The cusp of a capex supercycle

0:00Welcome back to the Barclays Brief. It's a bright morning here in London, but I've been thinking about how foggy the short-term outlook for markets feels right now. Headlines are shifting fast and it's easy to focus on one part of the picture at a time and then lose your perspective. Because beneath the more negative headlines we're seeing, a largely acyclical investment cycle driven by AI, but also energy and defence is gathering momentum. And the scale of capital being deployed here is enormous. Taken together, this points to something way bigger, a step change in global capital spending. So the question is, what does that bigger picture tell us?

0:43I'm Patrick Coffey and sitting opposite me today is Christian Keller, our chief economist in our research business. He's just published a chapter called Super Size Me as part of our global outlook. Christian, thanks for being here. Great to have you on. Thanks, Patrick. Thanks for having me. Good to be here. It was a tough one to get out as a global outlook in volatile times like this. I'm sure. How many rewrites did the front cover take? You know, as geopolitics moved on, we had to adjust it a few times. So, Christian, I was rereading that chapter of your Super Size Me on the train into work today.

1:20And in it, you argue that we're at the early stage of a major investment cycle in advanced economies. That kind of feels a bit optimistic to me. So what gives you the conviction that we're about to enter a new phase of much higher infrastructure investment? That conviction in part is driven by the fact that when you look at investment cycles, you really have to think of longer time spans. We looked at 150 years of UK and US data. And what you see is that while investment is very volatile. It does come in cycles, typically 10 to something a year cycles that you see an acceleration of investment.

2:00And typically that is associated with somehow new technologies where companies and economies need to invest in something. It is certainly punctuated by wars, financial crisis, and these types of disruptions. But ultimately what dominates is that kind of new technology and investment that comes with it and the reorganization of an economy. around that new investment. Yeah, exactly. And so by new investment, you're talking about AI, presumably. Exactly. In this case, I'm talking about AI and also all the infrastructure that comes with it. So the energy needs, but also the investment, IT equipment, et cetera, as AI becomes a proper general purpose technology diffuser throughout the economy.

2:42And then a bit separate, there's the defense cycle. I don't know, but we'll talk about it separately. Well, we'll definitely talk about that, I'm sure. Okay, so there's lots to cover here. Let's start with AI. Help me put the hyperscaler investment cycle into some context. I find as I get older, I'm increasingly drawn to historical analogies. I've been reading a lot about US railroad investment in the 1850s, the Apollo space program in the 60s, and the telecoms boom of the 90s. So where does the AI capex cycle really sit in comparison to those capex booms? Yes, I think historical comparison is actually a very useful tool because it offers perspective.

3:20As I said earlier, we have about 150 years of data. And the interesting thing is, if we look at the U.S. hyperscale investor this year, you get about to the average annual investment that was done in the U.S. during that railway boom you just mentioned. So we are in that category. And that is one of the biggest cycles. It's way bigger than the Apollo space program, a highway program in the U.S. in the 50s to 70s. And it's also actually bigger than the telecoms already in some ways. Has anything else in history been bigger than this AI CapEx boom? Well, we have some interesting artifact. We found that there was a purchase by the US of the Louisiana territory.

4:05So there is a bigger investment if you want to count that. But frankly, that was very particular. We threw it in there just as a comparison. But as I said, AI is now up there with the railroads. It should be a pretty big boom. And can you just put some numbers around that investment relative to other cycles for us? Yeah, look, when we look at the build-out of the US railroads, we have that as 2 % of GDP. And we think that's about where the US hyperscalers end up right now when we look at the committed investments. Okay. And that AI investment, is it adding up to a sort of notable increase in aggregate investment for the entire economy?

4:44Or is it more of a reallocation of investment? Yeah. That's a very, very good point. And we don't want to be naive here. For example, if you look at just the fourth quarter of GDP data last year, actually overall investment, despite very strong AI-related investment, wasn't that strong. So there seems to be some replacement going on. However, over time, if you see AI as a technology that needs investment, as I said, not only in the AI itself, the data centers, but then also in other equipment, AI related, and in the energy, it is likely to be additive. Not 100%, there'll be some offset, some replacement, but it's hard to see that you have such a boom that would then be fully offset by reduction of investment elsewhere.

5:28Yeah. And clearly, AI is the most visible part of this whole story. But your work suggests it's only one building block. How important interlinked then are energy security, defense spending, and economic resilience in turning this into a kind of broader, more durable CapEx boom? It's important to also to see that we are looking at this kind of as a cycle in advanced economies. And I think the regional distribution is not fully equal. So the AI is particularly done, hyperscaling is particularly done in the US, data centers. When it comes to defense, we see Europe as a region that has not invested much in defense.

6:06That's clearly coming back. But putting into context, well, if you look at general purpose technologies, it's always that you have a core investment. In the 90s, it was computers or so, but then you have a lot of investment around it. We see it now already, what we call the AI-related IT equipment investment. And then you have the energy. And the energy is largely to provide the electricity for the data centers, but there's more going on. There's a general electrification. There'll be a lot of investment needs in grid technology, et cetera, for the electrification that's coming independently of AI.

6:41I think those two are together. Defense is somewhat independent, even though the reality is that a lot of the defense will depend on AI. There's probably some interlinkage there as well. Sure. Just coming back, you mentioned advanced economies. Why do you focus on advanced economies here? Why not just global economies? Good point, because this is really where we saw, from a global perspective, the investment weakness. If you go back, we had ratios of 26 % as a share of GDP, what advanced economies were investing. And that then really came down all the way to last year as a 22 % or so. On a global scale, investment was relatively stable.

7:19What happened? China. China came in early this century, obviously, with joining the WTO. Massively high investment ratios, close to 50, coming down below 40 now, still very, very high. But you had this peak of China investment that seems to be coming off now. And as I said, after two decades almost of low investment, a lot of signals are now for those advanced economies, in particular US, Europe, Japan, now basically coming back up. I see. So you've got to strip out China, focus on the advanced economies, and you're seeing, you think we were kind of at a trough last year and you expect that investment cycle to increase.

7:56You also made the point that geopolitical shocks, clearly we're talking about Iran at the moment, aren't just risks to growth, but they're also catalysts for investment. How should we think about that shift, particularly when it comes to energy and defence? Yes, in the short run, something that is a key ingredient to all activity, or energy in general becoming more expensive is not a good thing. But I think what we've had also together over the last two decades with China is, you know, we had a high focus of all our supply chains globally, manufacturing, energy, everything really on efficiency.

8:28There were no redundancies. Everything was just in time. And obviously with the Corona crisis, I think it came home that people realized, wow, it is dependent on a few factories in China and the same town producing everything we need. And now the ongoing crisis with energy in the Gulf is bringing home to Europe in particular, that they are very reliant still on energy in certain regions. Places like China have much higher storage capacities, et cetera. So what I'm saying here is it probably dawns now to Europe or across the world, to the advanced economies in particular, that they need to go from just in time to just in case.

9:04And that does require investment in hardware. As I said, storage facilities when it comes to oil, larger warehousing or manufacturing goods, etc. And creating some redundancies, all this should be adding to investment needs. I love that expression, just in time to just in case. It's worth remembering though that AI, energy defense, obviously you talked about they're facing a boost in investment. They only account for about a third of all investment in advanced economies. So with that in mind, what are the biggest constraints that could limit how far this investment cycle actually runs? Some are very obvious.

9:40I mean, defense is typically at least led by the public sector, I should say, and the public sector is stretched. And Europe has just, to tell you the story of a country like Germany, that in the mid-60s was spending 5.5 % or so, over 5 % on defense, went all the way down to one and a half, is now back to two, wants to go to five. But Germany does have the fiscal space, but many other European countries don't. So fiscal space is an issue. But I think they will have to find ways, maybe through common European issuance, etc. You're right that interest rates are going up, in part because of the higher investment.

10:17And of course, the offset of not knowing whether investment in other areas maybe go down a bit. That could be offset. Understood. So let's just put it all together. If this investment cycle does play out how you expect, which is essentially advanced economies spending a lot more money, what does that mean for the macro backdrop? You know, for growth, inflation, the way that central banks think about policy over the next few years? If things pan out as we think, and, you know, a cycle we see always as investment growth, right? You would expect that to accelerate. Typically, these cycles are about a decade, you know, and they peak pretty much in the middle.

10:53So what you would expect is in the next five years or so, accelerating investment. What does it do? It means there's more demand for capital. And if savings investment, the savings ratios don't rise immediately, you would think, what is the price for capital? It's interest rates. So that famous R star, that real interest rate, you would think could be higher. At the same time, and that is much debated elsewhere, AI is, of course, potentially a huge pressure on wages. So you could see inflation, core inflation that's wage related to be weaker. Then again, what you will likely see is continued price pressures on commodities because it is commodity intensive.

11:31Energy and electrification needs copper. I mean, all these metals you've probably talked about elsewhere already, they may see continued price pressures. That's why people also talk about a commodity super cycle, which that seems to go hand in hand with that kind of a capex super cycle that we're about to see. Yeah. Christian, thank you so much. I spoke at the start about wanting a bit more perspective at a time where we're being bombarded by headlines. It's great to have you on and talk about this longer term piece because AI, defense, energy, economic resilience, they're so key in the market right now.

12:02And the numbers that you've written in the note are just so big. It's great to have you here. Thanks, Patrick. Great to be here. A couple of things that I'm going to be walking away from this conversation thinking about. Firstly, if you look back at 150 years of data, you can see that investment is volatile, but it does come in cycles. The typical duration of each investment cycle is about a decade, and the cycle shape tends to be symmetric around the peak. After two decades of subdued capital spending in advanced economies, we appear to be at the cusp of an enormous investment cycle, being driven not just by AI, but by energy security, defense, infrastructure and economic resilience.

12:46And it's worth remembering these aren't cyclical decisions. They're strategic ones. They're being reinforced by the very shocks dominating today's headlines. So when you put it all together, by the end of this decade, investment to GDP ratios could reach or even exceed the levels last seen in the 1990s. Thanks for tuning in to the Barclays Brief today. Do hit subscribe to stay up to date with the latest views of our thought leaders. And we'll be back again next week.