Why AI Makes Memory Demand Less Cyclical

Key Takeaways

The situation with the memory makers is fascinating. Um demand is exceeding supply. Historically, that's an enviable position to be in. And the shares of those companies probably reflect that. How How do you see investor appetite to to be exposed to the memory trade? >> Yeah, I mean the shares have been fantastic, but even after that, I think memory stocks still offer a very compelling investment case. And here's why. I don't think it's necessarily just driven by memory supply bottlenecks. Instead, I think there's something structural going on in the cycle. For example, memory's strategic value is rising and changing to a primary driver for performance. Think about how cloud service providers, one of their largest clients, they're still adding more memory despite the price inflation because this is still the most cost-efficient way to maximize their system-level performance. And when cost of the When token demand, excuse me, the token demand is rising on the back of the inference as well as agentic AI, I think the bottlenecks in terms of the capacity and bandwidth are becoming even more challenging. And this is a place where memory can come in and help navigate those challenges. >> Does it matter that investors will have the opportunity to have US-listed shares, ADRs of SK? What will How does that kind of change the mechanics of the market for you as well? >> Yeah, I mean it clearly the more exposure to more diverse type of shareholder base is clearly a very positive indication for any company in the world, especially when your fundamentals when when your fundamentals are improving. So, yeah, I'm super exciting news and looking forward to the development. >> A couple of months ago, I was sitting down with Jensen Huang, and I said to him, "Do we even need the textbooks anymore that would tell us historically memory is cyclical? It is boom and bust." All of the evidence suggests that in the context of HBM going into data center, it just doesn't behave the same way. What is your thesis on that? >> I mean, I agree with that. I mean, like I said, there's something structural going on. Yes, there is certain type of cyclicality cyclicality going on in the market, but at the end of the at the end of the day, what you need to remember is structural side of both demand and supply. But think about the supply side. Despite the fact that all the major companies are scrambling to add new supply, there are three factors that you have to remember. Number one, manufacturing intensity manufacturing complexity is rising. Number two, capital intensity to open a new fab is getting more expensive. And number three, the difficulty of technology migration. All together are pointing out that overall supply demand balance will stay in favor of the memory stocks. I mean, I think that's really great. >> I'd love to talk about your active managed ETF for a moment, TEKY Tecky. >> Tecky. >> Uh launched last April, right? 60 million in assets currently. What I find so interesting about the composition of it is something we've talked about quite recently on the show. There are the capital expenditure deployers and there are the capital expenditure recipients. >> Right. >> In the same basket when you look at the top holdings. >> Right. >> Is that a conscious active decision on composition? >> I think it's the composition from our bottom-up stock selection and portfolio construction, but like I said, that's exactly where majority of the capital is invested in currently. And it on one hand, there are big spenders big capex spenders that are trying to build new competitive positioning in this AI arms race. But on the other hand, we find a lot of exciting opportunities on the companies that receive this capitals. AI hardware supply chain, for example, I think that's why there's more to come. In fact, we just recently came from Asia where we sat down with a number of different companies in the supply chain. Everyone is telling us how they're seeing an extended order visibility from customers. How there are more conversations about LTAs. Everything collectively is highlighting that demand continues to substantially exceed the supply and that AI demand outlook remains pretty robust as well. >> For in the ETF context, it's a very competitive market. Um how closely are you thinking about flows? And again, I think people give you a lot of credit Selene for identifying SK as being just critical to the broader infrastructure build-out right now. But, you know, what happens next to your mind? How do you see the world changing uh in the next, I guess, 6 months to 12 months? >> I mean, even more than that. I mean, that's why we focus on AI big stack. And at the top of the stack, we have application layer. And this is, for example, where we are going to expect really large opportunities from physical AI. We think physical AI is going to be multi-trillion-dollar long-term opportunity. Essentially, AI is important because it's a primary driver for productivity gains. And historically, productivity growth tend to translate to massive economic expansions. I think the same thing is going to happen for AI. All the innovation we're seeing today is eventually going to open up very significant new end markets in physical AI, such as fully autonomous trans transportations, >> Right. >> as well as um humanoid robots. Um that's why we're really excited about physical AI. And in the meantime, that being said, it will still take a couple of years for this to turn into actual corporate earnings. So, here's what we focus on instead. Companies with um vertically integrated manufacturing excellence that can switch to the market with scale advantage, for example. Or open-source platform that can expedite the acceleration, adoption, and everything. And lastly, global technology companies that can massively benefit from the mass adoption of the application itself. Something to watch on and something we're still pretty excited about.