Transcription
This is a complete game-changer happening in artificial intelligence, and it's bad for Nvidia stock. Now, that's going to make me sound like an Nvidia bear. I've been very bullish on Nvidia. Uh, I own Nvidia shares. I'm not sure, like, I just want to be transparent about this. Uh, this has got me a little nervous.
Uh, but there's a specific date you want to write down. Uh, it's in November. We'll go through the exact date and why that date matters, but it could be a game-changing catalyst for what's going on in the chip industry. And it's really weird because it's totally the opposite of what we're hearing right now from institutions. Like, for example, Oracle is blowing money, taking on debt to blow money, and issuing stock to buy as many Nvidia chips as they can. Remember, just last week, they announced that they're buying $40 billion of Nvidia chips. Mind you, they're planning on spending $50, or sorry, $70 billion in capex, uh, you know, over the next basically 24 months, including the last about six months. So, sort of the past year and the next year, plus or minus $70 billion of capex spend. This is insane that Nvidia basically makes up nearly 60% of that, and then everything else, the racks, the buildings, the infrastructure, the CPUs, that's the other 41%. So, Nvidia is a huge beneficiary of this insanity for this AI boom in capex, and it's really driving GDP. I actually think it's probably the biggest contributor to GDP right now. Uh, and Goldman Sachs, based on their AI conference, says the same thing. And all this is bullish, right?
So, what's bearish? Trust me, I'm about to say what's really bearish, and it's scary. It's like it's making me nervous, and I'm more nervous that nobody's really talking about this. That's probably what I'm most nervous about. But anyway, Goldman Sachs, they're like, "Companies deploying capital behind AI buildout or buildout of AI remain capacity constrained." So, in other words, we still don't have enough chips to keep building out all these these products. But the problem is, this is potentially a debt-fueled capex expansion right now because companies like Oracle are like, "Oh, wow. Our investors love it when we borrow money or issue stock to buy more Nvidia chips." Then Nvidia goes up. Uh, then private valuations of OpenAI go up, the valuation of Oracle goes up, everybody gets rich. And this makes sense. And so a lot of people are like, "Oh, you know, why be bearish on this? You know, this, this is, this all sounds good, right?" That's what I thought as well.
And then I saw this, which I misread when I first read it this morning. I saw this headline: "Coreweave says Nvidia cloud contract valued at $6.3 billion." All right. Initially, it's really easy to brush this off. It's like, "Oh, cool. Okay. Coreweave is going to buy more chips from Nvidia because we see these headlines all the time. At this point, they're just dime a dozen." But what they actually say is, "Coreweave said its shareholder Nvidia will buy cloud services with an initial value at $6.3 billion." In a deal announced in April of 2023 that runs through 2032, Nvidia is required to buy any excess capacity not used by its customers, Coreweave said in a regulatory filing on Monday.
Okay, that's a lot. So, let's digest this for a moment. Okay, I really hate this. Let me explain this and digest this. This latest deal requires that Nvidia start buying unused data center capacity. Well, wait a minute. In order for Nvidia to start buying unused data center capacity, it means there's unused data center capacity. But wait a minute. If there's unused data center capacity, why is Oracle, Coreweave, NBIS, Iran, why are all of these companies, these data centers, throwing more money at building out more capacity? If all of a sudden, we're getting our first canary in the coal mine that we actually have too much capacity, right? I mean, think, think about it this way. Let's come up with an analogy. Let's say you have an ice cream stand and you're selling ice cream, and you're on the phone going, "Guys, guys, more, more ice cream, more ice cream. Keep ordering more ice cream. We, we need more shipments. Back up the trucks. We need more trucks. More trucks, more trucks." And you're standing at the ice cream stand, and your employee who's actually selling it to customers is looking at the ice cream bin and you're like, "Bro, we have so much ice cream. Some of this down here at the bottom, it's starting to go stale. Like, it's getting all frost-burned because the ice cream's going stale. And I'm trying to sell the frost-burned ice cream, but I can't sell it. And then I like the guy at the front desk is like, 'Hey, Nvidia, I need you to buy some of this frost-burned ice cream 'cause, like, we have too much of it. Get rid of the old crap at the bottom.'" All right, I guess fill it back up again. We, we really don't need to fill it up again. Meanwhile, you've got the guy in the back. This is good for the stock price of the ice cream stand. Keep the trucks coming. Keep them coming. Hey bankers, more loans. More loans. More debt. Hey attorneys, another filing. Sell some more stock. Okay, stock markets at all-time highs. Take advantage of it. Sell some more stock. Buy more chips. More debt. Let's go. Back up the truck. You're selling, right? Yeah. Yeah. Yeah. Yeah. Yeah. We're selling the ice cream. Frost-burned ice cream. Going to Nvidia. This is not good. This is like an early canary in the coal mine. And trust me, I do not want to be bearish here. I, I want my Nvidia shares to go to the moon because there are these leftover shares that I actually have a lot of, thousands of, many thousands of, and I'm like, "Oh, this is great." Uh, and, uh, like, so, so I want to be bullish here, but part of me now is like, "Oh, sh*t, right?" Like, this sucks.
So, the biggest concern is, you can't sell these data center products because how the cycle plays out, right? This is the cycle. It always plays out this way. The cycle is really easy. You end up with excess supply, then you get lower utilization, then prices go down, then there's a lower return on your investment for data centers. Then capex goes down. Then you have lower sales at Nvidia and Super Micro Computer and TSM. Then stock prices go down. Then people lose their jobs. And all of that falls onto an already weak economy. That's not good. It all starts with the frost-burned ice cream, the excess supply. It's always the cycle. It's always excess supply. Have you ever seen Kevin? Meet Kevin's real estate chart. Meet Kevin real estate cycle. Let me see if I could just Google it really quick because there's an old. Ah, yeah, there it is. Look at this. Uh, so this is an old, old thingy here that I made this back in like 2010. Okay, so I, I drew this cycle back in 2010. Let me put it to the desktop here. I, I wish I had a higher resolution of it, honestly, but it's, it's such an old chart from, uh, you know, when I was like getting into telling people to go buy real estate. Yeah, it's super blurry. Uh, I was telling people that we're probably here in 2010, and you should go buy real estate. But it always starts. I mean, look at this cycle. You could really call it the AI cycle instead of the real estate cycle. Look at this, uh, absorption of excess supply. Oh my gosh. The GPT moment. We're out of chips. Oh, we don't have enough chips. Prices skyrocket. Margins skyrocket at Nvidia. Then it says accelerated new construction. Build, build, build. Back up the truck. Then you get an oversupply of those data centers. That leads to low utilization. In real estate, we call that high vacancies. Then prices go down. The real estate cycle could may as well just be the business cycle or the AI cycle. So now we're like, we're here. We're at this accelerated new construction phase. Literally, the real estate cycle that I talk about in my courses probably applies to the AI cycle.
Now, like, if you're not already part of the Meet Kevin membership, you really should be. Some quick calls that I want to point out just from literally the last few days. One of them, well, first of all, look at this. Okay, Figure is up like to $40, and it's skyrocketing right now. It's up like 19% or something like that. And I wrote last week, this is a $45 to $55 stock. Uh, Coreweave, we talked about if we don't confirm $12.49, don't touch it. And what happens? We don't confirm $12.49. We reject. We literally reject off of $12.49. Don't trade that up. Then in the Meet Kevin membership, we talk about the most bullish thing for Tesla is popping onto the $414 line, which we bounce on twice, which is bullish. And I said, as long as the Q's or as long as Tesla doesn't dump it, open the trading strategy for today is the Q's slowly continue to slog higher. The next target is $590 on the Q's. Now, why do I mention that from the Meet Kevin membership? Well, because I think there's, first of all, obviously a lot of value in the Meet Kevin membership. But if you look at the Q's, what did the Q's do today? The Q's literally went straight up to $590. Well, what happened with Tesla stock today? Well, we know what happened with Tesla stock. It literally went down to $414 twice. It didn't dump open beyond our line. Therefore, the Q's were able to go to $590. And we had a double bounce at $414, which we also called that it would likely go towards. These are the things that you get if you're a member in the course, member live stream. So join us at the Meet Kevin, um, the Meet Kevin membership over at meetkevin.com.
But let's keep going with this chip analysis because this chip analysis is a problem. When we actually understand that this is a cycle that ends at some point, we start getting a little nervous because this partnership with Coreweave is a big red flag in my opinion. Look at this. Nvidia was willing to enter what are called MSAs in 2023, and they probably entered these MSAs with other companies to get as many sales as possible. Uh, MSAs are basically these contracts, material service agreements that guarantee that, or I'm sorry, they call them master service agreements that guarantee that Nvidia will buy excess data center capacity. Well, why would Nvidia need to buy excess data center capacity unless there were excess data center capacity, right? So, Nvidia now buying, uh, data center capacity means Coreweave can't sell all the data capacity. Last week, Nvidia got, or, uh, exited, uh, the data center industry. What does that mean? That means Nvidia no longer wants to build out data centers to license the actual data centers to you because they're already having to buy the excess capacity from Coreweave because it can't be sold.
Now, the problem is, the next earnings report is when we'll actually get a copy of the material services or the master services agreement. We won't get that until November 14th. We literally have to wait two damn months to get this MSA. This MSA is going to show that Nvidia has to buy these damn chips, or, uh, the output from these chips through April 13th, 2032. This is so smart by Coreweave that they did this. And if Coreweave pulled it off, because I don't necessarily think they're particularly brilliant, but if they're able to pull it off, then other companies probably pulled this off as well. It basically means that Nvidia went out there whoring themselves out going, "Buy our chips, take on debt, and dilute your stocks. Buy our chips. We'll guarantee they make money for you." Well, that's like the perfect way to set up a bubble. I mean, look at this. This is Coreweave. When they IPOed, they raised $1.4 billion by IPOing, by issuing shares, and then they took on another $3 billion in debt, all to go buy more Nvidia stocks. So, of course, Nvidia wants desperately their companies to go, uh, buy more Nvidia stocks. So, whatever they can do to prop up and enable these companies to go, great. So, if they can dance around like, you know, the naked heavenly maiden and, and they can go, "Hey, you know, let's go buy more of our chips, we'll guarantee that they're functional." Great.
But what is Jensen Huang? There's a lot of people like, "Oh, Kevin, Jensen Huang's really smart, you know, trust what he's doing." Yeah, I will trust what he's doing. Jensen Huang sells $13 million worth of shares. July 19th, Jensen sells 35 million, sorry, 37 million shares. A week earlier, Jensen is able and has filed to sell as many as 6 million shares, which is worth a billion dollars. That's a lot. That's a lot of selling. It's the opposite of what Elon Musk is doing, who just bought a billion dollars of Tesla shares that he made that back like instantly by pumping the stock. So, you know, does he really, uh, does it really matter? No. Uh, but, you know, it's interesting. It's, it's the opposite that's going on.
So now when we look at this material services agreement that comes out November 14th, what we're going to want to look at is the, the big questions now are, what price is, uh, is Nvidia buying, uh, data center capacity for? Are they, are they paying, uh, 2023 prices? Because prices are going to fall. So, if they're stuck paying 2023 prices, they're going to pay over market value. Or are they going to pay a market value? We don't know. We don't know the details. How much spare data center capacity is there? Are we talking about, you know, a day's worth of extra capacity that Coreweave is making Nvidia buy? I don't know. But this really changes the argument because I've always said that in the short term, Coreweave, NBBY, I ran, Oracle, great short-term place, but they're long-term bagholders. Basically, inside the bag are Nvidia chips. The bag itself are like the server racks, like Super Micro Computer. Super Micro is able to sell the chips. TSM is able to sell the chips. Nvidia is able to sell the chips. They all wash their hands of the chips, and somebody else holds the bag. Holds the bag are the data center plays. Coreweave, Oracle, NBIS, INEN. So, short-term good for them, longer-term bad for them. But this literally flips all of that on its head. It literally lets Coreweave take the bag and dump it back onto Nvidia. It's almost like a repurchase agreement.
So, I hate this for Nvidia. Nvidia, it, it now sharing in this data center capacity marks an inflection point in artificial intelligence. It's a red flag now. So far, nobody's really talking about this. I saw a Doomberg reaction that said, "Oh, this is good. It could concern e, it could ease concerns about Coreweave's overbuilding because there's a long-term risk of overdeveloping AI infrastructure." Yeah. No. Duh. Like, of course there's a long-term risk of overdeveloping AI infrastructure. That's the problem. But the fact that even the suits aren't realizing that this is bad for Nvidia, right? Because if there's over capacity, if Coreweave is like, "Man, we've got over capacity. Hey, Nvidia, start buying back." At some point, Nvidia is going to go, "Hey, ma'am, we'll sell you chips, but we're not going to keep guaranteeing we're going to buy back all your capacity." Because if they're buying back over capacity and they're getting ripped off, Nvidia is going to stop doing it. That tanks Nvidia chip sales.
Now, again, maybe this is just a really, really early canary in the coal mine. But I thought this person, Tony, this Tony guy's comment was right on. "Is it just me, or is all of this Nvidia, Coreweave, OpenAI, etc. stuff just a circle jerk of companies buying from each other at inflated prices to push up each other's valuations and then rinse and repeat? I'm not the smartest guy around, but this all seems very predictable and tiresome." I think he's got a really good point, and this MSA is really scary. Now, we'll see what price Nvidia has to pay November 15th. It's scary to me that we have to wait that damn long to get it. You know, somebody else in the chat commented this. A shout out to Brandon in the chat. Let's go, Brandon. Uh, he's never heard that one before. Uh, Brandon in the chat said, "At his company, the first thing they strike from their contracts are buyback agreements or arrangements." This is basically, you know, what this is is you're buying back the output of these chips. Uh, so now, to be clear, Nvidia is not buying back the chips, they're buying back the extra compute. This comes just as last week, Nvidia left the data center business, right? So, it's all sort of aligning. Like, if the data center business was so profitable, why would you leave it, Nvidia? You know, it's, it's a start of a cycle. It's the start of the canary in the coal mine. This is what the cycle looks like. Again, just like the real estate cycle. Excess supply, lower utilization, lower prices, lower ROI on data centers, lower capex, lower sales at NVIDIA, SMCI, TSM, lower stock prices, job loss, all fall onto an already weak economy. This is such a huge inflection point, and nobody's talking about it. So, either I'm wrong, or I'm just really early. And my wife already tells me I'm really good at being early. So, I'm probably correct. Just again, very early. Time for drilling stops. I'll tell you when I dump my. How about this? I'll make you a deal. When I dump my Nvidia shares, you'll be the first to know in the Meet Kevin membership. Uh, and, and let's just put it this way. I've got over 10,000 of them. So, uh, I will tell you when I dump them, you'll be the first to know in the Meet Kevin membership.
So, uh, to me, you know, then somebody I saw in the comments said, uh, you know, "Oh, this is good for Nvidia. They can control the data centers but not own it." And then I responded and said, "No, that's totally wrong. They don't control the data centers, and they don't get the upside other than, you know, more chip buying, but they are responsible for the downside." So, it's literally not control it, don't own it. It's don't control it and be responsible for the downside to squeeze out money in chips today. You're basically saying, "Hey, we could get three years of selling you chips, and if it hits the fan in the future, no problem. I already made my money." That's scary. You know, this is, this is the, like, this is the part where people wonder like, "Okay, when does it turn?" I mean, I don't think anybody's going to really pay attention to, uh, the Empire Manufacturing report from this morning. I don't think anybody's going to give a flying hoodie duty about this. You know, Empire Manufacturing, first negative read since June. Really. We're in a place where you, you have to look at. I mean, here's just the bottom line of it, okay? Uh, in, like, so inventories get used up. They wrote, "Inventories edged lower while the same kind of companies were buying less orders and shipments fell sharply." Okay, that is literally written right here. "Inventories edged lower and new orders and shipments fell sharply." What does that mean for the market? Well, it's bullish in the short term, A, because nobody's going to care about this report. It's bearish for the long term because it's a bad report. And then for employment, they say that employment came in at zero, but productivity went down. The average work week went down 5.1 hours. That is bullish short-term jobs report, bearish long-term.
So, uh, yeah, I don't know, man. All this makes me nervous. So, you know, if you want to play the short term, you need to keep, you need to do the following on core. Okay. So, some, some technicals. You can't play the core upside until you break out of $125. That's my opinion. Tesla needs to hold $414. Elon's pump probably won't be enough to hold it up alone. Okay, very important. Pay attention to that. In addition, uh, watch what's happening with Nvidia, especially that MSA. Okay, that's a really, really, really big deal. So, watch that. Uh, and then, uh, uh, yeah, as far as my top 10 stocks to buy. Keep in mind, I also have a top 10 stocks list not to buy list that we started. We started that, I think, Thursday or Friday. So, we're developing that all. We, we include it now daily at the bottom of each alpha report, starting with this morning's alpha report, which did really well. Make sure you're part of it in the Meet Kevin membership. All right. So there you have what's going on in the chip sector. Good lord.
>> Why not advertise these things that you told us here? I feel like nobody else knows about this.
>> We'll, we'll try a little advertising and see how it goes. Congratulations, man. You have done so much. People love you. People look up to you.
>> Kevin Praath there, financial analyst and YouTuber. Meet Kevin. Always great to get your take.