원문 보기
So, on the same day that we received the stunning Google reports, which we have already analyzed in a separate segment, information appeared about a mega-deal between Marvell and Lumentum, which is allegedly being prepared and could take place as early as next week. Well, the deal could be announced as early as next week.
This is simply crazy, because we are talking about Marvell, a company worth $221 billion, which Jensen himself called the next contender for a capitalization of $1 trillion. And another company called Lumentum, ticker lite, l-i-t-e, which is quite appropriate considering they work with fiber lasers.
Do you understand? Lite (light) . In any case. This is a company with a capitalization of 83 billion dollars. This is a crazy potential purchase for Marvell, which doesn't have the balance sheet to justify the move. That's why stocks are falling. And that's why lite stocks are rising. So let's get into a little bit of background.
The gist of what's happening is that Marvell's CEO is betting on co-packaged optics, CPO, and fiber laser interconnects. Of course? This is different from a copper connection. Who supports the copper camp ? Jensen says: use copper for as long as possible, because it's much cheaper, right? So think of it as a way of connecting a few different things.
Either data center with data center, rack with rack on the other side of the data center, or inside the rack— chip with chip, or even inside the chip itself, one part of the chip with another. These are four levels at which you can connect elements. Fiber has already practically provided communication between data centers, and they are gradually switching to fiber-optic connections between racks for faster data transfer with less latency, which becomes extremely important when you work with artificial intelligence and want to get answers quickly for customer support, voice AI, or agents.
All of this will become much larger, and investments in this area are just beginning. But the bottom line is that Marvell is making the argument that we will start to see a shift to all- fiber connectivity between racks. We will see how chip-to-chip communication moves to fiber, and then inside the chip itself.
Because the speeds will become so crazy that you will encounter the so-called "copper wall." Now Jensen says: use copper while you can. It's cheaper, okay? For Nvidia, this means higher profits. Broadcom CEO says, "Look, we'll use lasers, but we prefer copper." So, that guy, the CEO of Broadcom, I actually like him.
I think he is a very humble person. Broadcom, I have a few notes here. Broadcom holds 60% of the ASIC market. I'll just show them to you. This is actually on our research tab for course participants only, but never mind. Uh, Broadcom has 60% of the ASIC market. Meta and Anthropic have also joined Broadcom in joint development.
The real competition they see is, you know, Nvidia. But the GPU is , you know, your all-purpose workhorse. ASICs are very specific. It's in the name: specific. Specialized integrated circuit. Marvell only has about 15–25% of the ASIC market. So, Broadcom is the leader here. And that's why Marvell is taking a different approach than Broadcom.
Broadcom says, “ Look, we’re going to use copper as much as we can.” In fact, we have new copper products that can work very quickly. " That will be enough for us for a while." Marvell says, “No, man. We are going to switch to optics because we believe that optics is a safeguard against future obsolescence.” Well, that's what they did, but you know what I mean.
And that copper will be a thing of the past. Do you know what the CEO of Broadcom literally says? He says, uh, when asked about fiber optics, right? And that's literally his phrase. I wrote it down. This is on the research tab for course participants. Remember, you can become a member and get lifetime access to stuff like this at reinvest1.com.
This is just a sample of the materials we will be talking about in the reinvest research terminal later this year. But, in any case , he writes: "As for optics, oh, I try to avoid bright shiny objects." Obviously, you know, he's joking here, because fiber optics are bright and shiny. In any case, I like him.
I really think he's a smart person. I think the answer is to buy both, right? Both Broadcom and Marvell. This is not personalized advice. I have investments in both. Complete transparency. Uh, but uh, Google uses Broadcom to manufacture its TPUs. Marvell will also receive a portion of TPU revenue. But these are two very big differences between the perspective of Broadcom and Nvidia and Marvell and Lumentum.
Now you might think that this makes Nvidia pessimistic about fiber. No. Nvidia invested $2 billion in Marvell and $2 billion in Lumentum. In both . Nvidia has a total investment of $4 billion. This is really interesting. I think Jensen is trying to play it safe. Like, today we'll use copper. This is good for margins.
But if fiber optics works, we predicted it. We have invested in and call Marvell a future $1 trillion company. This would be a 5-fold increase from current values. OK. Well, that's cool. And they obviously want to ride this wave of growth. But what do they see? They see that Nvidia is going to force Marvell to integrate its products into Nvidia's ecosystem as much as possible.
Then Nvidia and their entire CUDA "wall of defense" will remain strong, and Marvell will not be able to undermine Nvidia's leadership. They're essentially saying, "Look, we want everyone to buy our GPUs." This is our ecosystem, okay? We have Ethernet, NVLink, Vera processors, Rubin GPUs, Blackwall, etc.
We have the whole stack, baby. We have switches, whatever . We have Spectrum X—it's copper Ethernet. We have everything. You don't need anyone else, says Nvidia. And then there's Marvell, like a little animal that says, " Yeah, but we design all these ASICs, all these people make custom chips." And they would really like to integrate them into your Nvidia system, into your ecosystem.” You may ask: “ Okay, but what chips do they integrate with?” The answer is, with many of them.
Marvell is developing the Trainium XPU, which provides about 15.5% of Marvell's revenue. 8.4% of their revenue comes from, I don't know how to pronounce it, Maia—a custom XPU for Microsoft . And also Google's Axion processor and some of their TPUs. And also developing a custom chip for Meta. So Marvell is making all these deals with Amazon, Microsoft, Google, Meta.
These are companies that are spending huge amounts of money on AI. And Nvidia is like saying, “Are you working with all of them? Good. Good. OK. How about this? We're going to create something called NVLink Fusion, and you can be one of the few who can use it. We will open our arms to you, Marvell.” You can bring your junk to our racks.
Oh, you want to ride my rack? Just connect to Envy Link Fusion and Marvell will be one of the partners. And then Marvell is like, “Great. We infiltrated the Nvidia stand . This is simply amazing. We're in deep, guys. We are deep. How can we go even deeper? Let's buy out this whole damn stack." That's actually what they say.
Buy out the whole damn stack. Because you see, Marvell is a prominent ASIC manufacturer, they can design, but they don't do everything inside the rack. So you have to understand: there is a rack, and there are cables. These cables can be manufactured by a company like Corning, which produces glass for fiber optic lines .
And there's also something that provides light, a device that emits light, right? Lumentum makes just such things. Of course? Marvell then makes the chips that collect or convert this light data, the data between the switches, but Marvell also has its own dedicated ASICs. So Marvell says, “Look, we can take over the GPU work for you .” We can do Marvell DSP.
This is, you know, from the server to the switch. We can provide communication between switches using our photonic chip. We could then buy Lumentum, which provides light.” The only thing they haven't bought yet is Corning, which, you know, is a major glass and fiber manufacturer in this vertical. So, Corning is a company with a capitalization of $113 billion.
This is a company with about 150 years of history. I'm actually really excited about Corning. I do not own Corning stock. I'm also really excited about Lumentum, and I'm a little mad at myself because we talked about Lumentum during the live streams for the course participants, I was talking about how much I liked them and how interested I was in them, but, you know, I wasn't sure yet whether I should buy them.
And so, I was like a cowardly baby and didn't buy them on time, and now they've gone up 10% because of this purchase. Of course, you know, it doesn't matter. But we talked about Lumentum because this whole fiber-optic story is just beginning. I think we are still at an early stage. I know some people say, "Oh, it's too late." DSPs, by the way, are digital signal processors.
It's like the brain inside the optical modules for fiber optics, and Marvell designs them, right? Lumentum is the manufacturer. So, by buying Lumentum, Marvell is also gaining manufacturing capabilities that they don't have right now, which is pretty impressive. By the way, Jensen's quote on this subject is: " Use copper when you can." " Use optics when you have to." Ahem.
Quite an interesting thought from, uh, Jensen. He always has some interesting remarks, doesn't he? But personally, I think it's very exciting because Marvell, uh, I was also watching another interview with Marvell. This was, well, an old interview, but it allows you to understand the CEO's train of thought, he believes that the next...this video...This sponsor— meetkevin.com/silo.
Wait. Points for promotions? You may have already heard of our today's sponsor Silo. This is essentially the same thing that Build did for rentals, but this time for investments. So, if you're not familiar with it, here it is. Silo is the first and only investment platform that awards points like credit cards, but to your brokerage account.
Every month they award you points based on the value of your account. You receive an annual multiplier of 0.5 x points on your total account balance. So, if you have $500,000 in your investment account, you will receive 50 basis points. That's 250,000 points every year. Damn, I'm losing points. Points can be exchanged for cash.
You know, many of us are probably thinking about this. Like, dude, wait, what? I want points. What can I do with them ? Here, please. Points can be exchanged for cash at a rate of 1 cent per point. Or you can use their travel portal to book tickets at a rate of 1.3 cents per point, depending on your membership level.
So, that's the main point. They are also launching their first points transfer partner. It may already be working by the time you are looking at this integration. Now, it's worth noting that while I can't reveal the names of the airlines yet, I can say that it's a Star Alliance partner with a one-to-one transfer rate , which is pretty cool since Star Alliance partners allow you to book flights on 26 different airlines, including United, Turkish, Swift Air, Lufthansa, Singapore Airlines.
So, if you're smart about how you use these points, you can fly in pretty comfortable seats, in business class or even first class, simply because your portfolio was in Silo and they were paying you base points from your portfolio. So , with the launch of its first remittance partner, Sylo is officially coming out of beta and moving to a long-term points accrual rate of 0.25% in November.
But anyone who signs up this month locks in those 50 basis points for at least the first year. And you may even be eligible to keep the old terms, but of course, check the details of their program. So if you want to lock in 0.5x points on your shares, 50 basis points, and get up to 250,000 points as a new member, be sure to use my link, meetkevin.com/sylo.
It's simple. meetkevin.com/sylo. S I L O to start by the end of this month. He believes that the next stages of growth are not just about transferring data. For example, data transmission is something they do using fiber optics. This is also data storage. And this becomes a very deep topic. Sorry this is so narrow-minded and geeky, I apologize, but I have a belief that...First of all, I...How should I put it?
I think the next boom won't necessarily be directly related to memory, but the area that sold off along with memory and could recover more quickly is data storage . It sounds crazy, but generative AI consumes so much damn cache memory over time, and robotics will only make it worse. What will happen is that you will need more storage.
And eventually you can use a lot of external storage where you have all the NAND, SSD, well, everything that's in that room over there. And all GPUs and ASICs are here . Well, how are you going to put these two things together? What about fiber optics? Quick, little one. See how all these things start to come together?
You know, connecting to each other. This is madness. However . And I'm not here to say that I can definitely name you the company that will show the best results from this entire stack. I think in this environment, spoiler alert, just between us, okay? Don't tell anyone. I think it's worth betting on the equipment, this is not personalized financial advice.
But I think you have to be very well diversified within that. Because if you invest all your money in one of them, it can do like Google, right? So you diversify among these hardware players and just bet on this whole AI ship going up. You know, unless you're a bear in the market. And then you do the opposite.
But if you're a bear, you might not even be watching this. Good. So let's take a look at the real fundamentals. So, the last fundamental analysis of Marvell that we did during the live broadcast for course participants was on October 5th. Remember, everything we're looking at, much of it will be available in our research terminal in the future.
This is part of what will come out as part of Reinvest One. You pay once and get lifetime access. But Reinvest Research is coming soon. You will also gain access if you join our live streams for course participants, where we conduct a lot of fundamental and technical analysis. We'll look at these ideas and try to understand, "Hey, what's up with the technicals?
Where are the fundamentals? What's worth buying?" We are just trying to understand more. We are trying to find out more. This is a learning exercise where we try to expand our knowledge every day about what the hell is actually going on with artificial intelligence. And a lot is happening . We think we're doing a good job with this, but we'll do even better when our research terminal is released.
And the terminal will literally have a module about building a business on agent AI, or improving your business with agent AI, or how to be a better consumer with agent AI. You get live streams for course participants, nine courses, a research terminal, a stock terminal, a real estate terminal. You get it all in one package at reinvestone.com.
We think this is an incredible offer and one you will really want to join in 2026. It would be like full autonomous control, where suddenly after 2026 you would no longer be able to get lifetime access. And then you say, "Damn, I should have gotten lifetime access." Because I think in a year people will be like, "Wow.
This is actually really cool ." We'll see. It's still early. OK. So, this is Marvell's operating performance report. This is our revenue from Marvell. We have a 36% increase in net revenue. The cost of goods sold increased less, so their pricing power increased. So, we see a strengthening of pricing power in Marvell.
Very optimistic about Marvell's pricing power. I love them. They also got rid of the automotive business, and they had free cash flow with a yield of only about 1.68%, which is pretty low , but there is some free cash flow. They also borrowed and issued some stock. So, you know, they're also in this pursuit, saying we need to build a lot here .
Oh, you can see my note here. I wrote: Broadcom has some optical ambitions, but copper connectivity is their significant advantage. Marvell is included in CPO optics. Of course? Yes. I wrote that I considered Marvell to be the discounted half- sister of Broadcom. It's a little more expensive, which is somewhat ironic because I actually think Marvell's stock price is higher than Marvell itself right now.
But I mean market capitalization. Of course? In terms of market capitalization, we're talking about $200 billion versus about a trillion dollars for Broadcom. Maybe even more, maybe one and a half trillion. I don't remember exactly. But, in any case, when I say "discount", it means that Broadcom's valuation is lower than Marvell's.
But Marvell is the less promising half-brother that is betting on optics, and this acquisition of Light… Damn , it annoys me. No, actually it's great. I am very optimistic about Marvell. I would buy Marvell shares on a dip because I want to see a merged company. I mean, I would be very happy if that happened.
Obviously, Marvell shares will fall and Light shares will rise. This happens every time there is an acquisition because Marvell has no money. Yes, they don't have $83 billion in cash. They have about $6 billion in free cash. So it will be debt and equity. This will lead to capital erosion. Yes, Light has some cash on its balance sheet too, but not much, man.
They have $3.8 billion in debt, and $3.4 billion in cash and accounts receivable . So there is no real money in their $80 billion capitalization. So , Marvell will have to take on debt and issue shares for this deal. So yes, Marvell stock will fall in the short term. But they are creating a real giant.
It's just a pleasure. Thanks for the mention from 8236. I like Reinvest One. Thank you, Kevin. Oh, yes. I appreciate it. So, take a look at this. Here's what I wrote. Yes, that's what I wrote on September 25th. Oh, September 25th, it's my passion's birthday. Well, that was my high school love. I don't know why I still remember her birthday.
This just makes Lauren angry. Don't tell her I said that. Um, that's pretty obvious. Uh, I got rejected by a girl in high school, okay? I invited her out for pizza, and she said no. And I still haven't forgotten it. Er , in short, this is a powerful play in the field of photonics for the artificial intelligence megaboom, as has already happened with memory.
So, there was this mega-boom, and stocks have actually fallen a lot since we did this analysis . Uh, and I thought they were cheap, they were a little expensive back then. So over the last, you know, 4 weeks, stocks have fallen a little bit . Uh, I said, at a PEG of 1.2, there are cheaper options like Broadcom and Marvell, they're actually cheaper.
However, I said that there is still growth potential here if this growth can be sustained, and it should pick up from the " hardware 2.0" rally that I see coming, you know, after the Iran situation. They see that net income increases by 1.8 times, and gross profit by 2.7 times. So you're talking about huge pricing power here.
Their operating expenses increased by only 14%, and their margin increased by 41%. That's attractive, man. Yes, we had a loss here. Why did the damage occur? This was due to debt repayment. So you should ignore this nonsense on the right. This is more of a one-time occurrence. Uh, it's Sidney Sweeney's birthday.
Was her name Sarah? No. Her name was Kristen because of the "game ". Er, in short. So, um. Good. Investment activity, and let's see, 100 billion. Good, good, good. Accounts receivable, 3.8 billion accounts here. We've already discussed this. This is financial information . Assets, yes, we have already considered that as well.
Good. So, let's take a look, let's dig into this a little more. Good. So Corning has huge supply deals. Corning is also an interesting option. I, uh , let me do a quick assessment of Corning. Er, Corning is GLW. They have a PEG ratio of about 1.39. Great pricing power. But they have an outdated business, right?
So, half of the business at Corning is so-so , and the other half is great for fiber optics. It's just not such a clean business. So you're actually paying a fair price for the entire company when only half of the business is thriving. Still, you know, Dell did well when only a tiny portion of their business thrived.
And honestly, Dell has become cheap. Dell has become significantly cheaper. So, uh, that's very interesting. Now, if I look a little deeper, now inside the racks the GPU-to-GPU connections are 100% copper. But copper cannot scale beyond 3m at the speeds needed for next- generation clusters. And the higher the speed, the more the distance that copper can cover is reduced, literally by half.
It seems to be said that for every doubling of the speed you want to achieve, the distance covered by copper is halved. That's why Corning is seen as a huge opportunity for scaling with fiber, and that's why Marvell says, "Oh, come on. " Even if we have to suffer losses in the short term, we need a full stack.
We need a laser, we need processors, we already have DSPs, we have ASICs, we have the connectivity to get inside the Nvidia rack. We have almost everything for full vertical integration. This CEO acts exactly like Elon Musk. And I like this guy . I-I think he's more ambitious than the guy from Broadcom.
It seems to me that the head of Broadcom is more conservative. This really increases the execution risks. But I like this guy . I respect him. I respect this acquisition. I'm a big fan. Uh, and even though the company I own, Marvell, is going down on this, and Light, the one I wanted to buy, the one that slipped away, is going up— I have to say, overall, it's brilliant.
Big fan. And this only confirms that we are in for an artificial intelligence boom. It is just beginning. And you get a deal with Iran, man. Well , baby. Who is Kevin without the Iran deal? But that will change, we will head into the " hardware 2.0" rally, and these valuations, which are now super cheap, will lead to a euphoric rally.
Then people will say, "Oh my God." This is what a bubble actually looks like. This is my opinion. Maybe I'm wrong. So, for now, I'll just play it. Oh man, your boyfriend Kevin just lost a bunch of money. Oh, but it's okay . I'm with you. Together with you. They drove. Well, because. Squeeze. Buy on the dip.
Buy on the dip. If you liked this video, watch this . I think you 'll be delighted. Why not promote these things you told us about here? I don't think anyone else knows about this. We'll try some advertising and see what happens. Congratulations, dude. You have done so much. People love you. People look up to you.
Kevin Paffrath, financial analyst and YouTuber, Meet Kevin. It's always nice to hear your opinion.
Comments 0