00:00 Speaker A
As investors digest Meta’s expanding partnerships with chip companies, it puts into focus the increasing CAPEX spend that Big Tech is investing into their AI infrastructure. Wall Street eagerly awaits to see the winners and losers of this trade. and so has my next guest, Thomas Carroll, Stefel Equity strategist. He says the AI trade is splitting off into hardware winners and service losers. Thomas, it is good to see you. So, let’s just start there.
00:26 Speaker A
Uh, you say we’re seeing this split now between as you put it, AI hardware winners and AI service losers. Explain that for us, Thomas. walk us through it.
00:35 Thomas Carroll
Yes. Thank you for having me on. Um, yes, the the crux of this note was, if we think about the market over the past three years, it’s really been about, you know, going long the entire AI trade, right? But what’s been interesting to us over the past, you know, really last quarter, as been as you said, the splintering of AI services, which are the perceived losers, which are you and your previous guests were just talking about,
01:03 Thomas Carroll
and the perceived winners, uh, the AI hardware names, which are, you know, have a perceived safety attached to them, and the market has really moved into them. But the idea of this note was that we really, you know, questioned the overall safety, if we were to believe the market’s, uh, opinion on the overall losers of the AI services trade.
01:25 Thomas Carroll
And so if I were to take you into like what we’re really seeing in the AI Services space, uh, you know, as you talked about your previous guest, there is significant pressure on these business models as we move forward and it’s very easy to see across the credit curves. uh we’re seeing, you know, you know, a significant move higher in credit risk premiums, whether it be for very safe names, even to the high yield space and the tech space as well.
01:49 Thomas Carroll
And that’s what’s really in our opinion that’s driving down a lot of these AI services names, uh, fears on the extended amounts of CAPEX they’re doing with the uncertain ROICs on the back half of this CAPEX. The question we have is if we were to believe the market that it is existential for these AI services companies, well, if you look at the, you know, the primary revenues of the AI hardware names, it is the capital expenditure from the AI services names. So,
02:14 Thomas Carroll
there’s an inter linkage between the two and the splintering of this trade uh has us somewhat cautious, which fits with our overall S&P 500 view coming into this year.
02:25 Speaker A
Thomas, you actually compare here this moment to the late 1990s, which is interesting when hardware and services diverge sharply and and then we saw volatility explode there. What are the the sort of the most important similarities between now and then, do you think?
02:41 Thomas Carroll
Yes, I would say uh that that particular study got a little bit of the goosebumps up on the back of my neck uh when we first observed it. Uh if we think back into, you know, the late 90s, right, we just remember the tech bubble, the boom, the bust, but it actually was obviously a very long period. And the Nasdaq actually peaked at the beginning of 1999, but the market held on for roughly nine months.
03:02 Thomas Carroll
And so what we observed then, which is very similar to what we’re observing now, except the GPT of the time then was the internet, which obviously, you know, we are talking through the internet. It has changed all of our lives. You do not doubt the importance of AI, but it was very interesting that hardware took off in a not in an exponential uh pattern relative to internet services at that time. And when that trade did peak, it went up 95% in a nine-month period and when it did peak, it was kind of the top
03:28 Thomas Carroll
for the market at its time. And so, we are observing, you know, a GPT similar to, you know, maybe potentially even bigger than the internet. We’re seeing a very similar splintering of, you know, services relative to hardware.
03:41 Thomas Carroll
Uh the similarities kind of have the goosebumps back up on on our necks a little bit as we observe it, especially given how thin the margins are for the overall S&P 500 entering into this year.
03:51 Speaker A
I I did think it was interesting, Thomas, you note here how AI services, you say trading at relative valuations we haven’t seen since you say the global financial crisis. I mean, I guess as as an investor then you’re trying to decide, okay, at that level is that an opportunity or or it’s it’s no, it’s a value trap. How how do you make that call?
04:13 Thomas Carroll
Well, I think the question is why is that occurred, right? Because we’ve continued to see forward earnings go up for these companies, but yet we’re talking about a 10-year rating of the price earnings multiple. And that’s where we what I brought up the overall credit risk that we’re seeing rising in the system. It’s really getting attached to these names. And the question that we have is, okay, if the CAP X is beginning to become existential for these companies business models
04:36 Thomas Carroll
and it’s dragging down this valuation, you know, it could be a value trap if that is correct. If you know, the credit stress begins to broaden. I think the bigger question that we’re having is that do we begin to see that broaden out to the general economy because the overall AI space has been so important to the US economy through the wealth effect of the K-shaped consumer, but also to the industrial production side as well.
04:57 Speaker A
You mentioned this earlier, Thomas, I just want to come back to this rising credit stress in in parts of the tech sector. What are you seeing exactly, Thomas? Um what signs, what signals? and how concerned do you think investors should be about that?
05:12 Thomas Carroll
Yeah, uh for us, it, you know, if we when we look at the overall AI services space, you know, it could include the biggest hyperscalers, right? They’ve also they have seen their credit risks rise, you know, the Oracle chart that has gone around has been very, you know, seen by a lot of investors, but it’s really true for most of the hyperscalers at different frames, but also out to the, you know, the broader credit risk, the high yield space. We’ve seen a near doubling since the back half of last year
05:32 Thomas Carroll
in the overall, you know, what we would call, you know, the high yield credit spread of markets then. Uh that’s been a significant driver for the valuation compression that we’ve seen in the AI services space. Uh and that’s kind of the the that is something that we tend to see during, you know, the initial stages of stress that’s being built in the system in regards to really any sector of the economy but specifically here as a due to tech.
06:03 Speaker A
Thomas, great to have you on the show today. Thank you.
06:05 Thomas Carroll
Thank you for having me here. Appreciate it.
