Nvidia Can't Buy Enough of It. I Found 2 Ways In

I want to thank today's sponsor, Copper One Resources. In 2020, we all stood in a grocery store looking at a sign that said, "Two packages of bath tissue per household. You had money to buy more, but at the end of the day, it didn't matter." Now, I have been on the other side of that situation where I sat in a meeting and I was told from a supplier, "Hey, we cannot sell you that." Not because it was out of stock, because they did have it, but it already belonged to somebody else. Which leaves us with a very interesting question. And and there's no polite way to put it. How much would it take for somebody to break their promise? And I guarantee you Nvidia is asking that question right now, which seems a little bit strange for them to be doing because on paper, Nvidia just had a fantastic quarter. The most profitable chip company in history should not be the one doing the asking. When those earnings came out, everyone started to talk about the same number. The gross margin came in at only 75% and it gets guided down to somewhere around 72 to 73%. So the story here kind of writes itself. So even if they are getting less profitable, then that means there's somebody out there who must be taking that money. Except here's the thing. Two years ago, this business ran at 72.7% margin. And nobody called that a collapse at that time. I mean, just the opposite. They called it the best margin in semiconductor history. So the fall everybody is describing lands right about where Nvidia lived before this big boom kicked in. But once again, somebody is still taking that money. A margin is only what is left after you pay all of your expenses. So I went ahead and I looked at what Nvidia pays. Their cost of goods nearly doubled in a single year to $63 billion. Costs just happened to grow faster than the sales did. And that gap is the reason why the margin started to fall. And here's the interesting part. Nvidia actually tells you where it's going. There's a line in the commitments footnote called supply and capacity. A quarter ago, it read 119 billion. Now it reads $279 billion. That happens to be more money in one footnote than Nvidia sold in product all of last year. And the filing says exactly what it is for. Data center infrastructure systems primarily memory and manufacturing facilities. Back in May, Micron and SKH Highix, two memory chip companies, produced 17% of the entire global stock market's return for the month. Not 17% of the chip sector, 17% of everything, every listed company in every country added together. So when you go back to the month of May, the whole stock market went up about 5%. And a fifth of that came from two memory companies. So, if you own an index fund, which I I genuinely hope that all of you do, that was your performance for the month of May. And it all comes down to two companies that may or may not have even been on your radar. And I guarantee you almost everyone who's talking about this topic of scarcity is looking completely somewhere else. And it happens to be that most everyone is talking about electricity. And I completely agree with them because when you go back to July, Elon Musk gave the economist the cleanest version of that argument that anyone's ever really given. The the constraint right now, I think, is actually slightly more on power and cooling than it is on AI chips because the power demands of the AI chips are very, very high. That's him in real time pretty much ranking them out for us. And he puts power and cooling first with chips, kind of a slow second. And and here's the thing that I couldn't stop thinking about. The video is 14 minutes long and and it is careful and it's very well researched and it never once says the one magic word of memory. But here's how they do sum up the whole thing. You can think of like the the constraints on AI are essentially electricity and AI chips. >> And really, it's just those two inputs, power and chips, >> two inputs. Now, go back to Nvidia's footnote. $279 billion committed. And the reason that they gave was the procurement of memory and infrastructure. But I want to point out they do pretty much everything with purpose. They led with the word memory. Granted, at that time, memory had been having a a pretty severe pullback, but yet it was the first constraint that was listed on Nvidia's invoice. So, how exactly do we tell which one is actually the constraint? Well, there's a book called the theory of constraints that that I highly recommend that you read. But I truly learned about the constraints like this type of one on the job, which of course I'm sure many of you have. I'd spent years as a merchant working for Target and also Amazon. and the job is pretty much the same one every year. You you sit across from a supplier and you work out what you're willing to pay them for their product. In most of those meetings, I would tend to have a little more leverage or have an upper hand. It wasn't because I was smarter than them or anything like that. It was because if that supplier wasn't willing to move on their price, then I'd typically have three or four other suppliers who would step right into their place. Then of course there's video games and electronics where I'd sit across from the suppliers and I'd be asking for more Nintendos or more Xboxes. And it didn't matter how much money I tried to put on the table because you simply got the allocation that they gave you. At the end of the day, they couldn't build more of them overnight. There there was a factory somewhere running flat out all the time. And that was the number. But the allocation is only half of the equation because not only did they decide how many that I got, they also decided at what price I was able to sell them at for the rest of you. So just think about that the next time that you shop for electronics. Every store you go into, they all have the same price. Like that's not exactly coincidence and it's not exactly fair competition either. That happens to be the supplier telling every retailer in the country what the sale price is going to be. And if you sold it for a dollar more or a dollar less, guess what? They stopped shipping you product. That to me is the definition of constraint. Whoever owns it decides how much you're going to get and what you can charge all the way down to the very shelf that it sits on. And you don't have to be sitting in the same room with a supplier to find out who it is because scarcity always shows up in the exact same place. Gross margins going up. Revenue goes up for everybody that's in a boom. So revenue really doesn't tell you anything. Margin only moves when you have the leverage to keep that extra money. Let me try to illustrate that a little differently. In a restaurant, you can hire more cooks, you can buy more food, and you can add more tables. But with only one oven, it serves exactly as many dinners as that one oven can cook. So take any company in this chain and look at its gross margin over the last four quarters against the four prior. If it went up, then chances are that they own something scarce. Now, if it went down while revenue exploded, well, in that case, they're probably just really busy. One fair warning before we run this. I ran this already against 24 companies across the whole chain. Memory and storage, the gear inside the building, the the people generating the power and the crews doing all of those installs. 13 of them had passed. And every single one of the 24 happens to be making more gross profit than it was 2 years ago. So, nobody here is doing badly. Everyone, for the most part, is winning. It's just a matter of who's winning more. However, my question is a little bit more narrow than that. When the bill lands, who actually keeps the markup or or the margin? One little housekeeping note before we get started. Every one of these names has a position on my valuation ladder, and I'm putting that on the screen instead of saying it out loud because a couple of them are moving between the bands week to week or in some cases dayto day. And and for me, I'd much rather you just see the version from the day this video goes up than the version from the day that I actually film this. And if you're curious, I am sharing my full ladder to my Patreon members that, hey, I'm not a financial adviser and I do this for educational purposes. And of course, if you get any value from my content that I'm providing, then please consider pressing the like button. And I'd love it if you'd also consider subscribing. Let me go ahead and get started with the answer that almost everybody's already giving. Somebody has to build the power and the cooling gear that sits inside the data center. And the name that everybody lands on is Verdive. Ross Given put $50,000 into it on camera just a week ago. And I want to be very fair to him because his framing was actually very good in it. He said, "Don't buy the bottleneck. Buy the bottleneck that your customer can't engineer around." And I agree. That's exactly right. And it is exactly what I'm trying to do here. On Thursday, the morning after Nvidia's earnings report, Veritative gap straight up almost 4%. Somebody connected it immediately and then by noon, the stock had given it all back and then some. Think about what that tells you. The market looked directly at this, made the connection for about an hour, and it went right back to arguing about memory chips. And he noticed the exact same thing that I did. He reads that hour as the market getting a little distracted. I on the other hand I I read it as the market actually getting it right because the thing that it went back to arguing about was memory and memory is the word in Nvidia's very own footnote. Now I am the first to admit that I could absolutely be wrong. He has been doing this for a very long time and he put real money behind his version but we can both check the same number. So let me go ahead and run it. Now to me the demand is not the question either because customers are paying Verive months before anything actually ships where that shows up as deferred revenue and it went from 1.8 billion to 3.6 billion in a matter of 6 months it doubled. The increase alone is bigger than everything Verdiv had earned in profit all of last year. Inventory is up 73% over the same stretch which is hardware being staged against those prepayments. And nobody, and I mean nobody, prepays a supplier that they think they could replace. Now, we'll run the test. Last four quarters against the four before. Verdives gross margin is up 4.3 points on revenue up 26%. So, in this case, it passes. And it passes honestly, but here is the rub for me. It's only four points against customers who are prepaying them billions of dollars. That happens to be a long way from just owning a bottleneck. It says to me that for all of that prepaying, those customers probably still have a second phone number that they can call when they want to. Every company in this video gets paid because it owns a bottleneck. And underneath all of them is Copper. That brings us to our sponsor where this segment is disseminated on behalf of Copper One Resources Corp. Where copper hit an all-time high back in August, above $6.70 a pound on the comics, up over 40% in just 12 months. Billionaire Robert Freriedeland says there is no rational price for something that you absolutely must have. And Stanley Ducken Miller called copper the tightest position he has ever studied. So with copper at record prices, copper 1 is down roughly 75% in 2026. And when the Iran war started in late February, junior miners began selling off hard. And copper 1 was one of the few raising money right through all of it. And that delilution, well, that's why they hold the largest cash position and working capital in their history. And it goes without saying that cash matters. Their market cap is about 14.8 million in Canadian. And 10.4 million of that happens to be cash. If you go ahead and back out the cash and the market is pricing three copper projects at about 4 million Canadian, Majuba Hill in Nevada, a past producer, plus Redd and Red Hill in British Columbia. And Red is drilling right now. So, we're sitting with record copper prices, three projects, and cash that's in the bank in what is still a very early stage explorer with real risk. Please do your own due diligence and check out the link down in the description on Copper One Resources Corp. Moving right along, the next company should be the easiest win in this entire video. It makes the electrical connection hardware and the liquid cooling, which is literally the list that Musk just read out to us. Its data center business is targeting $2 billion in 2026, more than double what that business sold the year before. Two years ago, the entire company, everything it sells to everybody was just $3 billion. So, the tough item to call out is why is their gross margin going the wrong way? And that company happens to be Invent Electric. And on the same four quarter test, its margin is down 2.2 points while revenue grew 46%. Let's go ahead and just sit with that for a second. Demand is real. Revenue is exploding and they're getting less profitable on every dollar of it. Usually, that only happens when the customer has somewhere else that they can go. A bus bar and a cooling loop are things that a competent factory can probably build. And there are a lot of competent factories. So, this looks to be managed more like a commodity and the margin happens to be flowing to the buyers. And for this company, my ladder lands it in the same place that it lands with Verdive. And that happens to be the whole problem with owning electricity. Anyone with a factory can build the equipment. The scarcity happens to be with the queue and the permits and the years of waiting behind both of those. But let's be straight about it. The shortage is completely real. It's also almost impossible to own. So, forget the building aspect for just a second and go back to the word that led in that footnote. So, let's go ahead and move on to the next company that got spun out of Western Digital about 18 months ago. And all it does is flash memory. This is not the memory that sits next to the Nvidia chip because this is not high bandwidth. This is the memory that stores everything that those chips produce. And and every one of those data centers, they need an ocean of it. So, just a single year ago, it kept 26 cents of gross profit on every dollar that it sold. Last quarter, it kept 85. That company is SanDisk. And on the four quarter test, it's up 41.4 points, which is more than Micron. And and it did that while revenue, it nearly tripled. And the part that kind of surprised me a little is Nvidia does not buy from SanDisk. There's no contract between them. There's no line in that footnote pointing here. That simple little test found the company on its own in a very different corner of memory with a different customer list and it found the exact same fingerprint which means the scarcity is not just one company's goody year. It's the entire category. So all of that lands on memory and it raises the question this whole test was built to answer. What does it actually look like when a business owns the thing that nobody can get? Start with Micron and I want to give you two facts side by side. The first is the test. Take the same quarter, one year apart. Micron's gross margin went from 37.7% to 84.6%. That's not a good quarter against a bad one. It's the same 3 months of the year, 12 months later. And the number more than doubled on the full four quarter test. It's up 35.4 points where 85 cents of every dollar of memory they sell is completely gross profit. I'm sorry, but a company running physical fabs is out earning Microsoft on the margin line, and Microsoft sells software, which costs next to nothing to just simply copy. I'll tell you right now, nobody else in this video is even close. And it matches the only hard evidence that we have about where the money went, which is Nvidia committing 279 billion and saying primarily its memory. In this instance, Nvidia is a company with really nowhere else to go. Now, there's a few options, but not many. and the capacity is really already spoken for. Micron said so themselves on their own earnings call. They have signed 16 long-term supply agreements running out to 2030, covering about a quarter of their revenue at prices locked in already in advance. And on the demand that they can't cover, their own CEO said this. Some of our key customers were only able to fulfill only 50% to 2/3 of their demand in the medium term. and we currently do not have line of sight as to when memory supply will be able to catch up with increasing demand. Their biggest customers are asking for memory and Micron tells them exactly how much of it they're actually approved to get. The second fact is the price and this is the part where I have to correct myself on camera because I had this completely wrong the first time that I ran it. Micron trades at around six times next year's earnings. And I want to be precise about why that number is a six and not a 13 because it does matter. Micron's fiscal year just ended. And they do not report it until the end of September. So the low teens you're going to see quoted in a lot of places is the multiple on the year that just closed. But the year ahead is a six. That distinction is worth half the multiple. run the discount math on it and the price only works if Micron's earnings settle around 60% below what analysts expect next year and they stay there. Overall, I don't think that's happening. So, the market is looking at a company running an 84% gross margin, nearly four times what it earned just 2 years ago. And right now, they're just refusing to pay for it. That is not the market missing something. That is the market pricing. The last time memory margins looked like this. They peaked at 59% in 2018 and five years later, Micron was losing money on every dollar of memory that it sold. At six times earnings, the market is not calling Micron cheap. Instead, it seems to be calling at the very top. Now, let me be straight about the thing that does not separate these names. Verdive sits at the 94th percentile of its own valuation history. Invent the 96th, Micron at the 98th. Nobody here is on sale and my ladder knocks every one of them down for it. Micron gets knocked down twice. Once for the percentile and once because the range of what this company could be worth is absolutely enormous. And my system sizes down when it can't really tell you tightly what something is actually worth. Even after both of those, the gap between price and fair value is about a third. And here's where it gets a little bit uncomfortable for me because the name that my system is least willing to knock down, it's not exactly the famous one. SanDisk has only been public for 18 months. It has no valuation history to be expensive against. So, it is the one name here with the percentile rule. It can't really touch it. Its margin of safety is about the same as microns. And it's trading below my own bare case for it. And of course, I'm going to be honest about how this is cutting both ways. like no history means no percentile check. No percentile check means one of my guard rails simply isn't available here. To me, that's not a green light. It's a missing instrument. I'm not going to just hand you that and then just walk away. The obvious argument against it is the cycle. Memory has historically been cyclical, but there's a lot of people talking about the fact that maybe that has changed. And in this case, an 84.6% gross margin really doesn't have any precedent. The prior peak was around 59%. And at the bottom of that cycle, Micron didn't just sell it for less, they was selling it at a loss. Before I make this sound like easy money, here's what owning this actually felt like at the start of August. Micron fell from well over 1,200 to 820, down 32%. And the list goes on. Marll down 45. Western Digital and Lamb Research both down 38%. KLA down 37%. that is real and you should hold it. So, what would actually break this? I think the answer is China and it's already moving. A Chinese memory maker went public back in July and more than half of its equity is state money spread across enough government hands that no single one of them actually controls it. At least not that we can see. It raised about $ 8.6 billion in doing it. That was the largest offering anywhere in Asia in 2026. And here's the real kicker. Analysts who track this think that by the end of this year it's going to be starting roughly nine wafers for every 10 that Micron starts. Now wafers are not bits. Their wafers hold less memory and they are on older equipment which is why they're still under 8% of the money in this market. And Micron is over 20. But almost everything they make is ordinary memory which is exactly the product whose price built Micron's margin in the first place. And the interesting part is that Apple's been trying to test this for the China market. iPhones and Macs. But seven senators from both sets of parties wrote to Apple back in July and they asked for a written commitment. And the commerce secretary said the administration is currently not in favor. Now, the legal side of this is simpler than the noise around it. What we're all watching is political pressure doing the work of a law that doesn't quite yet exist. And of course, I want to share with you what I got wrong when I first started pulling this thread. My instinct was that Apple would be funding China's buildout. But in reality, it's not. That company reportedly turned down Apple's request for a discount because there's two other companies that already have its output completely spoken for. A supplier who desperately needs your money does not refuse your money. However, the thing that scares me is what Apple would supply is much worse than money. Like I said, state money already pays for the factories. The thing that state money can't manufacture is credibility. And Apple is probably one of the few companies on Earth whose supplier approval would literally be a stamp of approval for any company. So, here's where I land. Three of the four names that I just showed you passed the test, and the two that passed hardest are both memory. My latter is willing to let me start a position in either one. And it got there before I knew anything about the Chinese competitor or senator sending letters. And in my mind, that's the whole reason of having a system instead of just an opinion. And that same system is what tells me what breaks it. A statebacked competitor putting capacity on exactly the product that is short. And the interesting part again is the markets already rehearsed that trade repeatedly. And you can watch it happen. Micron fell 7% the day its lobbying fight with Apple became public. It fell another 9% and then another 10% in just 2 days after that company listed in Shanghai. It rose 4% the Monday Washington pushed back. Then the yo-yo comes right back again almost 6% on a report that the administration might actually allow it after all. So you're going to see this change before anybody actually announces it on a number that you now know how to check. Now I know some of you are already typing ready to send that comment. What about the turbines? What about the fuel cells? What about nuclear? So I tried to get ahead of you. So let me go ahead and put the whole test on the screen because I ran all of it. Quite honestly, the answers aren't even what I expected because as you walk right down the list, most every single one of them are very much down. I do want to point out that G Vernova does pass up 2.2 points, 20% gross margin. That's the profile of a company with a very long order book and very little say in what it can actually charge. And then there's the one that beat everything except memory. Constellation Energy is up 24.5 points. That is a real pass. And it is the honest exception to what I said earlier about electricity being impossible to own. Because it's true, you can't buy a permit and you can't buy a spot in the interconnection queue. But you can buy reactors that were built 40 years ago and are already licensed and already running. That isn't equipment. That is the scarce thing itself. And there's no factory anywhere that can just make more of it in this decade. So if you want the electricity trade, that is the shape to look for. It's something already built, already licensed, and already running. And let me be clear about the turbine makers and the fuel cell companies because they are going to do enormous business in this buildout. And I'm not waving any of that off. They happen to be selling something that a factory can simply make more of. The test that I just ran is about what happens when a factory can't. And and to be straight with you, I have not put constellation energy through my own valuation work. So that's a direction to go read more about and not a recommendation from me. Go ahead and pause this if you want to. The four biggest passes on this board are Constellation and three memory names and and the largest one is a company most people watching have never owned. Which means you don't have to take my word for any of this. Go run it yourself tonight. Pick any company that you already own in this chain. pull its gross margin for the last four quarters against the four quarters before and find out whether or not you own something scarce or just something that's really busy. As always, thank you so much for watching and I'll see you next