저자: Kevin · 외국인
요약 보기
Elon Musk(SpaceX/Tesla)의 AI 하드웨어 대규모 지출이 Nvidia·Marvell·Enphase·ARM 등 AI 하드웨어 종목에 수혜를 준다는 주장. 랙 주문 13,000→최대 68,800 시나리오, 비용 2,750~3,200억 달러 추정. Nvidia PEG 0.76·적정가 795~850달러, 2028년 매출 상향 여지 강조. Marvell은 NVLink Fusion 수혜, Dell은 마진이 낮고 고평가 경고, Schneider는 밸류 부담, ARM은 고PER로 부정적, Enphase는 잉여현금흐름 수익률 근거 저평가 주장. 영상 말미에 Nvidia를 '가장 싸다'며 최선호로 꼽음.
다듬은 글
Hello everyone, this is Kevin. Hey guys, in this video I'm going to be looking at nine stocks, and I have to say, a lot of people have been asking me, " Kevin, what do you think about Tom Lee saying that 2027 is going to be an incredible growth year?" And I have two things to say about this . First, this guy is always looking to grow.
Secondly, he is not optimistic enough about 2026. Look, I actually think so, and we've mentioned this in our SpaceX analysis videos before the company went public. I truly believe that Elon Musk will spend every cent he raises on equipment like a drunken sailor. Just last week, their earnings and financials report confirmed that 85 cents of every dollar of capital spending goes directly to AI hardware.
So, in this video, we'll break down which stocks could be great opportunities in the AI hardware sector, given that we expect Elon to spend wildly on macro and mini hardware, plus the upcoming "terrafab," which some of the stocks on this list are directly tied to. I just want to be extremely clear, because sometimes people who only read the headlines or don't watch the entire video don't get the point.
Maybe if I say this at the very beginning of the video, I can solve it. We have over a seven-figure investment in SpaceX through our venture capital fund. We bought this asset at a valuation of about $330 billion. Obviously, this means we are in the black about 5 times. Okay, great. We'll see when we actually get the shares.
We don't have them in our hands yet, although some of the blocking periods have already ended. We have no idea when these shares will finally be with us. When I shoot a video where I describe under what conditions I would be willing to buy SpaceX shares— below $80. This doesn't mean I think the price will drop to $80.
This means I don't want to buy them until the price drops below $80. I would not increase the position relative to our initial one. But this video is not about investing in SpaceX. This is not an attempt to be a bear about SpaceX. This is not an attempt to be bullish about SpaceX. The sole purpose of this video is to say: if we believe that Elon will spend money like a drunken sailor, then who benefits from it?
Okay, now that we've got that out of the way , let's get started. And we'll look at these nine stocks. Keep in mind that I intentionally did not include the memory sector here. Memory will be a topic for a completely different video. We might even do an in-depth analysis of some of these companies. You can write in the comments whether you want to see a more detailed analysis.
But we'll focus on just nine different stocks outside of the memory sector. So, first the number of racks. Baseline scenario: Foxconn's current order for Mac mini hard drives—13,000 racks. That's about $52.5 billion in spending. This order has already been placed. The fact that this order is going to Foxconn will really impact one of the stocks on this nine-position list.
There's also a false hope being spread right now because of the SemiAnalysis article that everyone in X is talking about. When that article came out, I left a comment within minutes of their post saying, "Hey guys, there 's a mistake in your material." But everyone doesn't give a damn about it . But SemiAnalysis' bullish forecast is that we wo n't actually see 13,000 Mac mini hard drives.
This is the current forecast. The current forecast from SemiAnalysis under the bullish scenario is 68,800 rack orders. So, 68,800 divided by 13,000 is about 5.3 times more orders. Which translates into a cost of approximately $275 billion. Perhaps even up to $320 billion in costs. And that's where people say, "Okay, what's this based on?
Is this based on net profit projections?" Well, that's the problem I have with the SemiAnalysis article. You know, my argument is this: how can you claim that Elon will earn $15 to $20 per hour of GPU work, and in the same damn article write that Anthropic will only pay $3 per hour of GPU work? You can't play the best-case scenario both ways by giving a discount for LLM and a premium for Elon at the same time .
It doesn't work that way, because the LLM developers pay Elon, okay? If you're going to use $15 per hour GPU, apply that to both sides. Of course, no one answered, so it doesn't matter. But people in X or those who just do short summaries through GPT don't care because they don't have the patience to understand the nuances.
And that's good, because it creates opportunities for people like you and me. So, my point is that it doesn't really matter, because Elon doesn't care. This is just between us, okay? He doesn't really care about profits. I know. I know this sounds crazy. But I think Elon will literally expand the offering no matter what, because he's obsessed with the idea of, "Oh my God, we need so many data centers." Like, we need to conquer space, and then we'll have to produce millions of chips for all the Optimus robots.
Oh my God. Keep in mind, that's still ten years away, right? Regarding Optimus robots. It doesn't matter. It doesn't matter if there will be FSD or if there will be orbiting satellites. It doesn't matter. All this nonsense doesn't make any sense. All that matters is that the richest man in the world believes it.
And if the richest person in the world believes this, he will spend a lot of money, regardless of the return on investment . And there are a few big stocks that stand to benefit from this first and foremost. The first on the list is Nvidia. It doesn't matter if Elon makes a profit, they will increase every time an order comes in from Elon.
It's no wonder Jensen flatters Elon and calls him, you know, the greatest or something, right? The best of all time. That's because he's an incredibly big customer. This is a real whale. You know, if you were a real estate agent selling $500,000 houses, and you were working with a buyer who was buying their first home, and a whale came along and said, "Hey, we want to buy 10 houses for $500,000." Who will you answer the call first?
Yes? I mean, I just happen to have a 5090 graphics card here. And these things, frankly, are real beasts. I mean, it's really amazing, because most of what you see here is just cooling. But I probably bought this same chip for around $2,500 when it first came out, which wasn't cheap. They are now selling for around $4,900, meaning they have almost doubled in price .
We have, you know, most of this—it's obviously cooling, okay? Here we have cooling radiators. There are three fans here. Some models for smaller chips are sold with two fans. Here in the back you see some capacitors. Underneath this is actually the crystal and the "brain". Crystal is actually a chip, right?
And this is just a printed circuit board, this textolite. And most of it, about 85%, is the frame and cooling. And this becomes quite important for what we're going to talk about, because a lot of the other promotions are related to some of the other components of these GPU chipsets. So, Nvidia has apparently confirmed a direct deal with Elon Musk .
And Elon calls Nvidia chips simply the best. With 13,000 GB300 server racks, you'll start to see Nvidia's forecasts increase, likely starting in Q4 2026. And this is what I consider part of the second hardware rally after the ceasefire. Let me just clarify this. In our live broadcast, we predicted that there would be a second hardware rally.
And this is exactly what we were talking about in July when we were buying the dip , because the Leo Fold blew up. I'm not trying to force this on you . I'm just trying to give you an understanding of what we were talking about, okay? Some of these things, that's normal. I'm just going to share this with you.
I appreciate you being here. The second hardware rally, in my opinion, is largely driven by Elon's spending, as it will significantly revise up the forecasts these companies are making for real growth. And keep in mind, when we look at Nvidia, this thing has 75% gross profit. And out of their total revenue of $81.6 billion, they earn 71.4% of their net profit.
This is madness. Some of the stocks we 'll look at have net returns of around 9%, 18% , 20%, 28%. These are normal net profit figures. 71.4%—I mean, that's not what happens in the memory world for this AI sector. These are crazy profits. And of course, the biggest problem for Nvidia and one of the reasons it's stuck in the plus or minus $200 range is that people were worried about the growth projections and the possibility of those projections turning negative in the future.
But even though there is the potential for these predictions to become negative in the future, the only thing that can actually turn this negative into a positive in the longer term is Elon. Nvidia is now up just 20% year-to-date, despite being the largest direct beneficiary of SpaceX. And this has nothing to do with Twitter monetization.
This has nothing to do with monetizing ads through Grok or Grok subscriptions, and it really has nothing to do with the contract extension for SpaceX, which is the subject of much debate. People are debating whether Anthropic or Google can, or will they actually want to, continue at such crazy rates?
Are they doing this just because they're trying to pump up their investment in SpaceX? Nvidia is sneezing about all this. If Nvidia does indeed become the beneficiary of this $320 billion version of the all-in, what will it lead to? Possibly up to $100–150 billion in additional revenue for Nvidia. This is about 30% more than Nvidia's revenue projections for fiscal year 2028.
I mean, just think about it. Just look at it right now. So, now it's $81 billion , and if you calculate it on an annual basis, it's about $324 billion . So if you add another, well, about $140 billion, that would be about 30% of additional annual revenue for the company just because of one buyer. And this buyer is a person who is crazy about AI.
And maybe he will be right. Maybe he will just be incredibly insightful on this matter. He has acted flawlessly many times before . He has done fantastic things in the past. You know, betting against Elon—you have to guess the right moment. You can bet against Tesla at its peak, you can…and I feel sorry for those people, but I just saw it, even bookmarked it.
Someone just opened a $ 3.2-3.3 million short against SpaceX. Uh , and they posted a screenshot here. But what amazes me is that they are actually misleading people , because they actually only invested $1.5 million . They just put it in a 2x credit ETF on the, you know, short side. And I'm like, "Bro, you're taking 2 shorts when the price is literally at the bottom of the Fibonacci?" For example, on Friday morning people thought I was bearish.
And again , when people watch the whole video, I even said on Thursday morning, "Listen, people can't reasonably sell their blocked shares because they do n't have them yet." This puppy will roll back, and its next test is 133. We said that, you know, when the price was around 114. And look at this.
133— this is literally the first line of retreat, and that's where we stopped. I feel sorry for the guy who is shorting now. But again, this video is not about SpaceX. It should be about these actions. So, SpaceX now plans to use GB300 in most of these new facilities. But macros and stuff—by the time they actually start installing these GPUs, Rubin might show up .
And if Ruben is available, and we eventually get to the full 10 gigawatts that Elon is going to spend, which would require raising significant capital—more on that in a bit— then Nvidia's revenue could actually skyrocket, you know, total spending could go up from the estimates that we've seen, from $275 billion to $320 billion.
This could increase to around 400–480 billion. This is becoming pointless. And then Nvidia's share in this will be even greater. So their 2028 projections , solely because of Elon, Macro, Mini, Terrafab, the expansion of the original Colossus— only because of Elon can we literally raise the valuations for Nvidia.
You may have to increase their forecasts by 50%. So it is ironic that projections for the coming years may be underestimated. And really, the person to look to in the capital expenditure cycle is Elon. I mean, if this guy, well, for the purposes of this video, we'll still call him a trillionaire, even though he's a little on the verge of it.
Oh my god, you're a marginal trillionaire. You're just a nine hundred billionaire. You know, it doesn't matter. If that's what he wants to spend money on, fine. What's interesting about this is the Rubin Plus, which is significantly more expensive per rack. You know, the rack probably costs about 4 million dollars.
NVLink is needed , it's another 500 thousand to a million per rack. And InfiniBand, you need Nvidia to pay for the shipment. Yes, there is a risk that Google will cancel the order if Elon doesn't make it on time, but honestly, I think it's unlikely because Google is "locked in" and if they canceled, SpaceX would suffer, they would just shoot themselves in the foot and be forced to report poor earnings.
There is no chance that Google will cancel the order. They are strongly connected to Elon. In addition, the Vera Rubin processor is on the way, which, to be honest, I think the market has forgotten that Nvidia produces its own CPU. I know a lot of Nvidia bulls will say, “I know, Kevin. Just don't tell them.
I want to continue buying cheap. “But it’s actually a fantastic thing. I have an AMD CPU, it's probably ...one of the computers I have costs about 10 thousand dollars. There's the best AMD processor , the best power supply, and the 5090 graphics cards. We got some processors, in fact, we just ordered six more 5090s.
Um, that's for what we do with reinvestment. But yeah, I mean AMD is a good processor. But Vera processors are server processors based on the ARM architecture. And while some analysts include them in their price targets, I don't think people understand: if Elon buys up all those Vera processors, the pricing power in the processor market will become enormous.
can look at Nvidia's valuation. So if we go to the website app.meKevin.com, which as you know, course participants have full access to, then of course you can download this. You can also use the app on your phone, the Meet Kevin app for Android or Google. But if we go to app.meKevin.com, the web portal, there are my notes there, of course, but we'll skip part of it because we're talking about that now .
The moving average of price power fluctuated slightly. Some of these declines in pricing power are expected to be temporary. And in fact, their profits returned to 75% according to the results of the last quarter. We are going to update this software to use a larger exponential moving average that will give more weight to recent price power indicators.
And this will actually increase the estimation of pricing power in our algorithm. But what's really interesting is that this particular stock, in my opinion , has a new target valuation based on a PEG ratio of two and a half to a little higher, which raises the share price to $795-$850. That's roughly the fair value that we determine for this asset.
Of course, momentum can push the price higher or lower. For example, Leo Fold can lower the price, while market momentum can raise it. But the company is currently trading at a P/E ratio of 25. Growth projections for the next 4 years are 34% per year, which means it is currently trading at a PEG ratio of less than one, which is simply crazy.
The current PEG I see here is 0.76. This is a really low figure. This is the price/ earnings ratio to growth. We use it more often than just P/E ratios. If you only use P/E ratios, I believe you are unfairly valuing stocks because you are not taking into account their growth. But, you know, one of the reasons why memory stocks seem to be underperforming in terms of growth—that's a separate topic for another video , but it applies to Nvidia as well —is that people think growth is cyclical, and therefore the growth rates won't be long-lasting.
So this is where someone like Elon can obviously help keep these costs moving forward. But, you know, it's not just Elon. For example, we have some AI, or I would say some AI insights right here. Blackwell inference acceleration, NVLink 72 deployments or , yes, deployments scale among hyperscalers. This accelerates the economics of token generation, directly reducing the cost of withdrawal by 60% compared to previous generations.
There will be even more token cost reductions for Vera Rubin level products , you know, when they come out. This will be truly incredible. In addition to this, the expansion of the TAM of processors, which I believe is critical. The new Arm-based Vera Rubin processors , developed in conjunction with Rubin GPUs, are aimed at agent-based AI orchestration.
This alone creates a revenue opportunity of $20 billion per year. Not so much now, is it? Because if you look at that $20 billion of their annual revenue, let's say in 324, that's, you know, well, maybe 7-8 % or something like that. But I actually think Elon could create such a huge demand for them that prices would go up and you would actually get a higher margin processor, right?
And then , of course, network income, that's critically important. This will also be important for our next action. And there's even more that you can see here, which I think is very exciting. Red flags—those would be things like, you know, delays in H200 deliveries to China. I'm less worried about this.
I think the market has already really factored in a lot of this with respect to China. We do n't even see much volatility anymore when news comes out of China regarding this particular company. Of course, heavy reliance on vendor funding, cyclical funding, hyperscalers accounting for 50% of data center revenue, exposes Nvidia to a potential slowdown in capital spending.
Of course, we already know that. The whole argument here is that Elon, he's going to spend. And the thing about Elon is that as long as the stocks perform well, like stabilize somewhere, which they probably will, around here, he can call the banks and just borrow more and more money. And he's going to spend every dollar of it on equipment.
Now the next company is Marvell. So Marvell is exciting because they build custom semiconductors for companies like AWS and Microsoft. They call them XPUs, essentially ASICs, and products that, you know, other Nvidia competitors could use in their server racks for their data centers. The problem is, you don't want to buy an XPU from Marvell and then be locked out of using Nvidia products, right?
Well, don't worry. There is a solution . Smart old Jensen said, "No, no, no, no. If you need to use, we would of course prefer you buy Nvidia." He literally said it from the stage, it's great. " Obviously, we'd rather you buy Nvidia. But, if you buy Marvell XPUs, we have this new architecture called NVLink Fusion.
And that's a fancy way of saying, 'Keep using your A6s, your XPUs, whatever .'" If you want to adopt the NVLink Fusion architecture, we can allow you to integrate with our Nvidia hardware, like our GPUs, our Vera processors, or whatever. " This now allows Nvidia to essentially swallow up these other data center chips , and Nvidia just takes control of everything.
Google TPUs haven't joined the NV Fusion partnership yet. Marvell is an early adopter, but many people actually see this as a structural advantage for Marvell, because as long as Nvidia wants to do this in the data center: come here, baby. Come here. Come here. Come here. Come here . Come here. Right?
Uh, Marvell says, "Hey, I've got these chips here." I want you to hug me, Dad. How do I get inside? "Uh, here, we'll make this little product for you, and it'll let my hand do it. Right ? Now, this is interesting for SpaceX, because the Terafab D3 and AI6 chips, the D3 for orbital missions, the AI6 chips, for example, for FSD and Optimus robots, the AI5 and AI6 chips, they're going to need NVLink Fusion to work with all this Nvidia hardware that our good old Elon is spending money on.
So NVLink Fusion is becoming more well-known, more popular, and indirectly indicates that ASIC manufacturers like Marvell can continue to sell their own silicon to anyone without worrying about not being able to integrate it into, say, Terafab or whatever. So technically you could see Marvell chips inside Terafab.
Now, it's more likely that these will be Tesla-specific products, especially because they want to manufacture them themselves, or Starlink or SpaceX products. Again, they want to manufacture it themselves. That's fine. It's going to be something on kind of a parallel competitor to Marvell. But that's not a big deal.
The whole idea is that the more popular NVLink Fusion becomes, the more confident customers feel about buying products like Marvell's custom chips. And that's great. So you don't have a direct beneficiary here. It's more like, "Oh, sure." It makes NVLink Fusion more popular and visible to people. Hey, we need chips today.
Hey, let's buy more Marvell chips. Obviously, there are other promising directions for Marvell. It's CPO, future base stations, AI ideas, like the AI RAN that Jensen and the Marvell CEO are talking about. It's the idea that in the future, your mobile phone, like a Google Phone, iPhone, or iPad, can do a lot of the computing directly, but you need access to the data.
And you can get that through base stations, where T-Mobile has its antennas or something like that, right? Marvell and Nvidia have partnerships to make chipsets that will allow for that . So instead of just having data in data centers, some of the workload will be in base stations—the nodes for 5G and 6G.
It’s more of an option for the future. In the near term, “Okay, this will popularize NVLink, probably good for Marvell .” What’s their valuation now? In terms of valuation, we think a fair price is around $370. That’s about 69% of the upside potential from current values. The growth forecast is about 35%.
$4 EPS as of Jan 27. 35% average growth over 4 years. The PEG ratio is now about 1.5, we need to check. Let’s see. We’ll find it. We have a moving average, analytics. The current PEG ratio is 1.43. Fundamentals point to a range of about $350–377. Momentum could push the price higher or lower. The margins are growing, and that’s good.
I like to see it's on the chart. If we go to the analytics here, we see... There it is. One of the first things you see is a strategic partnership with Nvidia to integrate custom chips and optical networking directly into the Nvidia ecosystem. Jensen still favors copper cabling as much as possible. Marvell believes that we will move to fiber optics even inside the racks and the data centers themselves, which are currently mostly copper.
If we get to the point where Broadcom stops pushing copper ( although they have optical solutions) and we really move to optics, that's a huge potential for Marvell. So it's unclear whether Terrab is going to be an Ethernet-based data center or an optical-based data center. If you figure that out and you figure it out that it's one or the other, that's either a bearish or a bullish signal for Marvell.
It's probably more of a base case scenario that's positive for Marvell. That's something to watch out for in Elon's comments, right? Uh, accelerating interconnect growth is great. We have forecasts for their revenue growth in optics, custom semiconductors and designs, the target is $10 billion. Uh , the data center interconnect business is great.
Switches... I mean, we could read... You can pause the video and read some of this. Uh, there's a lot of really interesting information here about what this company does. Uh, so this is just a quick overview. Okay, great. Now let's move on to the next company. So the next company is much smaller. Uh, this one...
I mean, you should look at it. Marvell's market cap is now $191 billion. Nvidia's market cap is $5.4 trillion. The next company we're going to talk about is quite tiny, which increases the potential volatility. Well, relatively tiny. It's still a pretty big company. Uh, this company is called Lattice Semiconductor, LSCC.
It's a company that specializes in field- programmable logic integrated circuits ( FPGAs). So they make them. Uh, historically, there's been one to two of these in every server rack, and we expect that to increase to three or four per rack. Now, keep in mind that these are relatively cheap products.
So it's not like you're making huge profits, although server revenue is about 38% of Lattice's revenue. And that's expected to increase. But because they're cheap, their growth potential is lower. Uh, but because of the small size of the company, it could have more volatility on the upside , especially as we see more FPGAs inside actual racks.
They've also acquired a management platform, which should further increase their margins. But remember, every GB300 and Vera Rubin has FPGAs built in. So let's look at a potential target price for them. Unfortunately, their margins are falling a little bit as they try to integrate the company they recently acquired.
Management expects them to grow. Their current PEG ratio is only 0.42, which is low . That's based on the growth that we're seeing and that some analysts are projecting going forward, which effectively means a doubling of earnings per share. If we see growth slowing to 40% or 50%, as some other analysts are projecting, then the future price of the stock could be lower.
So right now it's trading at a P/E of 61. If you're expecting 100% growth, that's how you get a PEG of less than one. But if you're expecting 40% growth, that's how you get a PEG of about 1.5. Right now, the average valuations that we have for the next 2 years are very volatile. There's no good 4-year outlook for this company.
It's too small. So I think on the bright side, the valuation could be to $763. In the pessimistic scenario, it could be about $225, which of course gives less upside potential. If we go here to LSCC, we'll see a price of about $130. It's been down since Leo Fold. And again, that's a market cap of about $18 billion .
So this is a small manufacturer of components inside the rack. I think these things cost a few dollars a piece, honestly. Let's find out how much an LSCC programmable logic integrated circuit inside a server rack, like a GB300, costs. Let's find out. Let's just see what it says. So it's a few dollars.
Yeah, right. So a single chip inside a server rack from Lattice is usually $15 to $50. So they're cheap. And again , the idea is, "Hey , can they get a hold of themselves and put a few of these on every motherboard, or not?" yes? "Look at this. A single off-the-shelf server rack can contain anywhere from $1,500 to $5,000 worth of Lattice hardware and software.
Pretty interesting, right? The smaller the company, the more speculative it is. So of course there's more risk. Now let's move on to the next thing, which is cooling the chips. On my AMD processor, on my main big computer, I use water cooling. This thing is fan-cooled, it has three huge fans, and it needs a lot of cooling.
This little die gets really hot, especially if you're running 4K products on it . So just imagine the data centers, how much cooling they need . And that's where Vertiv comes in, of course. Each GB300 rack uses 125 kW of energy, which is, you know, like two and a half average households use in an hour .
It's just crazy. And then they have thousands of these racks, right? But in any In that case, they need liquid cooling and chillers. Nvidia will get its money regardless of whether SpaceX contracts are renewed, but guess who else will get paid? Vertiv. And LSCC, right? So that's good for these guys .
Now that we really think about it, we're like Vertiv, thinking, "Okay, what are the margins for this company and how much pricing power do they really have ?" "They actually have quite a bit of pricing power. If you look at their margins here, you'll see that their operating margins are growing. We're getting close to 20% operating margin.
And gross margins are holding at about 37%. For a manufacturer with 18.8% growth, that's actually pretty good. The company has a forward EPS estimate of $6.68. The growth expectation over the next four years is about 23% per year. Maybe if you just take the next three years, you get about 28% average growth.
That's because there's actually a downside at the end , but because Elon is drunk selling or drunk buying here, selling, you know, maybe his financing, right? Or the available stock, which is terrible, but actually buying this equipment. We could potentially see their margins grow. And we could see a higher average growth rate.
So if we assume an average growth rate growth of about 28% over the next 3 years, this particular stock could go in a pretty wide range, and it really depends on that. It's a smaller company, so it all depends on the growth rate. We'll take 23, we'll take 28. The growth rate that we're going to use here is going to be 28.2%, at 28.2% with those margins growing.
We're looking at a fair value of potentially $390. So I think the range for Vertiv is probably $ 270 to $390. If I open up the chart of Vertiv now, you'll see that it's actually already in that 270 range. So the growth potential is a little bit limited on the more conservative end of that valuation.
Although there's definitely still potential on the more optimistic side. This stock is also more volatile, simply because we have less of a forward-looking view on it. Smaller companies, I mean, this is a hundred billion dollar market cap company, so It's not that small. But some companies just don't get as much attention from analysts as some of the giant corporations.
Let's look at some notes on Vertiv. What's interesting is that operating margins are growing. We saw that on the chart. That's good, and this quarter they accelerated. Here's what changed this quarter. This is again from the Meet Kevin app. The management of prefabricated systems and their integration demonstrates a strategic shift to integrated solutions that combine power, thermal and IT infrastructure.
This increases the share per facility, shortens development cycles and creates higher barriers to entry for competitors. Essentially, it structurally brings more people into the Vertiv ecosystem. Nvidia is already deeply integrated into them. So I would call it an Nvidia proxy. Management confirmed that the services margin is in the 30-35% range.
It's a smaller share of their revenue, but services is worth paying attention to, as is the 800 VDC architecture. That's what Jensen was talking about . I think it might be Vera Rubin. Vera Rubin, let me check, DC. The Vera Rubin architecture will include 800V DC power. That's what I thought. Remember, I said earlier that Elon would probably go to Vera Rubin as soon as they came out.
The point is, GB200 doesn't need 800V DC, but you can use it. So if someone is building an 800V DC data center, they can use GB200, and that's what they're going to need for the next generation of infrastructure—Vera Rubin. Anyway, it's good to know that Vertiv is part of this stack for the broader adoption of liquid cooling, not just at the chip level.
But data centers themselves also get really hot. Have you ever been in a room with a lot of these beasts ? It gets hot in there. But anyway, I like that line. The backlog has stretched to 12–18 months of supply. Here again, it's going to be a situation where Elon says, 'Look, I'll pay more.'" Let me go ahead .
"He doesn't care if the company loses money. It's like in the Economist interview where they ask, "When are you going to start making money?" "And he's just joking around and laughing. He's trying to get to Mars, man. He doesn't care about next quarter's EPS. Oh, man . These companies do that. So, um, well, let's see.
Self- powered and microgrid integration. The rise of behind- the-meter energy solutions is driving demand for Vertiv's integrated power and energy storage management technologies. By positioning itself as a systems orchestrator for microgrids and DC architectures, Vertiv is capturing the adjacent revenue streams that complement traditional data center infrastructure projects.
So, actually, there's a lot of talk about these future data center expansions trying to do everything they can to be independent of the grid. It's politically unpopular. It's unpopular because of the delays, mostly because the grid says, 'Hey, we can't get you.' 'You're just going to have to wait.' And data centers—it's hard, you know?
So a lot of these companies or developers are implementing not just Bloom Energy-style fuel cells, but also solar, wind , and everything they can. It's probably too early for nuclear, right? And of course, gas turbines are a big and interesting topic right now, and Boom, I know, is trying to take their supersonic jet engine designs and sell them as gas turbines.
I always think that's very interesting. But yeah, it's actually a very interesting point. This company has a growth forecast of only about 15.7%, which, frankly, may be quite low . Net margin is about 17%, and they expect to have $13.46 per share at the end of 2026. That puts us, you know, a little below a PEG ratio of 1 , to be honest.
So there's definitely upside potential for this stock from a valuation perspective. And we're also seeing margin expansion, which is fantastic , and that's also growing their cash flow. We have, you know, expansion geographies where they place their products. Of course, there are risks associated with relying on supplier partners.
Those are some of the risks. I've read all of that. So, actually, that's how we arrive at a price target of about $390. Of course, there are potential pricing advantages with Eaton, and that's exactly what you need to continue to see this momentum in Vertiv. The next company is called Eaton. Eaton is an interesting case because even though they're also an electrical manufacturer and a liquid cooling supplier, their data center orders are up 85% year over year.
They're leading the transition to 800 VDC, and their margins are expanding. You have the liquid cooling integration and the synergies, blah blah blah. Great. Lots of good stuff. There's one red flag here. First of all, we have 4 days to cover, increased short squeeze risk, okay. But I want you to You saw this.
Balance sheet strength is a complete mess. Why is it 0.06? Let me show you. I'm going to use the annual report. This is the last quarter, so I'm actually using an older report. But I want you to see a year-over-year picture of what's going on here. So first of all, look at their current debt. Their current debt, mind you, none of it is deferred.
It's literally the current portion of long-term debt, accounts payable , other current liabilities, accrued compensation. All of that is net debt. $9.3 billion that's up over the last quarter. You see that difference, right? It's getting worse. On top of that, look at this. Long-term debt, pension liabilities, post- retirement benefits.
Here we have the first deferred income taxes, but that's a drop in the ocean compared to the rest of the current liabilities. Dude , I have, you know what? $ 9 billion to $11 billion in current debt, plus another $12 billion in long-term debt. And if I look at the cash, man, where's the cash? I have like $ 700 million, maybe $800 million in cash.
The cash has exploded. That's not cash, it's gone. When I say " exploded," it usually means growth. No, in this case it just disappeared. Got it? Yeah, they have accounts receivable and inventory, but man, they're barely making ends meet . So if I get to the financing, they still have to pay dividends.
For some reason, they're buying back shares at all-time highs. That's also amazing to me. And so, of course, they have to borrow more than they refinance because they have limited cash. So there are some issues with Eaton, even if they can make a good profit on orders from SpaceX, we'll see. There are some red flags here.
You have $13.46 of earnings per share. So if I divide that by the 448 price that we have right now, I'm looking at a P/E ratio of 33 on a projected growth of 15.73. That gives me a PEG ratio of about 2.1 already at 17% margin. Man, there's not much room for growth left. And so sometimes I, you know, don't want to just mention companies that you're like, "Oh yeah, this is really exciting." I can be excited about it, but much less than others, you know ?
I mean, the growth potential is much lower. Maybe their growth will really surprise, but as you can see, we have a red flag in the balance sheet and the moving average, which is pretty obvious. So when you look at the chart itself, and their margins, even though we talked about a 90 basis point change in the quarter, we still see this curvature that is not ideal for this particular games.
Next up—Schneider Electric. It's a French company. A lot of people are excited about them because of the data centers, but they're only about 20% of their revenue. It's an ADR, an over-the-counter stock. ADR, an American depositary receipt , again, a French company. So it's a little bit more difficult to value because they're forecasting $11.76 in earnings per share, but you have to divide that by five to get the earnings per share for the ADR.
So that's $2.35. Okay, 70 divided by $2.35. That gives me about 30 PE with about 14-16% upside. I already have a PEG ratio of about two on this thing. They're only about 13-14% net margin, so the fair price for it, as much as I hate to admit it, is probably lower at only $58. So, I'm not that excited about it from a momentum perspective, and I'm not excited about valuation.
So I'm not going to spend a lot of time on that company. Let's move on to Dell Technologies instead. Dell Technologies is interesting because there's a huge potential risk that Wells Fargo just reported: SpaceX has placed an order for GB300 servers with Foxconn. They may literally bypass Dell and go directly.
It's not guaranteed, but it's a big deal because the reason Dell stock has skyrocketed so much is because of its AI server backlog. The AI server margin is still expected to be in the mid-single digits. They don't disclose the AI server margin clearly because, frankly, it's not that high, even though they're creating a lot of value.
And that's why people are really excited about Dell stock, and I understand them. It's a low-margin business, but it's a high-volume business . That company is also at a relatively high level. If you look at Dell here, we're at, yeah, a relative high . It's at a high. 453 $453. If I divide that by their projected EPS, I get a PEG ratio of 24.5.
And we expect to grow maybe 19% on average over the next 4 years. That gives you about a 1.26 PEG. But their margin is about 8%. So it's hard to justify a figure that's much higher than the current PEG ratio. Now, I want to be honest and warn you that this is a beta product. We're still working on refining it.
That's not fair because the baseline fair value estimate for Dell is only assuming 10% growth. That's revised down to about 19%. So the realistic future value is closer to where they are now. So again , the fundamental growth potential is lower. I just want to let you know that this is a bug in our platform.
And I want to make that clear . So, I hope that's helpful. Okay. Now, before we move on to the next stock, let's do a quick Let's take a break and talk about how much we love the free cash flow metric. Software companies are really good at generating free cash flow. Like Service Now or Salesforce. Salesforce has about a 13% free cash flow rate.
That's so good and so incredibly attractive right now. But there's actually one hardware company that has almost the same level of free cash flow rate as some software companies. That's crazy, and it's the next company on our list. Look at this. This is a free cash flow rate chart. It's basically free cash flow divided by market cap.
So if your market cap is a hundred billion dollars and you're generating five billion dollars of cash flow, that's a five percent rate of return, right? Simple math. Look at this. Nvidia is about two percent. Marvell is like Palantir, about one percent. Eaton is a little over two percent. Dell—a little over two percent.
Vertiv— about two and a half percent, right? Okay, cool. Where's Enphase? What? Almost seven percent free cash flow yield. And folks, that's probably one of the few things I like about Enphase. Well , maybe not the only thing, but one of the few . Mainly because in their last earnings call they talked about some kind of delay with their IQ intelligent solid-state transformers.
That's probably a 2028 data center project . It's going to take time. They're doing tenders right now and they're saying they're opening up a lot of things. So unfortunately it's going to take time for that to become a real data center driver and be priced in by the markets. But maybe it'll be priced in by 2027.
It's unlikely to go up until they start selling or until interest rates come down. Neither of those things look like a near-term prospect. So Enphase—it's not a short-term investment, but it's cheap. It's worth looking at . It's a stock with a projected EPS of $2 by the end of the year . The stock is currently trading at around $40, or around $41, if you look at it.
If we take Enphase, which has been in a slump since COVID due to a lack of interest in solar, and we look at a price of around $42, I'll divide $42 by $2.04 of projected EPS. That's a price/ earnings multiple of 20.5. They expect growth to be 18.6%. That gives a PEG ratio of around 1.1. With a net income of 27%, the stock probably deserves a price of around $70.
The upside is the interest rate cut, the free cash flow, and the data center option. That's my take on Enphase. So, it's an undervalued position. Of course, the margins there are not bad. If look at the operating margin, it's down a bit. Their actual operating margin is projected to almost double. It's expected to accelerate as they finally bottomed out on solar revenue, which has been falling precipitously since COVID.
So this company has its risks. I think the fair value of this stock is about $77, plus the option to lower rates and the data center outlook in 2028. It's worth remembering that it's not a favorite right now. It's like a stepchild in the family, if you will. I have nothing against stepchildren. Okay.
The next company to look at, and the last one, is ARM. ARM is very interesting because companies like Nvidia pay royalties on processor cores, not GPUs. They don't pay for GPUs. They pay for CPU cores. We're seeing more cores, which is driving ARM's revenue from licensing. You have to understand that ARM has very tight China ties in how related party revenue is reported.
They don't actually have a dual listing. They're just listed here, and then you get related party revenue that's just reported on the China side, which is a little bit weird because it can cause wild swings in reported revenue. Weird. They're also now into hardware manufacturing, which I think is a big risk to the company's cash flow.
That's one of the reasons they're so low on the cash flow chart. However, OpenAI and Meta are buying their specialty silicon, and volumes are expected to ramp up in 2027. Their AGI CPU revenue is expected to come in by fiscal 2031. But I think specialty silicon is an expensive play because of the capital expenditure required to get there.
That company is projected to have EPS of about $2.23 by March 2027. And the stock right now, yeah, let's let's see. And then we can calculate the P/E ratio. The stock is now, oh, $282. $282 divided by $2.23 in EPS gives me a multiple of 126. I'm projecting this company to grow 47.4% over the next 4 years, but that still gives me a PEG ratio of 2.67.
That essentially brings me to the valuation of the company that you're already targeting. Plus, you have the capital expenditure risks. So I like Enphase's cash flow yield , but honestly, it's almost like the value play of them all because they're in a declining industry. It could benefit from data centers as an option in the future, as well as lower interest rates.
And that makes it something of a value play that's just generating cash right now. The big favorites from Elon Musk's buying spree, in my opinion, are definitely Nvidia, Marvell, and Lattice Semiconductor in descending order. The others are also good, but you can see my analysis of them. Nvidia is just incredibly cheap.
Of all of them, Nvidia is just incredibly cheap. Of course, this video is not personalized financial advice, but Nvidia is just incredibly cheap. Anyway, I hope you enjoyed this review, have us here for? . Meet .
Comments 0