Dan Niles: Why Semiconductor Bulls Should Be Watching This

analyst during the dot-com bubble, very famous one in fact. Dan, great to have you with us. >> Great to be on, Melissa. >> Um, you do point out that there is a seasonality impact here, um, at least through the beginning part of July, but what are you anticipating as we go into earning season? Do you expect that, you know, memory could come under pressure, that maybe hyperscalers, uh, could resume some sort of an uptrend? >> So, a couple of things. We're probably going to get a pretty good hint tomorrow because Samsung is going to report tonight. And, you know, they're obviously one of the big three. It's Samsung, Hynix, and Micron. And their profits should be up almost 18x year-over-year when they report. And so, we'll see how the market responds to that tomorrow because the Korean stock market, it's up 90% year-to-date. And that's being driven by Samsung and Hynix, which are roughly half of that index. And so, that's going to give you your first kind of read into this. The second one is SK Hynix goes out with a massive IPO or listing in the US, I should say, not IPO, on Friday. And so, that'll give you another data point of now you don't just have to own Micron, you can actually own Hynix as well. And so, that's going to give you one more. And then along with SpaceX, you know, how does that react to being finally, this is the last add for a while, being added to the Nasdaq 100. You're asking longer term, I just go back to one simple thing. You've gone from token maxing in March to now people trying to control their AI bills. And there's a great post by Brian Armstrong, CEO of Coinbase, where then this was, I think, uh, June 26th or so. And he puts out this post showing how token usage at Coinbase continues to go up. But he's cut his AI spend by almost 50%, 5 0. And a lot of companies are going through this, and that includes Meta Microsoft Uber etc. where they're trying to get their AI bills under control. So, I'm wondering which of the three big hyperscalers ends up guiding below forecast for September after reporting a huge June because of that. So, when they start to report, that's when I'm going to go back to being, you know, pretty cautious on the guidance. >> Hey Dan, Mel just called you a famous internet analyst in the dot-com boom. I was there, you were famous, but um but you know, you made some big calls. You put your neck uh um out there. And you know, when I look at what's going on in memory and I think about in market cap terms and how much these stocks are up and where, let's say, the Nasdaq or Korea's um you know, index is, the Kospi is, I say to myself, well, Nvidia is a greater market cap than probably the top 10 names combined in the memory space. And you look at the sort of price action it is driving a lot of sentiment as far as tech investors are concerned. Does that trouble you that you're seeing a pocket of these names globally, okay, relative to some of the names here in the US that are actually like drive They're really driving the train. Uh and I'm just curious, does that make sense to you? And I'm just curious how you're thinking about that. >> Well, I look at Nvidia and I go, they're under assault from a a bunch of different areas, right? You've got Google, Amazon, Meta. You had an announcement that Anthropic's looking at Samsung to produce their own chip. And so, they're getting, you know, pushed around by all of that news flow. Also, don't forget the big one with the gentic AI, which you could argue showed up in March with the uh sorry, January with the formalization of open claw on January 30th, you move from 8 GPUs to 1 CPU to something closer to 4 to 1 or 1 to 1, which really benefits the legacy microprocessor vendors who were sort of left for dead up until very recently when that ratio switched back. So, I think it's to your point, I think it's very company specific where you got to got to go through the names one by one and go, "Okay, these guys benefit in this regime shift. These guys don't." And the one thing with memory, don't forget, is that this new architecture coming out and in gen tech, Vera Rubin takes it at a minimum like three and a half times more memory than the Blackwell generation did. Now, that's just started to ramp. Obviously, there's talk about that getting pushed back. And so, those are all things I'm watching right now. And this is not a sell-off. I mean, if you go back to 1995 or '97, you had 50% drawdowns in the socks and it finished that whole climb from the end of '94 to March up 850%. You look more recently, I think at the beginning of or middle of 2025, the semiconductor index went down 40%, 4-0 in a drawdown. So, this is minor. That's why I'm very curious to see what's how Samsung and the memory stocks trade tomorrow after Samsung reports tonight and then Hynix on Friday. What does that do to the sector?