The Information's TITV
00:11:07 · Barton Crockett
[00:11:07] S1: Well, look, you know, I think they're two separate things. I mean, I would say in terms of their AI execution, clearly, were dead in the water and now, you know, they're they're treading water, you know, and they're certainly not in the race, you know, their capabilities are not, you know, leading edge, but at least they have something that's credible. And, you know, they did that by, in part by working with Alphabet and in part by changing their team around and, you know, and that's helpful that they're making progress there. You know, I think the real question for Apple is what is the next big thing that really accelerates growth? And this is a stock you love to own when they're coming up with some disruptive new innovation. Obviously, did that with iPhone, they did that to a smaller degree, you know, with the watch and, you know, what could that next thing be? You know, maybe it's robots, you know, not based on the form factors we're seeing in Home View or anything related to that, but, you know, they have skills all around that. Maybe in, you know, coming years, they could come up with something that is exciting, and an Apple robot running around your home could be pretty cool. Maybe they do that in AR glasses, you know, where Meta's been showing some traction, you know, and maybe they they are able to innovate better in AI, and we're certainly seeing, you know, Meta innovate with their muse and, you know, and there is every reason to think that Apple could do that. I'm not sure they've shown the skill to do it, but, you know, if I was to bet on their disruptive innovation, it'd be more on the hardware side than on the AI kind of software front, but we're looking for something, and, you know, and one of those things maybe is it.
00:12:48 · Akash Pasricha, Tore Svanberg
[00:12:48] S0: Right. Well, Barton, I want to thank you for coming on. That is Barton Crockett from Rosenblatt Securities here on TITV. Marvell agreed to buy Lumentum, a market leader in lasers and optics technology used in data centers. Our read here at the information, the deal comes to around $80,000,000,000 Marvell shareholders will own two thirds of the new company. Lumentum shareholders will own one third. That is assuming that the deals go through after what is likely to be a careful antitrust review. Here to offer some analysis is Tore Svanberg, Managing Director and Senior Semiconductor Analyst at Stifel. Tore, welcome to the show. It's great to have you here. [00:13:27] S2: Yes. Thank you for having me on. Appreciate it. [00:13:30] S0: Okay. So we'll get to the price here in a second because the valuation is really interesting. Just remind us though, Marvell and Lumentum, what type of chips, what type of lasers exactly do they do? How does their technology intersect with each other even before this acquisition? [00:13:47] S2: Yeah. So if you think about Marvell, they are a networking semiconductor company. They do serve both optical and copper connections. But as I'm sure you know, optical connections are becoming more and more important. And Lumentum, of course, they sell optical components and lasers. So by combining the two under one house, Marvell sees the opportunity of this extreme co design to develop some of the best solutions for all the four networks within AI infrastructure.
00:13:47 · Tore Svanberg, Akash Pasricha
[00:13:47] S2: Yeah. So if you think about Marvell, they are a networking semiconductor company. They do serve both optical and copper connections. But as I'm sure you know, optical connections are becoming more and more important. And Lumentum, of course, they sell optical components and lasers. So by combining the two under one house, Marvell sees the opportunity of this extreme co design to develop some of the best solutions for all the four networks within AI infrastructure. [00:14:21] S0: So just to be clear, these are the chips, these are the lasers that they connect different chips together within a server. They connect different servers together. They also connect different data centers together. Is that right? [00:14:35] S2: Yeah. So when you think about the four AI networks, there's basically scale across, scale out, scale up, and scale in. Today, most of the AI infrastructure is in scale out and scale across, and those do use optical connections today. However, the next generation networks are going to be scale up and scale in. Those are based on copper connections today. But because the speeds are expanding so exponentially, that's where optical technology is now set to solve some major issues. [00:15:08] S0: Right. The idea here, just to recap for folks, scale out and scale across, this is between servers and then between different data centers. Scale in and scale up is basically within the chip and then up inside the servers. Is that right?
00:15:08 · Akash Pasricha, Tore Svanberg
[00:15:08] S0: Right. The idea here, just to recap for folks, scale out and scale across, this is between servers and then between different data centers. Scale in and scale up is basically within the chip and then up inside the servers. Is that right? [00:15:26] S2: Exactly. So scale out is about connecting the racks together. Scale across is basically connecting data centers together. Scale up would be basically looking at racks that are quite large, quite tall. And then scale in would be millimeter type connections between compute and all the other various components within an AI rack, such as NIC cards, storage, memory, and so on and so forth. [00:15:56] S0: And so Lumentum, they play largely in the more microscopic scale in category. Is that right? [00:16:07] S2: No. So scale in today is all copper connections. [00:16:10] S0: Copper connections. Okay. [00:16:11] S2: But next generation scale is going to be based on optical. So today, Lumentum plays predominantly in scale out and scale across, which is where optical technology is used today. But as I mentioned before, as speeds continues to move higher and higher, copper runs out of steam as a connectivity medium, and therefore you are going to need optical connections. [00:16:34] S0: Got it. Okay. So now let's talk about the strategic logic then for this deal from Marvell. Sounds like Lumentum has some pretty forward thinking technology here that is going to become important. Do you think that this is a good deal for Marvell and ultimately we're going to get to the price, what do you think of what they paid for it?
00:16:56 · Tore Svanberg, Akash Pasricha
[00:16:56] S2: Yeah. So first of all, today, semiconductor companies and optical companies are not one on the same. They have different capabilities, different IP. But what's happening is that with the convergence of electrical signals and optical signals, or in this case, semiconductors and photonics, I think Marvell sees an opportunity to bring it all under one house and therefore be able to deliver some better architectures, some better products, and certainly some better performance solutions to their customers. As far as the price tag, yes, people look at the headlines like it's quite staggering. This is one of the biggest deals in semiconductor history. But let's not forget what's at stake here, right? We're talking about AI infrastructure, which is probably the place to be in technology over the next five to seven years. And what they are bringing to bear here is some of the most complex solutions you can find. I do understand for investors it's confusing again because we haven't really seen anything like this before. I think we're all used to thinking in verticals. But I do believe that by bringing these two competencies under one house, Marvell should be able to address next generation AI infrastructure in a much better way than what they are today. [00:18:16] S0: What do you think Lumentum shareholders are reacting to here? Because, I mean, look, typically a deal gets announced, the target, right? I mean, those investors get excited and the acquirer, those shareholders get nervous. Here, it was both of them getting nervous. So Lumentum's anxiety is what exactly?
00:18:16 · Akash Pasricha, Tore Svanberg
[00:18:16] S0: What do you think Lumentum shareholders are reacting to here? Because, I mean, look, typically a deal gets announced, the target, right? I mean, those investors get excited and the acquirer, those shareholders get nervous. Here, it was both of them getting nervous. So Lumentum's anxiety is what exactly? [00:18:37] S2: Yeah. I mean, I think this has more to do with the length of when this deal is going to close, right? So it's not going to close until first of the twenty eighth. There is going to be some regulatory scrutiny. So I think what Lumentum investors are worried about is, could this guild actually go through? Will it get approved? And given the timeframe, how long this is going to last, could that cause some disruption to their business model? Could competitors come in? Could customers get nervous? So I think that's what Lumentum investors are basically worried about. Do you think it'll go through? I personally think it will go through. There's no regulatory reasons why this wouldn't go through. Certainly if you just use the laws out there and the legality, there's no reasons why this should cause any antitrust anxiety. But keep in mind, this is going to need both China and US approval. As you know, politically, there's been some concerns there. But again, just purely from an antitrust perspective, I think the deal is going to go through.