Google Just Confirmed the Biggest AI Trade Yet

Ladies and gentlemen, the chip names are about to head a lot higher. And Google just signed the release form saying it's okay to dilute in order to buy more chips from Nvidia, AMD, Broadcom, you name it. This is the AI infrastructure boom starting with Google. And we have many other topics that I want to cover in this video. But first, Google just raised $80 billion led by Birkshshire Hathway, which considered the top line of the top line of value investors signed off on potentially buying $10 billion of Google stock in secondary markets at an average price of $350 per share with the additional amount coming from potentially at the market offerings or other private raises. But this is because they're growing so much on the cloud side. They're now bringing in over $20 billion of revenue, growing at 63%, a lot higher than where they were growing just two years ago, roughly sitting at around 20 or maybe even 30% on average. But now that is more than doubled, if not tripled up to 63%. They're growing faster than their competitors. So this is market share taking. Although AWS is accelerating in their growth and so is Microsoft Azure, they are not growing as fast as Google. And so Google clearly has something here where customers are really wanting their compute and so they want to step on the gas because well it's profitable growth. The interesting thing about this $20 billion that they're bringing on this is also at the highest margins that they've ever put up. This step up in margin from essentially 3% up to 11% and then hovering around the 17% was great and that margin expansion is good on the amount of overall revenue that they continued to grow by. But the fact that we've gone from 17% up to 33% year-over-year while also bringing up the revenue from 12 billion to 20 billion is an unbelievable feat for Google stock. Google understands that the need for these artificial intelligence tools is essentially unlimited and customers are constantly paying more and more for compute. And so Google has found themselves in front of an infinite money machine. Metaphorically speaking, of course. But imagine if every time that you put a dollar in, you're getting a $130 back. How many dollars would you put into this machine? This machine being setting up data centers and having an unlimited wave of customers come and ask for these products at higher and higher prices. Would you go out and raise money? Would you use every last dollar that you potentially had? Well, that's exactly what Google's seeing here. And that's why they've cut off their buybacks. They might keep their dividend for now, but we could potentially see this come to an end also. And now we've actually started up well raising money. So now instead of their overall share count going down, it's going to start creeping up. Now this is on top of all the cash that they're generating from the business before capital expenditures roughly between 45 to 50 billion in cash per quarter, which should continue to grow because you're getting cash from Google search, you're getting cash from YouTube, you're getting cash from all these different areas of the business, but it's flowing all into Google Cloud. They're doing this because their backlog year-over-year has gone from 92 billion dollars in Google, the blue square here, up to 467 billion. But they're not the only player. The most interesting part for me being an Nvidia investor primarily is that you're seeing the backlog across the board go from 732 billion just across these four businesses up to $2 trillion. That's almost a 3x in the overall amount of token needs or compute need for Google, Amazon, Microsoft and Oracle. So whenever this news came out, this sparked a lot of excitement across the actual infrastructure layer. Now depending on how big you are, like Nvidia for example, the stock didn't move too much actually. In fact, by the time the market closed today, the stock was actually down. Yet Marll was up 28% at the time that I took the screenshot. By the time I'm recording, it was up like 32%. This is due to another piece of news, Nvidia, where Jensen Wong on stage specifically said that Marll will be the next $1 trillion company. Let's have a listen. >> So, what I just said, so we're talking about connectivity today. >> The next trillion dollar company, ladies and gentlemen, >> let's do it together. >> The reason why he said let's do it together is Nvidia is well an investor of Marll, right? So, they actually acquired shares in early April. I actually think it was March 31st to be exact. They bought in at $91.84 here, but now today it's over $280, making their $2 billion stake in the company already worth over $6 billion in just a few months. And this has been across the board in every single one of their private stock investments, where you've turned $13.6 billion of equity investments into $52 billion that goes towards Nvidia's stake. Everywhere we look, we're seeing a faster demand for customers wanting more clawed or chatbt tokens. But the truth is that we just can't get access to enough chips. These bottlenecks in compute is a massive benefit to companies like Nvidia, AMD, and Broadcom. Google is the greatest sign of this that we've seen in public markets raising. Even though they are one of the most cash flow positive businesses on planet Earth, that demand is not going to go completely just to Google. There's going to be multiple beneficiaries and I think ones that are more peer plays than just Google because whenever you buy Google Cloud, you're also getting Google search and YouTube and Google Network and some of these businesses are not growing nearly as fast. This is where this brings me to Coree because although Cororeweave is not actually the AI infrastructure play, they're one of the companies that are able to charge these higher and higher prices for compute that has a direct relationship with Nvidia. So, they actually get access to the chips and yet their demand is skyrocketing. By the way, they're a pure AI infrastructure play. It's essentially like buying a stock that's just Google Cloud and Google Cloud only. Yet, their growth is also exploding. Revenue backlog has gone from $25.9 billion last year to $99.4 billion this year. Now, 99.4 billion is so interesting cuz that's bigger than what Google was just one year ago. So, if they can continue to step up this backlog, that's going to be an extremely exciting feat for them. Yet they're being priced as if they're cheaper than Nebius, a company that is much smaller than them. Nebus is only doing roughly 300 maybe, let's say, $400 million in revenue. Coreweave is doing 5x that per quarter at $2 billion a quarter. And that's just right now. It looks like revenue is decreasing down to 111% per year, but Wall Street actually has this accelerating and essentially showing that potentially next quarter will be the low at 111% growth, then expected to go up to 154%, then 190%, then back down to 143, and then 124% which is even faster than where we are today. So over the next five quarters, we're going to essentially go from $2 billion per quarter. year-over-year that'll get to five and then 5.75 billion dollars at more and more profitable growth. You can see this in adjusted IBIDA that although the margins are going down to 56%, Wall Street is actually anticipating that this is where they bottom. If you look forward, Wall Street is actually expecting this margin to improve back up to 60%. And while we're see growth essentially go from $2 billion in revenue up to $5 billion year-over-year in adjusted IBIDA we actually expect more growth because of that margin expansion. So from 1.1 billion up to 3 billion year-over-year and then 3.5 billion the next year. Now the biggest thing that is stopping companies like Cororeweave from being profitable is IBIDA is stripping out all the interest taxes depreciation and amortization. But it's the depreciation that's really getting hit for Cororeweave. the chips that Coreweave is buying are essentially depreciating over a six-year rate. And so all the chips that you have, the more and more chips over the course of time, that means even more and more depreciation over time. Every single new chip that you get starts that six-year schedule once again. So going forward, the depreciation is expected to continuously increase. But whenever we look at the pricing of chips, just between February and the end of May here, the pricing on Ampia is going up or Hopper or even the Blackwells are all rising in price over time. So their initial total cost of ownership on these chips might actually be better than what they initially expected whenever they were buying Hopper or Ampier or Blackwell chips because although they expected to depreciate them over 6 years, customers are keeping these chips for longer, paying more for them, but yet the cost to operate them has not changed. Let alone the fact that Cororee is going to be one of the first companies to get access to Vera Rubin chips, which will be the top-of-the-line best total cost of ownership chips that you could possibly get access to. And because of that relationship with Nvidia and Cororeweave, having invested so much money into this business, they give them access to these chips to make sure that they succeed. Now, unlike Google, Google has a lot of other businesses that can help fund Google Cloud's growth. Coreweave, their growth is purely coming from the cloud segment. So although long-term it will be profitable and they do see a lot of clientele and demand, they need to actually fund this growth through dilution and debt, which is what they've been doing over time. And it's one of the things that scares investors because between the times of essentially December of 2023 and March of 2026, debt has more than 10x. You're talking about like a 12x in terms of the total debt along with them diluting shareholders as well. It's the biggest scare for investors. But this debt is actually getting a better coupon rate over time because banks are giving them the benefit of the doubt the larger that they end up getting. Another reason that I'm such a big believer in Coreeave is just where they rank in terms of performance via the other clouds that are actually in this space. And I mean literally every other cloud in this space. Google Cloud that's seeing a lot of this growth according to semi analysis which is the golden standard for understanding data centers and how they rank up is actually ranked in silver. Now I actually expect them along with AWS to go into gold by next time the ClusterMax 3.0 0 ranking comes out and coreweave will essentially well hopefully keep in platinum tier. You could also see Nebus potentially make its way into platinum as well as they've differentiated themselves versus Azure and Oracle as well. But there is a real benefit to being such a strong provider of products and having that relationship with Nvidia is that you are actually a part of the conversations of extreme codees. Coreweave has a direct relationship with Nvidia engineers and in part is setting up a lot of their software and hardware with Nvidia engineers to make sure that they are creating the best products for their customers. This according to semi analysis means that they can charge much more in terms of our testing. Coreweave clusters meets all criteria and sets the bar for other Neoclouds to follow. It is for this reason that we are aware of multiple examples where coreweave is able to command a higher price for managed slurm or kubernetes clusters by roughly 10 to 15% per gawatt per hour versus their direct competition such as nebus fluid stack cruso lambda and together AI. Indeed coreweave's pricing is closer to the pricing of the big four hyperscalers. The main challenge that we perceive for Coreeave going forward is to continue innovating in differentiating versus their competition so as to maintain their pricing power. We believe that they're successful doing so as they've been able to get access to GB200's and GB300 NVL72 generations which are the products that Nvidia is essentially promoting. But as you ended up seeing back here, they have been able to get access to Vera Rubin and so that essential pricing power continues going forward. So going back to the Google news, I think that this is extremely beneficial to not only Google but also the AI infrastructure layer like Nvidia, but that goes down all the way to all the component layers, right? The co-ackage optics, the memory layer, the actual chips, that's for Nvidia, Broadcom, AMD, all those companies. They're going to see more and more orders because the model layers like Enthropic and OpenAI are asking for more compute. Sarah Frier was just on the All-In podcast and she said, "I don't know where you guys can get more compute. If you can find any, let us know. But like for real, they're fully locked down and everyone is way over capacity. That's why Google's doing this. But the fact that the market actually did not oversell Google on this news, I actually think gives a lot of runway for companies like Microsoft or Amazon or Meta or Oracle to do the exact same thing. Meaning that these companies can expand beyond their cash from operations and actually expand into debt markets to really accelerate their buying. That's going to put more pressure on TSM, on Samsung, on Nvidia to be able to fill up those data centers with chips. So, those companies are definitely going to work. But as I continue to go down, it's a massive green signal to companies like Nebius and Cororeweave that they do have more customers and the demand is very real because even the hyperscalers are willing to go into debt or dilute or stop buybacks in order to fulfill their customers needs. I've been a huge buyer of Google and then eventually Nebius, but I think it's Cororeweave's turn knowing that the demand is there. They have access to the chips. They have the compute and the money to do so. And now it's just a matter of time for them to fulfill those orders and that the market actually sees them do so. But ladies and gentlemen, thank you all so much for watching. Let me know in the comments down below, how are you playing this? Are you buying the Neoclouds? Are you buying the hyperscalers? Are you buying the AI infrastructure? Let me know who the real beneficiary of this is. Until next time, thank you all so much for watching and bye for