Rob Campbell [RC]: On today's podcast, Jeff Mo and I start with a simple question: Are we moving from a more intangible economy back towards a more physical one? Jeff lays out the case for why growth may increasingly depend on building things again—whether it's data centers, grid equipment, factories, defense systems, even satellites—and what that means for investors.

Along the way, we'll discuss how those bigger picture themes connect to bottom-up research, where the U.S. midcap team is finding ideas, and why there's no substitute for humility.

[DISCLAIMER]: This podcast is for informational purposes only. Information relating to investment approaches or individual investments should not be construed as advice or endorsement. Any views expressed in this podcast are based upon the information available at the time and are subject to change.

[RC]: Jeff, great to see you again.

Jeff Mo [JM]: Great to see you again, Rob.

[RC]: Well, let's dive right in. I heard two interesting sound bites at some of our recent research team meetings in the last couple of weeks. The first wasn't from you, but from one of our colleagues, Sandro, on our global credit team. He reported that producing one unit of GDP today requires about a third as much oil as it did back in the 1970s.

The second statement, though, was from you. It was a thought that the economy might be entering a phase where there's a higher amount of CapEx needed per unit of GDP growth than we've had over the last couple of decades.

Can we start there? Obviously, your two frames of reference are a bit different. Sandro is really looking backwards. You're thinking forwards. Help us understand the context for why you thought that was important to share with the team.

[JM]: I don't think these two statements are necessarily inconsistent. Oil is a very narrowly focused view on one type of energy, and it's true that, statistically, if you look at the numbers, we do need less oil as a global economy to continue economic growth compared to the past. I think the rise of renewables and, generally, a less energy-intensive economy have probably caused this math to be true. In fact, it's the point of a less energy-intensive economy that kind of brings me to what I said.

Over the last 25-30 years, whatever frame Sandro was using, we've seen a rise in the intangible economy. You had the internet boom, almost 30 years ago now, and with that, the rise of a professional class. I mean, certainly, it started before that, but office workers as a percentage of the total workforce in most western countries is at, or close to, an all-time high right now. You've kind of had a lot of growth that was caught in the intangible part of the economy: intellectual property and services.

If you look at the lived experiences of many consumers, we are also consuming more intangible things, whether that be Netflix, social media, travel, and so on. However, our belief is that that could be changing or at least the reason to make physical investments is starting to get more compelling again. The first one that is already kind of in full force is what we've talked about before, with AI data centers being built out to support the advancements in AI, as well as the excitement and the boom in usage of AI across many consumers and businesses in the world.

I think there are several other themes that are causing the economy to become more of a physical economy requiring more capital expenditures per unit of GDP growth.

[RC]: Let's get into those. I'm thinking back to my Econ 101 days where economic growth is driven by labor, capital, and this productivity measure. If I think back to my textbooks, it was total factor productivity. What I'm hearing from you is that, over the last, let's say 25 years, whether it was the internet boom, there was just a lot that was driven by that productivity factor over the last little while in comparison to either labor or capital. Now, you're saying that capital might play a more central role in driving growth going forward.

[JM]: We may be reaching some of the limits for the internet 1.0 era for productivity improvement. Certainly, I think there's a lot of hope for productivity improvement from the AI models and people harnessing them. But I would say capital is becoming more important because in many cases you need capital to achieve those productivity improvements, whether it's a data center or something else.

Probably the easiest one is infrastructure and we've talked about this for years and years as an economy. There were the bills during the Biden era, IIJA in particular, that really helped to put more money, at least in the U.S., towards infrastructure. A part of that infrastructure investment was recognition that there was underinvestment. But a part of it is also trying to reorient the economy towards a slightly more domestically oriented economy in the U.S.

For the last 25-30 years, there was a view of kind of outsourcing manufacturing—outsourcing physical stuff from the U.S. to China and other lower labor cost jurisdictions. Now, whether it's the rise of robotics, or the COVID supply chain shock that caused people to realize that we need a little bit more productive capacity near to centers of consumption, or just geopolitics, you're seeing more reshoring.

And whether it's building the infrastructure to support that or just straight up building more factories in the U.S., that's causing a rise in physical capacity. It's not just the U.S., for example, they're building more battery factories, auto supply chain, and defense manufacturing companies in Europe, as well.

Generally, throughout the world, it could be a geopolitical reason as well—just more of a home bias to wanting sources of supply closer to the home country.

[RC]: Yeah, a bigger margin of safety.

[JM]: Exactly. A bigger margin of safety and just kind of generally not trusting the other guy that used to make it for you. Chips being a great example of that.
The second one that dovetails to that rising tension within the geopolitical environment is defense. Whether or not this comes to be, the U.S. defense budget is growing and increasingly growing at a rapid pace. NATO countries have made a 5% of GDP commitment, even though they're maybe fudging the numbers a little bit or expanding the definition of what expenditures count as defense spending.

Nevertheless, that's still going to cause a big drive in defense. Certainly, Russia is spending a lot on defense and Ukraine, of course. China is also growing very quickly. Generally, the world is getting to be a bit more of a dangerous place. People are building more physical stuff. There's only so much you can do trying to hack your opposition in a real war.

[RC]: That might actually be an interesting case study for your original statement, which is that defense spending isn't the most productive way to grow your economy, and it is quite capital intensive. If your economy is going to grow, and if that's one of the areas that's driving that growth, it almost by definition requires a lot more investment.

[JM]: I mean, the hope is you make all this stuff as a deterrent, and then you never use it. Certainly, there is economic production of it, but it doesn't actually, unlike a road or something, doesn't add to your productivity of the economy. One area that can add productivity to the economy is the space economy. We're seeing a lot more satellite launches on both spectrums, both the consumer side, as well as the military side. 

On the military side, I think one learning from the Russia-Ukraine war is that signal jamming is a real concern in modern warfare. Many countries are racing to develop more of what's called optical-based communication methods, where you send lasers to a satellite, and then the satellite redirects the message back to the ground again. You need more satellites in space to be able to do that, and so, the major countries that can afford to do that are pushing in that direction.

Obviously, things like Starlink and other low earth orbit satellite constellations are becoming a big phenomenon. Internet connectivity is seen as, especially in war, really important, but also people are realizing, “Hey, this concept of blanketing the earth with thousands of low earth orbit satellites actually gives pretty good connection.” It's additive to productivity, whether it's remote connections, or even, I think most airlines in North America now offer Wi-Fi through a satellite provider—more areas of productivity that people hadn't thought of previously.

All this is driving a higher demand in commodities, maybe ex-oil, but I'm thinking of things like copper. Copper is hitting close to decade-level highs now, maybe all-time highs, depending on inflation adjustment. I haven't looked at the chart that far back. Several other metals, especially more advanced metals, and niche metals, are hitting all-time highs, both in demand and price.

Part of that is just the general need for that material, and especially advanced technologies. But a part of that says we've actually under-invested in physical commodities probably for the last decade because we had a few different things that have caused the economy to not consume as many commodities in the last decade.

The first was after oil had its crash in 2014-2015, when OPEC, for a period of time, decided to pump as much oil as it could. It didn't cause an overall economic recession, but it caused an industrial recession in the U.S., and that dampened demand for commodities.

Around the same time, China started to reorient its economy away from an investment-driven economy to a consumption-driven economy. That is still ongoing and has its bumps along the roads. Certainly, during COVID, China—especially because they were one of the strictest lockdown countries in the world—that dampened the needs of their physical economy.

Generally speaking, through that 2015-2016 era, all the way until 2022, when supply chains really got gnarled up, there was just under-investment in the physical supply chain, whether it's upstream mining, materials processing, or capacity to make intermediary industrial products like bearings, valves, transformers, etc.

Now, as the economy is starting to demand more physical products, you're seeing these kinds of weird, huge discontinuities. Memory being the most extreme example, where the prices of DRAM or NAND is going up at 40–50% a month, month over month, which is just insane price increases. I think this is probably the only time in my career I can think of that I've seen the price of a product go up that quickly, absent things like weather spikes, natural gas, or something that's very technical in nature.

I think this is a theme that could be with us for at least 5 years, maybe the next decade or two.

[RC]: It's interesting, on your comment on memory. My understanding is that it's even changing, not just the pricing, but the way those businesses operate, in the sense of the types of contracts they have with their customers. Whereas it used to be just go in and buy what you need. Now, because of the supply-demand imbalance, there are a lot more longer-term contracts entering into this, which potentially kind of smooths out some of the demand for some of those providers.

[JM]: Yes. With memory, it is an extremely cyclical industry, even before the AI boom. It has its own idiosyncratic factors and these big price swings are somewhat more typical for that industry.

But I'll give you another example. As a team, we started analyzing a company in the electrical supply chain. They make certain products that go into the electrical infrastructure, or the grid infrastructure. This is a company that's had between 0-3% growth for the last decade per year. You're steady at GDP-type growth industrial company. They had about a 15- 20% price increase on their main product in the last quarter. That was a year-over-year number, not a quarter-over-quarter number. Still, for an industrial product that has generally just been a very stable, steady product for the last decade, that's a big price move in a part of the economy that maybe you don't expect as much cyclicality.

[RC]: We've done a really good job of explaining what led to that statement of yours. I do want to come back later to why you might be wrong, just to play the other side of it. But if we take it as, okay, this is one of the big tides that we see, not just starting now, but probably over the last couple of years, I want to ask you about a couple of ramifications.
The first we've kind of touched on already, which would just be cyclicality. If we're in an era where capital spending investment is a greater proportion of GDP growth, should we expect more booms and busts as a result of that in a more cyclical kind of world?

[JM]: That’s an interesting question. Booms and busts are probably more controlled by consumer confidence ups and downs. I mean, the last two severe recessions, so to speak—2020 and 2008—were all due to consumers suddenly pulling back on spending: in 2020, because they were forced to by the government, essentially; and in 2008, because of a sudden lack of confidence that they had no more wealth anymore, because their mortgage was underwater, and so on.

At least in the west, I still believe that's probably what's going to drive cycles. I actually think that capital-driven cycles, as long as investment in capital is driven by medium and long-term views on the productive nature of that capital, will be less boom and bust.

I've heard my colleague, Peter Lampert, who is head of our international equity and emerging market equity strategies, talk about this concept that some of these AI data center-exposed companies, or companies that benefit from the build-out of data centers, are actually an insulator if the economy perhaps goes into recession. For example, if oil prices stay very high because the Iran-U.S. tensions cannot resolve, at least not very quickly, maybe the fact that in many of our portfolios, we own some companies that benefit from AI CapEx and that is a ballast for other, more economically sensitive parts of the portfolio.

The decisions that are being made right now by the hyperscalers and others are due to the fact that there’s a belief that AI productivity will be very high in the medium and long-term. We're happy to make these investments today and continue forward. In some ways, I think the boom and bust could moderate because businesses tend to be an equation: consumption, plus business investment, plus government investment. Technically, plus-minus trade surplus. The government and the business side tend to be a ballast to the consumer side.

[RC]: Should we expect, as well, with more capital expenditure, more investment, does that necessarily mean a higher savings rate that you need, maybe lower consumption? Almost by definition, if you're saying that capital spending is going to be a greater proportion of GDP growth going forward, you're suggesting that consumption is going to be a lower amount. Maybe, just to go to the portfolio, I've noticed that we have reduced our posture in consumer-facing businesses. I wonder if those two things are linked.

[JM]: Yes, absolutely. We have been positioning our portfolio slowly as we find good ideas to do so, away from a more consumptive-driven economy to more of a building or productive-driven economy. I would say that's less to do with the fact that we think in this GDP equation, consumption must therefore fall. First of all, on my basis, some of these companies were looking less attractive. And secondly, the probability of a recession in the shorter-term has increased. Some of that was driven by that.

I think if you take the longer-term view, it needs to come from somewhere. The way it probably balances out is as corporations start to need more capital again, they will demand more capital. I'll give you a great microcosm. The big four hyperscalers—Google, Amazon, Microsoft, and Meta—those were companies, and Apple too, although they're not doing this kind of AI data center build-out.

For years and years, they had just huge amounts of net cash on the balance sheet. And while they still technically have net cash on the balance sheet, most of them are now raising debt in the bond markets. If you follow their CapEx plans for the next 5 years, they're probably going to have a net debt balance sheet by the end of all of this.

Each of these companies, I don't know what their cash balances were, say, 3 years ago. Let's say they each had $300 billion of cash. That was technically in a demand deposit account at a bank, and banks now had that excess capital to lend out. Now, if you go from positive $300 billion to negative $300 billion—just those four companies alone—that's a $600 billion swing in the amount of cash available for the rest of the economy to borrow from.

As the demand for capital goes up, well, the loss of supply and demand, the price of capital goes up. Interest rates go up. And, when interest rates go up, especially American consumers, many of them have a proclivity to consume non-borrowed money, that has to go down. I do think you're right that long-term, the economy shifts from a lower consumer-driven, consumer spending-driven economy to a more industrial investment-driven economy.

[RC]: Can we connect some of these bigger picture ideas with the day-to-day, bottom-up work that you and the team do in the U.S. mid-cap space?

[JM]: Absolutely.

[RC]: Would it be fair to say that there's a bigger picture theme? Does that drive where you’re looking for investment ideas on a bottom-up basis?

[JM]: Yes, it does. I would say idea generation, the primary method at Mawer is still always idiosyncratic. We look at hundreds, thousands of companies every quarter, and our primary criteria and assessment is still: does this company look like it might have a competitive advantage? It might be a wealth-creating company. Does it have an excellent management team?

A couple of stats you can look at in terms of consistent execution and things like that. And might it trade at a discount to intrinsic value? So, I would say that, at the core, is still what I call idiosyncratic idea generation. It could come from any sector; it could come from any place, the idea I mean.

[RC]: You're talking the first cut of the universe, the first shot at, okay, there's a hundred companies. Let's spend a couple minutes on each to figure out: what are the 20 that I really want to dig into?

[JM]: Yeah. And I think that part hasn't changed and will never change. But for example, if after a couple of minutes you come across Company A and Company B and both look kind of interesting, but you read the business description of company B and you're thinking, oh, actually this could fit the theme or how the economy is reorienting a lot better than company A. We then might bias to look at company B first. We might still end up buying company A after our intensive analysis research determines that A is indeed the better investment. There's a lot more diligence to go through after that.

[RC]: Yeah, obviously. That's it. Not just two minutes of buying stocks.

[JM]: But the fact that in that instance, where after two minutes, both of these companies are reasonably similar in terms of their probability of us ultimately buying it, but we now probably bias company B, whereas in the past, maybe we didn't have a bias, for example.

You can see it in our portfolio. We've probably been thinking about the defense theme for over a year now. A company like Northrop has been kind of holding for us for a while. Some of the companies that benefit from the AI data center build-out have made it into core parts of the portfolio, most notably SanDisk, which started as the peripheral part of the portfolio, but through its amazing performance has become a higher weight in the top 10 weight.

Even a company like OSI Systems, they're the world's largest manufacturer of border control or security scanning equipment. They'll benefit a lot from the movement of physical goods. Right now, a big area that they have growth in is cargo scanning at borders. And so as the economy becomes more physical, just more stuff will move around.

Those are some examples of companies that we have already built up into higher positions in the portfolio. But I would say in the bottom half, or the tail of the portfolio, there are several companies that we have found and identified that are more of a general industrial economy, as well. You mentioned robotics earlier. I would say there's a couple of companies that we think have promise there, but in those cases, their valuations today are a little high.

So, they're more in the tail part of the portfolio. Robotics is still more of an emerging theme, but in terms of things like what Tesla is talking about with Optimus, their humanoid robot. I mean, that's not in production yet. And the true value add of that type of machine is still a little bit unclear in the economy. Some of those areas are still emerging themes.

[RC]: You mentioned SanDisk and some of these other companies have done really well as this particular theme has been rewarded. There's a flip side and the way I've heard you and others describe it, and I don't even think this is our term, is this notion of the “bits to atoms” trade. It has produced its share of losers thematically over the last while too. We've talked about that on the podcast with other guests, but clearly, we have some of that exposure within the U.S. mid-cap portfolio as well.

[JM]: Yeah. We will continue to have some of that exposure. For those of you who listened to us speak in the past, our viewpoint when constructing portfolios is one of creating inherent contradictions. So, while we think about all these themes, it boils down to a bottom-up basis, which is, if we believe this company has growth for the next 5-10 years, greater than GDP, when we put that in our discounted cashflow model, this company looks more attractive now, all else being equal from a valuation standpoint, that would cause us to lean in that direction.

At the same time, we will say, if the particular theme you're exposed to is AI data center build-outs, and that’s what's causing your revenue growth to be maybe higher than GDP. If AI doesn't generate the productivity booms, or at least not as much as people expect, then perhaps we need to also buy a company who will do better if AI turns out to not be as transformative as some people think it will be.

An example of what's colloquially termed an AI loser would be a professional service company or a software company, where if AI becomes amazing, it could just create a brand new software program in two hours and be just as good as SAP or something like that. And if you're in that camp, you probably sell SAP and buy something in the AI data center complex, or if and when they become public, the AI model companies.

But our view has always been one of inherent contradictions. We're not trying bet on a theme, even though we spent the entire podcast talking about themes that we are seeing or thinking about. We're trying to keep a balance but not in a kind of head in the sand way. But we're just saying there's a lot of babies being tossed out with the bath water these days in the AI losers’ bucket.

And so, we have kind of picked away at it and we've actually added some companies to the portfolio that fit in that category, as well. Ultimately, our goal is to create a broadly diversified portfolio of idiosyncratic ideas on behalf of our clients. And there should be companies that have inherent contradictions.

If it's a period of time where companies that, let's say, benefit from AI developing faster and faster and they just look really attractive on a bottom-up basis, then maybe the portfolio will have a little bit more of those. But it’s still from a bottom-up stock selection process. And where the top-down kind of overlay comes in is, even if we find all these great bottom-up ideas in AI CapEx or AI beneficiaries, we want to make sure that the size of that bucket doesn't get so large that we become unbalanced. Hopefully that makes sense.

[RC]: I think it does. It also partially helps to answer a question that I promised to come back to, which is, where might you be wrong on this particular theme? And I think you answered this. You just make sure you have the appropriate awareness. You're still looking from a bottom-up basis, but in a world that has become extremely thematic (it seems like the market is moving very much in themes) that awareness is just that much more important.

Jeff, can I ask you one last question before we wrap up, which is if you agree with the premise that over the long-term stock prices should follow genuine wealth creation, what does this all say about the ability to generate wealth as an economy for a standard of living, but just in individual investments looking forward?

[JM]: At an economy-wide level, we're still pretty optimistic for the economy for the future. Many have described this AI boom as the most significant boom since the internet boom. At the time, it seemed gimmicky—anything with a “dotcom” attached there, the corporate name was suddenly trading a billion dollars higher. Then, there was a crash and there was a lot of, }I told you so”, that was not very helpful. But 25 years later, pretty much everyone in the world is like, yeah, I can't live without the Internet. At this point, it's completely transformed the economy, transformed our lives.

When I travel into the mountains of India, Indonesia, or somewhere, there are people with smartphones on Facebook, following the memes of today. By the way, I would say that memes are actually driving markets more than themes these days, but I'll put that aside. At the core of it, it's still a solid economy and there's still a huge productivity gain coming from this new technology. How much of that productivity gain will ultimately be achieved is actually not about the technology, in my opinion. I think it's about the adoption.

[RC]: Thank you so much for this conversation. I know we covered a lot of ground, and it's given me a lot to think about going back, but I also think it's just a wonderful example of what the life of a portfolio manager looks like. Contemplating a bigger picture theme, really trying to understand all the drivers and the angles, what the other side might be, where you might be wrong, and then going back to the bottom-up and understanding exposures within portfolios, what might benefit, what might not, having some contradictions, and then, like you said at the end, being super humble about one's lock on that degree of knowledge.

Jeff, thanks so much for taking the time to walk through that with us and look forward to having you on again soon.

[JM]: Thanks, Rob.

[RC]: Hi, everyone. Rob here again. To subscribe to the Art of Boring podcast, go to mawer.com, that's M-A-W-E-R.com forward slash podcast, or wherever you download your podcasts. If you enjoyed this episode, please leave a review on iTunes, which will help more people discover the Be Boring, Make Money philosophy. Thanks for listening.


Companies Mentioned
TSMC (Taiwan Semiconductor Manufacturing Company)
Google (Alphabet)
Amazon
Microsoft
Meta (Facebook)
Apple
Tesla (Optimus humanoid robot)
Northrop (defense exposure)
SanDisk (memory/AI data center exposure)
OSI Systems (border control/security scanning equipment)
SAP (software example)
Starlink (SpaceX satellite internet)