AI stock corrections are currently driven by shifting market sentiment rather than weak earnings fundamentals, says Roundhill. While investors worry about long-term durability, earnings momentum remains strong. Roundhill noted, "Right now, the bears are winning because there's this void." Many tech firms still face pressure as people sell first.
Question
We've seen these corrections in the Kospi and in these individual names before, but the question is: Is this time different? Because it comes on the back of debt mounting at some of the companies, not all, and also an unprecedented increase in the kind of interest rates we've seen in the US. Is this time different?
Answer:
Right now, a couple of things are different. Number one, I think the narrative's different. So, when we think about a lot of these tech names, especially around memory and the infrastructure build-out, people were talking about short-term pricing power. That was a reason to get into these trades. The reason to stay in these trades is the longer-term durability of earnings. So, people are now questioning that, and it's harder to prove out in real time that this isn't a commodity-like industry anymore. Instead, it is more driven by long-term supply contracts and so on. So, the narrative's changing, and when the narrative is changing, people sell first and ask questions later. So that's what's changing. What's not changing is the earnings revisions in the near term. Earnings momentum is still really strong here. We have a lot of exposure in the memory space at Roundhill, and this is a place where we feel a little bit uneasy in the short term, but we do think the longer-term earnings trends and the supply bottleneck will continue to exist.
Question
In that sense, nothing has fundamentally changed.
Answer:
No, and if something has fundamentally changed, we won't know for some time. So that's the problem. Right now, the bears are winning because there's this void. I can't prove that this industry is going to look better in 10 years. I'm not really sure that commentary in the upcoming reporting period is going to help. We might need a couple of quarters where this becomes a little bit more of a "show-me" story, unless you start seeing acceleration in other places. Now, there are some other areas. I heard you mention CoreWeave before, and some of the neoclouds. Those have actual debt-related problems coming up right now. So that's more near-term pressure there. But I would say that Micron, Samsung, and Hynix don't have the same debt pressure on their own balance sheets, even though some of their customers might.
Question
This CoreWeave—is it a big issue? I mean, have you looked at this in any detail, or is this like a one-off? What's your sense?
Answer:
So, it's a little bit of a messier story because, unlike the memory names that I mentioned, where revisions keep going higher and higher and higher every time we talk, CoreWeave has seen revisions move slightly down for 2026, 2027, and 2028. So, it doesn't have the short-term tailwind, and then you throw in the headwind of debt continuing to become more expensive. If you start worrying about whether these guys are going to have the capital to continue the build-out, you can understand why there's more pressure there in the near term. So, I get it. Anything that's run up a lot, and anything that's had any sort of momentum in the past year, has underperformed the most in this sell-off. So, it does feel positioning-driven in that sense. But again, the longer-term fundamental earnings story is still intact. This is headline fear and repositioning for now. Fundamentals are to be determined.