Yahoo Finance

Chip stocks, well, they're mixed this morning. AMD, Intel and Qualcomm pulling back just a bit, but this comes after yesterday's rally. You can see Nvidia up another nearly a half of a percent today, gained 8% yesterday. Some of that recent increased optimism surrounding the sector coming as CEOs in the industry tout strong demand.

The demand on it is so great. Uh, and everybody wants to be first. Yeah. And everybody wants to be most. And everybody wants to be And so the the intensity is really, really quite extraordinary.

I believe this is a market that will grow to, you know, $400 billion by 2027, which is huge, and it is an AI super cycle.

So let's talk about that outlook for chip demand here going forward. For that, we want to bring in John Vin. He's Keybank Capital Markets Equity Research analyst, John. It's great to talk to you again. Let's start with Nvidia, given the massive move that we saw in the market yesterday, closing out with an 8% gain, lots of optimism surrounding some of the commentary that we got from Jensen Wong. Anything in there surprise you, or is it pretty much much of the same and just reassuring some of those skeptics out there on the street?

Yeah, I think this bounce that you saw on Nvidia yesterday was, you know, probably, you know, recovery based on kind of an overreaction on earnings, right? Obviously, Nvidia just, you know, recently reported earnings that kind of meet met and exceeded street expectations, but they kind of missed some of the higher buy-side expectations. So you saw a little bit of a pullback there. I think what really drove the recovery you're seeing more recently is that, you know, Oracle came out, talked about very strong, uh, accelerating growth on the, uh, AI side, and they also said that their outlook for next year is going to result in a doubling of CapEx. So I think that's reassuring investors that the AI trade is not over yet.

Is Oracle in that sense the best non-chip AI play in the market right now?

Um, I don't cover Oracle. I cover the semiconductor. So I would just say that, you know, Oracle right now is one of the key important reads in terms of AI hyperscalar demand and spend going forward. You know, obviously, they're off cycle, so they're they're giving us the most recent read on AI chip spend at this point.

I knew you'd had an extremely informed reading. I know I know you don't cover them, John. I just got to, uh, when we're on the topic, I got to lob it into you and see see what we come out with. But great insights as always. As we think about, though, for chips, who is best positioned to really kind of catch some of the overflow that Nvidia is not able to handle for themselves?

Um, I would say kind of three names that we would throw out, you know, right now. We think it's such a, uh, expensive market. We heard Lisa Su talk about it being a 400 billion market by 2027. The three names I would kind of highlight would be an AMD, a Marvel, and a Broadcom, right? So AMD, they're going to be, you know, we think they're going to be the lead, uh, second-source alternative on the merchant side. Uh, Marvel and Broadcom are going to be focused more on AI hyperscalar custom ASIC, uh, development, you know, obviously Broadcom owns the Google TPU franchise there. And, you know, they've got a pretty strong position, whereas Marvel, you're starting to see them get some good traction in the market with with AWS.

John, when it comes to some of the maybe lesser known names, you and serious logic, I believe is one of the names that you like when you talk about focusing on Gen AI and the further destocking process. Where do you see the value there?

Yeah, we like serious logic. Um, they're kind of a proxy for Apple. You know, 80% of their revenues is is Apple driven. And if you look at the opportunity ahead for serious logic, right? There's potentially a replacement cycle, um, if Apple starts to to really get traction with, um, with Apple intelligence there. So you could start to see unit growth. And on top of that, this is a company that has demonstrated, uh, content growth on pretty much every single new generation of iPhones. So on top of unit growth, you also have content growth that compounds. And it's it's kind of a proxy and a play on on Apple.

So with that in mind, I mean, as we're waiting for some of, especially coming from Apple, waiting for some of these generative AI features to really hit critical mass coming through either via operating system updates or other consumer technology, where do we still need to see a lot of the industry be built out? Where are some of the gaps that still exist where investors would be apt to pay closest attention to?

You know, obviously, uh, everybody saw the Apple launch event the other day. And, you know, I think they made some progress, but I think if you take a step back, you know, I think the industry is still waiting for that kind of killer kind of AI app. And I don't know if we really saw it the other day. So I think, um, you know, you're definitely seeing some progress, but I think the the broader market, um, is still waiting to see what that killer application is going to be. That's going to drive potentially a replacement cycle in in AI assets.

What does that mean for the future of the data center? We've heard Nvidia talk about this, the possibility for 1 million chip data centers in the future. Oracle's talked about needing a data center to be powered by nuclear reactors. So the data center itself seems to be set for a massive overhaul. How much more spending, though, do you think needs to come on the front end of that from some of the very companies that are looking to play in this AI space?

I think in the near term, uh, we think spending is likely going to be unabated. You know, obviously, the debate and controversy is, you know, when are you going to start to see the monetization models for for all this AI spend? And you really haven't seen it yet. Um, you are obviously seeing some pretty good growth, but what we look at in our perspective, at least in the near to medium term, at least going into 2025, is that all these AI hyperscalars are going to have to aggressively spend the next generation platform that Nvidia is releasing. Blackwell is going to basically lower your TCO by by 30x. So if we're still waiting for that monetization model, um, operationally, if you're a hyperscalar, you've got to go with the lowest cost TCO solution out there until you get that monetization model. So we don't see any sort of letup in spend over the next 12 months here.

John Vin, who is the Keybank Capital Markets Equity Research analyst, John, always a pleasure to get some of your insights. Thanks so much for joining us.

Thank you.