEx is sustainable, is that 2 to 4 trillion, if you were to add up all the revenues of Microsoft, Google, Apple, Facebook, Netflix, Amazon Web Services. So, literally the most…

JA: So, this is almost adding up all of the internet effectively.

JB: Yeah.

JA: We, as consumers, see.

JB: The most ubiquitous companies on planet Earth. So, Google - billions of searches a day. I think Microsoft has 400 million paying users for Office. Meta - billions of data users.

JA: Apple has 200, sells 200 million iPhones a year or something.

JB: Yeah, Netflix is almost a majority, I believe, of households in the developed world.

JA: Yes.

JB: Their combined revenues are 1.4 trillion this year. So, we need to get to a number that's twice what all those businesses do in 5 years from a starting point of 13 billion.

JA: That's going to require quite some growth, quite some adoption. I think the stat that really blew my mind was that if we hit that $4 trillion of revenue, that'll be $500 per person for every person on Earth, pretty much, which is a huge amount of money. I mean, if you subscribe to ChatGPT, it's something like $20 a month. So even if you are a subscriber, you're only paying them roughly $300 a year. And…

JB: Yeah.

JA: So that is nowhere near enough, you know, people need to be consuming more AI than every person on Earth subscribing to ChatGPT.

JB: Exactly.

JA: Which is quite a mind-blowing number.

JB: I think, you know, at Brunner, we're trying to be balanced, you have to look backwards and you have to look for real-world profits, we want cash flow. But we also don't want to have our head in the sand about incredible new technologies. So, we want to be opportunistic. We don't want to be naysayers about everything that emerges. But we also want to be rooted in common sense.

JA: Right.

JB: And we also know, probably more than average, about the stock market history. I think it's probably both of our opinions that there are reasons to be a little bit worried. Now, if you look at our holdings, we don't own Nvidia, which with the benefit of hindsight was a mistake because whilst there could be a big downcycle if CapEx does start to come down, Nvidia's profits will also go down and the stock will probably get crushed, but obviously it's an incredible company, it's very, very profitable and, three or five years ago, we should have owned it. But we do have plenty of other names in the semiconductor space that are related to this theme. So, why don’t you sort of just mention very quickly some of our holdings in that space?

JA: One of our largest holdings actually is in Taiwan Semiconductor, known as TSMC, which is one of only three firms that's actually physically capable of manufacturing the most advanced semiconductors. So, they're the chips… they manufacture chips for other people, so when Nvidia designs a chip, Nvidia doesn't own any factories, so Nvidia needs someone to manufacture the chips. They go to TSMC. Similarly with Apple, if anyone has an iPhone or an iPad, the chips in there, the processors in there are also manufactured physically by TSMC, and that's a very large holding for us, and it's not all about the AI theme. They have a huge amount of revenue from selling chips for smartphones. They have a round of revenue selling, even older chips for, you know, Intel process or CPUs, old-fashioned processing chips.

JB: I believe, actually, that Taiwan say they make pretty much every one of the main chips for Apple. So, I think the new iPhone, the A19 chips, rumoured 30 billion transistors. 30 billion on and off switches and in less than 1 inch square, which is pretty remarkable. They're the only company on the planet that are capable of manufacturing chips of that sophistication.

JA: So, TSMC are a beneficiary of the growth in AI and the growth in AI spend. Clearly, you know, Nvidia is now a fairly substantial customer for them, I think it's approaching 20% of their revenue, but we're not relying solely on the AI theme. We've got a business here where they've got a, it's a fairly diversified sort of business base, as I said, selling chips for Apple as well as others as well. And it has a clear competitive advantage in that it's only three people on planet Earth who are capable of manufacturing these chips – that’s Intel, Samsung and TSMC, so we've got a business where it's a bit diversified, but it's still a beneficiary of this theme. And our other two holdings in this space would class similarly, so we own Amphenol, a US business that makes among many other things, some interconnectors and sensors that are used in data centres, they're used in many, many other applications as well, but their performance over the last year or two has really been driven by rapid growth in spend of data centres. And similarly, Schneider Electric, who make lots of the high-voltage and low-voltage power equipment that's needed for data centres and for office buildings and all sorts of other things, has also been a big beneficiary of all the investment into AI data centres. In all these cases, these are what we think are very good quality businesses on a stand-alone basis and, even in the absence of the AI theme, they would be businesses that we would be interested in and for us, the AI theme is, you know, is gravy for these businesses, it's just an additional tailwind.

JB: I would argue as well, I mean, probably the outcome's a bit more controversial, but we have quite a few other businesses where there is a degree of exposure to AI. I mean, we have Microsoft.

JA: Absolutely.

JB: Microsoft Azure hosts a lot of AI applications. It's pretty convoluted but we think they earn about 1/3 of OpenAI, and they host a lot of their activities. Amazon, ditto, hosts a lot of AWS and of course, Google, which I would say, after ChatGPT, it's Gemini that's emerging as the sort of second-place AI company when you look at usage statistics. So, it seems like it's becoming a bit of a sort of two-horse race between Google and ChatGPT, with maybe Claude Anthropic being used in businesses for things like coding.

JA: Yeah.

JB: That seems to be the way that it's going. So, it touches on a lot of our investments.

JA: Yes.

JB: But I think what we're trying to say here is that we're a little bit worried and we're mindful of stock market history. We're mindful that this could be a big boom that proves not to be sustainable, and we have that at the back of our head at all times.

JA: I think one interesting quote maybe to leave this topic, was one of the presentations I saw in the conference in San Francisco was from Sequoia, who are a very big venture capital firm, investing in these startup businesses in this area, it's their bread and butter, and one of their partners was on stage, noting that they'd been saying that AI was probably a bubble for about 18 months. There is a widespread view that valuations are increasing very rapidly, a huge amount of money is being invested into AI. The returns are as yet unclear, and to a degree, speaking with other investors at the present at the conference, my general feeling was that people are trying to work out whether this is 1997 or whether it's 1999.

JB: Right, so clearly, we've got a year or two to go, or whether they're on the cusp of the…

JA: Have we just had the Netscape moment, which sort of famously kicked off the internet boom that then lasted for another few years, or are we very, very late in the process? It's unclear, and there's obviously lots of uncertainty, but maybe this is a good segue into some of the other things that we've been doing more recently, some of the more recent investments we've made where, we've looked at businesses which are perhaps less exposed to the AI theme and more defensive in terms of their fundamental characteristics. Less at risk of disruption, and we touched on some of them actually in past podcasts. We talked about Tesco in the last podcast and…

JB: Yeah.

JA: However, however good AI gets, it's not going to be able to manufacture groceries for me. Similarly, we talked about Kia, one of the world's largest automakers. AI might be very good, but it can't get me physically from A to B. Maybe talk a little bit about what else we've been doing recently and some of the thinking behind those decisions.

JB: Yeah, so there’s investing in AI, there's also what AI means for other businesses, and I think generally there is a sense of nervousness around anything whose business is being constructed around intellectual property, because there's this fear that if you automate intelligence and if intelligence becomes mega abundant then anything that is relied on intelligence to build, it becomes a little bit fragile. So, these sort of names that were previously admired companies, asset light, information rich, have been out of favour, I would say. And I think I can understand why, if it's suddenly much easier to write software code, are the barriers to entry for software companies coming down? And I think that's a reasonable conversation to have, and there's a debate there, whether it's true that it means that software companies are suddenly going to be irrelevant, I doubt. But you can understand why people are having that conversation.

JA: And that’s why people are worried.

JB: Yeah, exactly. So, I think it’s about increasingly finding solace in the physical world. So, you know, an iPhone replaced lots of different things, but it still can't make toast.

JA: Right!

JB: So, where can we find businesses that will indisputably not be affected, and so, we've recently bought a couple of new holdings that I think are reassuringly physical in their nature. You mentioned Tesco and Kia and I think they are good examples. But in Japan, we have MonotaRo, so this is one of the companies I saw on my travels. So why don't you tell us a little bit about MonotaRo, if you don't mind, because you know about it too?

JA: So this is a good example of finding a great growth business in Asia. So, MonotaRo is, effectively a distributor of maintenance repair and overhaul parts. Essentially what they're doing is selling bits and pieces to business customers in Japan to fix machines to keep their operations running. It has a huge catalogue of items, about 20 million different parts are available, and, if you order from them, they'll get it to you next day. It's a business has historically been very much done off line. If you're running a little factory in Japan and a part of your machine breaks, you go and run round to a physical venue and try and find a replacement part. They probably wouldn't have 20 million in stock like MonotaRo does, but that's the route you would go. But this business is becoming increasingly online, and MonotaRo's done a great job, growing from the small and medium-sized enterprises up into the larger businesses, over the last few years. It's had a very good history and track record of growing, well above 10% a year, they're medium-term guidance that they give the market is that they expect to continue growing in the mid-teens range, so maybe 15% or so. It's a really strong growth business, and one thing that we particularly liked with MonotaRo, is that the largest shareholder is a US industrial distributor called Grainger, who has an exceptional track record and exceptional history. One of the things that lots of investors struggle with in Japan is the balance sheets and lots of these companies run them with huge amounts of cash, they're not very shareholder friendly. However MonotaRo having Grainger as their largest shareholder, keeps themnvery disciplined, makes them a little bit western, if you like, in how they manage their balance sheet and how they return capital to shareholders.

JB: Yeah.

JA: So, that's a really good, a nichey business in a sort of single market, effectively a long way from here, and regardless of what AI does, machines are still going to break, they're still going to need repair and maintenance and MonotaRo will be there to serve their hundreds of thousands of customers in Japan.

JB: So, it's basically like an Amazon for Japanese business?

JA: It’s absolutely an Amazon for Japanese businesses, but, you know, the parts they're selling are not really things that you would find on Amazon because it's spare belts and plugs and sockets and Amazon wants to sell books and DVDs and things like that.

JB: Yes, exactly.

JA: Maybe, you could talk about another one that we made, which was actually in the US where it's definitely something that AI can't do, and that's the investment we made in a company called Federal Signal. Now tell me what Federal Signal is.

JB: So, Federal Signal, yeah, it's a company which makes speciality industrial vehicles.

JA: What's a speciality industrial vehicle?

JB: So, examples of a speciality industrial vehicle would be, for example, street sweepers.

JA: OK.

JB: Road line painting machines, and one which I particularly like is sewage vacuums.

JA: OK, so this is something that really can't be done by AI.

JB: No, so if you have a fatberg lurking under your road, one of these guys would come down with a sewage vacuum and pumps and blasts it away, and biologically this is probably going to remain a sort of necessary feature of municipal life. I would say.

JA: Unfortunately.

JB: So, they sell a lot to municipalities to utilities, water utilities and so on. What's great about these businesses is they're quite small end markets, so they’re niches, they're number one in pretty much all of their end markets, so they have a good R&D budget. They can come up with innovations that are useful for the user. I feel slightly sorry for the user, but they make the user's life easier; they're more effective.

JA: Right.

JB: And very few other people are really participating in these end markets so they can eke out good levels of profit, so another sort of real gem of a business. I think both MonotaRo and Federal Signal probably show the benefits of us travelling. We hit the road, we go and see as many new businesses as we can, and sometimes you find these things that are relatively small, maybe medium-cap, but just tick every box. I mean, in the example of Federal Signals, so niche, it allows it to grow, strong market position, great cash flow.

JA: Very defensive, presumably, because, you know…

JB: Yes.

JA: If you're a municipality and you've got some road lines that need painting, it's not really a discretionary…

JB: Exactly, and if the sewer's blocked, you've got to unblock it. So, we've talked about AI and this fear of a down cycle in AI capital expenditure and the history of semiconductor expenditure is distinctly one of massive cycles. So, at a time where you might be a bit fearful about down cycle in some of these sort of sexier investments, we think actually unblocking sewers is the way to go.

JA: Well, what a note to leave the conversation on!

JB: So, thank you very much for your attention and thanks for listening. I'm Julian Bishop and this is James Ashworth. We're from the Brunner Investment Trust. Hope you hear us next time as well.

JA: Thank you. Goodbye.

JB: Goodbye.