Hyperscalers: Hypergrowth, higher risk

0:00Hey, everybody. Welcome back to the Barclays Brief. It's Ronnie. I'm back home in New York after a great nonstop week in Asia. No surprise, through lots of investor meetings, everybody want to talk about AI. But the narrative bent a little bit. And for the first time, concern was a bigger part of the discussion than optimism. Most of that concern is around the viability of hyperscaler capex and broader infrastructure commitments. The good news is I have Andrew Keches in the studio with me. Andrew co-runs U.S. high-grade credit research and focuses on TMT. It's perfect timing for him to be here with us.

0:31Andrew, welcome to the Barclays Brief. Thanks for having me. And Ronnie, it's funny you mentioned that in the intro because I was coming into work today and sort of wondering to myself if there's ever been a topic that has felt as all-encompassing and consuming as AI has been. And of course, I get in front of my computer and see all these announcements. A steel company is now producing data center steel and a paint company making data center paint. So I think everyone is chasing this to some degree and has a lot of incentive to attach themselves to it. It definitely feels like it's all AI all the time.

1:01But let's start with hyperscaler CapEx and that CapEx growth. The numbers being thrown around are simply jaw-dropping. Consensus estimates for hyperscaler CapEx are almost 400 billion this year and expected to ramp to north of 600 billion in 2027. Can you just take us through all that, please? Yeah, jaw-dropping, mind-boggling. There's a lot of different ways we could probably describe the CapEx numbers. And if I were to summarize how the market feels right now. It's somewhere between excitement and nervousness, as every time someone talks about this. And the reason really is because I think the market still feels like AI is absolutely one of those once-in-a-generation technology shifts that's going to change a lot and have a massive impact on the economy.

1:43The question, though, and the part around nervousness is just how much is this going to cost? And the reality is every time the hyperscalers report, forward capex estimates move higher. It just continues to happen. So we've gone through a couple of different phases in terms of, you know, euphoric. I sort of, the analog I would give is back to the automakers with EVs, where every time they upped their spending targets and talked about EVs, their equity would react positively and everyone felt really good about it. And, you know, now we're several years into it. And some of those, some of those investments may not have paid out the way that I think the automakers thought.

2:17I don't think that's necessarily how people are thinking about AI. I think people genuinely truly believe that this is going to have a massive impact, but no one knows how much this is going to cost. And these numbers just keep moving higher. So I think that's really the nervousness part of it all. Well, we definitely know the numbers are going to be huge. How will all this spending be funded? So far, it's been mostly a cashflow story. So I think the really important thing to put in perspective here is these companies are some of the biggest in the world. They had massive technology and internet platforms prior to this, And they have this huge backbone of free cash flow that they've been able to use to really subsidize almost all their spending to date.

2:53I think we're starting to see that shift. And so we're starting to see a little bit more of a move into the debt financing markets. And I think that's what's really caught everyone's attention, at least very recently. In my conversations, that is for sure the thing that has everybody's alarm bells up. Can we spend a little bit more time on that? We're moving from internal free cash flow funded buildouts to external debt funded buildouts. How does this all work? So I think it starts with the cash flow. And if you were to go look at forward estimates, you'd say that the hyperscalers are all still going to generate a handsome amount of cash over the next many years.

3:24And it's probably growing. The nuance to that, though, is that most of them pay a very large dividend and a lot of them like to buy their stock back. So if you think about it after shareholder returns, the actual like the true net cash flow they're generating, especially as CapEx continues to move higher, is actually thinning out. Now, some of them are still in a very comfortable position, but I think that's why we're starting to see some of these hyperscalers come to market to issue debt to fund it rather than just rely on cash flow. And the other aspect to keep in mind here, again, huge companies, right?

3:54Trillion plus dollar enterprise values. But almost all of that is in their market cap. They don't really have any meaningful net debt. In fact, some of them have negative net debt. So as a credit guy, I would sit here and say that I think it's responsible to use a little bit of leverage. The issue is, though, that a little bit of leverage can mean some pretty big dollar numbers when we're talking about these companies. Right. And so we've seen a lot of that manifest in a fair amount of investment grade issuance in the credit markets. Can you tell us about how the market's absorbing and digesting that?

4:21The short answer is that I think it actually overwhelmed the investment grade market in recent weeks. So we had coincident timing of a number of deals, both public and private, that came in a short period of time. And it was just a lot of supply to digest. So it had a real impact. The investment grade market saw spreads move wider on this as everyone sort of had to make space for this issuance. And the hyperscalers themselves were issuing it at, I'd say, pretty meaningfully wider levels. And they had to, in other words, they had to give some pretty meaningful concessions to get these bond deals placed.

4:49That all said, these are still very, very high quality companies for the most part. We're talking many of them are AA or even have a AAA rating out there. So as the market digests it, I do expect, and we're already starting to see, the market is going to be able to absorb it and the credit market's starting to recover. All right. So let's bring all these risks into focus. What could go wrong? Well, I think one part of it is it's not all just going to be funded in the public debt markets. Private credit is absolutely going to play a role. And so I think we're already starting to see that with a few deals.

5:18And so as you start to lay out the funding and it starts on balance sheet, then starts to move off balance sheet, we'll have to see if there's SPVs, different entities out there that are funding it. you start to have a more dispersed set of risks here as you have to track all this off-balance sheet obligations of the company and factor that into the credit quality of them. So I do think, if you were to summarize that, we're talking about big numbers, they're getting bigger. We're talking about funding moving off balance sheet. And I do think the combination of those is somewhat reminiscent to past market cycles where there's been drawdowns or big equity volatility.

5:51So I think there's a lot of discomfort around it right now. It's an interesting dichotomy because most equity analysts in the tech space are very optimistic about the prospects of AI. What are most credit clients you speak to thinking about and worrying about here? I bet your world is pretty optimistic about it because we're using my world now to fund it going forward. Yeah, we need your money. I would not say that all credit investors are bearish on this by any means. In fact, there's still plenty of bulls out there who say, look, this is going to be a challenging couple years of transition and credit metrics can get a little wonky.

6:24But as long as we believe in the future state, this is an opportunity and we should be underwriting to these. But I think the where credit and equity differs is really just comes down to what we just said, where the credit market is going to play a bigger and bigger role in terms of funding this all. And again, we don't know what the ultimate end state of AI is, and we don't even know how much it will cost and how long it will take to get to that cash flow inflection. So I just think from our side, the risks are a little bit asymmetric. And so it's not surprising to the least to hear that your world is excited and mine's a little bit nervous about this.

6:55Now, I have a question for you. I've heard you speak about AI pretty bullishly in the past. So talk me through where you're at. Are you still a bull? Sure. So look, I don't think it's a right now thing, but I do think that the narrative is bending. And I think that personally, my risk antenna is up in a different way than it was certainly a few weeks ago. And I'll tell you why. This ROI of AI theme is going to continue to stay in focus. And I think that it's a little bit unclear as to how it's all going to play out. And at some point, and maybe at various points along this journey, the equity market's going to get uncomfortable with that.

7:26And then it strikes me that we're moving from capital-like business models in big tech to more capital-heavy business models in big tech, even if it's just for a short period of time. But equity markets haven't historically rewarded companies going through these shifts. And so that's something to bear in mind. And then all of this circular financing just has me nervous. And so I think at some point, this story is going to get challenged. And a lot of people are in it. And people are very exposed. So I think it's probably a time to just be respectful of the risks. And I think a point you just made, I would even take it a step further about going from capital light to capital intensive.

8:01Some of these businesses were software internet companies. And I would describe them as becoming infrastructure providers. And this isn't necessarily one-time capital that needs to be invested. There's a refresh cycle. And so you're on a little bit of a CapEx treadmill. It's going to be higher for longer. It may not always be this high, but you're absolutely going to have more durable CapEx going forward. And there will definitely be moments of discomfort associated with that dynamic. Thanks, Andrew. Thank you. This was great. We definitely covered a lot of ground, and I know that we'll be talking a lot more going forward.

8:29Here are my key takeaways from this session. AI is a powerful technology with vast potential. Reaching that potential will require enormous sums of investment dollars. And those investment dollars will only continue to flow if the right return hurdles are achieved. So we're going to have to keep a very close eye on this interplay. Clients can read more on this topic on the Barclays Live link in the show notes. And please don't forget to subscribe wherever you're listening to get notified every time we put out a new episode.