Reading the Tea Leaves (Ep. 1) – The AI Arms Race

Chris Zand: Hello, and welcome to our new series called "Reading the Tea Leaves." I'm Chris Zand, and today I'll be joined by portfolio managers, Nael Fakhry and Greg Hermanski. Over the years we've been asked to provide in-depth analysis on topics and areas of the market that we find to be particularly of interest or carrying significant future implications.

So we decided to start a new video series, that we will aim to update quarterly, that will explore these topics that are both top-of-mind for our investment teams and our clients. Coincidentally, we're all regular tea drinkers, and since we're going to be exploring current market trends and assessing their future implications, we've decided to call this series "Reading the Tea Leaves."

In this episode, we're going to be focused on an under-the-radar shift that's going on in the technology space, which we think hasn't received all that much media attention. The media, understandably, and news outlets, have been focused more recently on inflation, on interest rates, and this very important upcoming election we will have in November.

The predominant narrative, which we tend to agree with, is that the economy is continuing to hold up well, despite persistent inflation and higher borrowing costs thanks to a strong and robust employment market, and very attractive and strong wages.

Wall Street seems to agree, as the stock market has been rallying for the last 18 months, led by the so-called Magnificent 7 and other Mega Cap technology companies.

Today, we're going to take a closer look at an under-the-radar shift that's happening with many of these companies, specifically their massive investment in artificial intelligence and the infrastructure to support artificial intelligence-related activity. The scale is enormous, the stakes are extremely high, and the potential implications for the future of the stock market, and some of these stocks in particular that are major players today, is significant. To kick things off, I'll hand things over to Nael Fakhry and ask him to explain exactly what's going on in this space. Nael, over to you.

Nael Fakhry: Happy to do it. So as you know, we focus on Quality Growth companies, and that's the type of company we want in the portfolio. And as a key part of that, we really focus on obviously earnings and the conversion of that earnings into cash flow, and we focus on incremental returns on capital. And I think that is really important as a framework to think about what we're talking about here in incremental capital spend.

Last quarter's updates from the large cloud providers represented what we think was really a seminal shift in terms of their budgets, specifically the biggest cloud providers, Alphabet, Amazon, Microsoft, and Meta dramatically increased their capital spend and it was really related to the cloud parts of their business. It's driven both by underlying IT demand, but also by increased demand for AI, and they're all at the same time, very substantially increasing their investment in hardware and in the infrastructure required to support AI on a massive scale

Chris Zand: Nael, you mentioned dramatic increase. Maybe you could expand a little bit on how significant we're talking about here.

Nael Fakhry: Sure. So the numbers are really jaw-dropping. As you can see from this chart that we put together, collectively, the investment in cloud spending has gone from what was very large number last year, just under $120 billion by these four companies again, Amazon, Alphabet, Meta, and Microsoft to what we're projecting this calendar year to be about $180 billion. So that's a massive increase, about 50% on a year-over-year basis in just one year. And these companies, keep in mind, these companies have raised CapEx 50% on a year-over-year basis in the past. So there's precedent for this, but when you're talking about this scale of dollar spend as a baseline and growing 50% off of that, the incremental dollar spend is just jaw-dropping again.

And the other thing to think about is just in context of revenue, Google and Microsoft or Alphabet and Microsoft are collectively each spending about 15 to 20% of revenue on capital expenditures related to the cloud. And that compares to other capital-intensive industries like auto manufacturing or energy spending about five to 10% of the revenue on capital expenditures. So that really provides, I think, a framework for how significant the spend is.

All that to say that this spend really requires very serious scrutiny and analysis to ensure that the incremental returns are attractive and that the cash flows that come out of this spend ultimately are additive to the businesses. And that's why we think about things from a quality growth perspective. It's really important to think about the incremental returns in the cash flows that come out of these budgets.

Chris Zand: Wow, these are quite staggering, Nael. Thank you. Greg, can you talk a little bit about why this is happening now and try to put it in historical context. My sense is this isn't very typical for these companies year-to-year.

Greg Hermanski: Sure, thanks Chris. We think that there's two elements that are influencing this huge increase in CapEx. And the first part is just to acknowledge that to support AI, the infrastructure demands are massive. So if you're training a hundred million to a trillion parameter LLM model, it requires an immense amount of very expensive leading edge compute power.

The second element of why this is driving such a huge increase in capital spending, I think really goes back to the end of 2022. That's when OpenAI's large language models came mainstream to the public, and it was more visible how you could use LLMs to make it easier for people and companies to access the revolutionary power of artificial intelligence as a technology. So once people realized that they could access AI through the large language models, the floodgates seemed to open up and people were rushing to find new ways to monetize that.

Part of it is Silicon Valley and the venture capitalists, they wanted to start a new company so you've seen a large number of companies that have popped up that are artificial intelligence companies, and that was one angle of the spending. But the other piece was that a lot of companies that already have businesses in the market, they're trying to protect their businesses using AI. And so that's driven the second element.

Normally when you have new technologies, maybe it enhances one segment of the business, but now we have this revolutionary technology that is very expensive and people are trying to create new businesses and they're trying to protect the old businesses, and it's created this massive capital spending.

The second part of your question was to take it from a historical context, and there's no doubt that this is a huge amount of spending, especially from a dollar perspective. But if you go back and you look, and we added up Amazon, Google, and Microsoft from the period of 2012 to 2014 and looked at how the growth was when we had a new type of revolutionary technology and a move to cloud computing from on-premise computing. And over that three-year period, CapEx increased about 150%. So there actually is a historical precedent for large increases in technology spending when you have these more revolutionary changes.

Chris Zand: Very fascinating. Thanks, Greg. Nael, let me turn back to you. What are the implications for investors? Is this good news or bad news if you own these companies?

Nael Fakhry: Well, we think it depends. And simplistically, if the returns on the incremental spend are attractive like we've been talking about, then it's going to further cement the dominance of these companies as cloud providers. You think about the cloud business as it exists today. It's a really attractive business. There are long-term contracted revenues, high margins, and there's a very long runway of growth. So these cloud businesses are massive recurring profit engines for these businesses. And if AI fortifies these positions for just a handful of winners, that's incrementally very positive.

Now, there's obviously uncertainty on this incremental spend that we've been talking about, but we would say specifically for Amazon, Alphabet, and Microsoft, they have a really long and positive history of capital allocation and really effective capital allocation. And so obviously these cloud businesses that we really like, they built these businesses largely from scratch, and these are great businesses. And so that's one important consideration.

The other consideration is that even outside of the cloud, these have been really great capital allocators. They've built amazing businesses, and as Greg alluded to earlier, these companies have existing products that have incorporated AI for years now. This is not a new technology in many ways for these companies, so they kind of have an edge. If they can build AI onto their existing product lines, that gives them, we think a potentially really big edge. So we definitely give them the benefit of doubt to some extent, but we're watching carefully given this massive dollar spend we've been discussing.

Chris Zand: Thanks, Nael. The implications truly are fascinating. I'd like to wrap things up with this final question, what are the key questions that you'll be focused on going forward?

Greg Hermanski: We think that there's three key questions to focus on. One is what sort of incremental returns will these companies generate on the incremental cloud spend? We've kind of discussed that already, but that's something that we'll be monitoring. Also, will the current leaders be the new leaders as AI's implemented? And typically when you see technology switch from one revolution to the next revolution, you end up with new leaders so we'll be watching to make sure that our companies are able to keep their competitive moats and in addition, look for maybe some new opportunities with new leaders in the market.

And then the third element would be, is capital spending increasing to a new base level that people will grow off of from there? Or is this kind of a one-time spend to build a primary AI infrastructure and then that'll reduce to certain lower level and then grow off that lower level. So we're trying to figure out where capital spending is going to go long-term and what the growth off of that level is. That's a little bit unclear at this point.

Chris Zand: Very interesting. Thank you both very much. I'll look forward to speaking with you guys on this topic as it continues to evolve. I'd like to thank our audience for joining us today. Until the next Reading the Tea Leaves, feel free to reach out with any questions.