원문 보기
Welcome to the channel. If you think Nvidia is just a hardware manufacturer selling graphics cards to tech startups, then Jensen Huang's big announcement this week proves to me that we are entering a completely new era of technology investment. So Nvidia is expanding its business model from selling silicon to providing the land, gigawatts of energy, and data center buildings needed for the future of artificial intelligence.
Nvidia announced a $1.5 billion investment directly in SoftBank's SB Energy to acquire land, power, and infrastructure at the Port Spike technology campus. Nvidia is providing credit support of up to $105 billion for an initial capacity of 4.25 gigawatts. Although there is a possibility of expansion to a total of 8 gigawatts.
And all of this will be leased exclusively to OpenAI for a period of 20 years. This credit support ensures that the Ohio facility will use Nvidia's own DSX AI factory platform, limiting competitors' access to OpenAI's largest adoption pipeline, but importantly, this energy deal is underpinned by a separate arrangement where Nvidia partners with alternative asset giants.
We're talking about companies like BlackRock, Blackstone, and KKR to create a pool of half a trillion dollars in outside capital. Essentially, this shifts the risks of building data centers off Nvidia's balance sheet . This also ensures that its customers have unlimited financing to purchase its GPUs.
This means that Nvidia is actively funding the foundation of the rapidly growing and changing AI economy. They are stepping in to guarantee infrastructure for cutting-edge labs, whose growth is already being held back by a lack of physical data center capacity. By entering the land and utilities market directly, Nvidia is signaling that physical power is the main obstacle between them and a long-standing monopoly in the hardware market.
I will note that this mega-deal also fits very well with Nvidia's broader corporate strategy of turning AI computing power into a standardized investment asset class. In this video, Jose and I will analyze three strong stocks that we believe have every chance of benefiting from Nvidia's multi-billion dollar infrastructure grab.
Let's start . Before we continue with today's episode, if you'd like to get some stock tips from our analysts , check out the pinned comment and description. Using this link gives you a promotional offer as a thank you for viewing. Thank you, and back to today's episode. Hi, Rachel. Hello, colleagues.
Welcome back to a new episode. Great introduction, and first let me say that I am extremely excited about Nvidia's quarterly reports. They will be out in less than 7 days. So, friends, if you're interested in Nvidia's financial results, I'm sure Rachel and I will be discussing that at the end of next week.
So be sure to follow our channel for updates. So, yes, I have in mind three stocks that could be significantly affected by this important move. And I imagine that during Nvidia's earnings conference call, this will be among the many questions that will come up during the Q& A session, where they'll discuss, " Hey, AMD, Nvidia, how are you going to be able to provide this kind of credit support?" How many companies will you be able to implement this with?
We constantly hear figures in the hundreds of billions of dollars. So I believe that during the reporting conference we will receive a lot of valuable information. But we see Nvidia trying to reduce risks or eliminate bottlenecks in its supply chain. And the first stock I want to talk about, Rachel, is TSMC.
I mean, TSMC's second-quarter report, released in mid-July, showed that the AI development cycle is still growing at a healthy rate. Revenue grew 36% year- over-year, and HPC now accounts for 66% of total revenue. Nvidia mentioned that each additional gigawatt for Nvidia is approximately $50–60 billion in revenue.
All of this money will flow through TSMC's advanced facilities. To build these AI data centers, Nvidia needs processors, graphics cards, networking solutions, various optics, various memory solutions, and storage, all of which are built at TSMC. So for me, it's not necessarily an obvious choice, but it's as close to a win-win as possible if the AI ecosystem continues to develop and Nvidia continues to address resource and power supply issues.
Another thing that really excites me about TSMC is their forecast, right? In its latest report, TSMC noted that growth for the whole of 2026 is expected to be just above 40%, and the Ohio plant will not start operating until 2028. So, if we're already seeing this type of growth for TSMC this year, can we imagine the revenue growth potential when such multi-gigawatt, single-location data centers become a reality for TSMC?
So, their market predictability is getting better and better , and that's exactly what we heard in the TSMC management report, right? They are confident in their demand forecasts for 2027, maybe 2028, 2029, 2030, that's why they increased capital spending, right? Confidence in revenue projections gives them confidence to increase capital investment.
And all this gradually begins to form a single picture. Why did TSMC tell us about the crazy growth in AI revenue? Why are they so confident in their capital expenditures ? Well, because Nvidia sees such a demand for its products that it can say, " Look, we're going to need so-and-so chips in 2028, in 2027, and so on." Now, it's important to note: I'm certainly excited about TSMC, but there's one point that should be considered a pessimistic scenario—the factories that TSMC is building domestically, here in the United States.
They lead to gross margin erosion, right? I mean, they were talking about losing 2-3% gross margin in the initial stages. They are also making significant expansions across many other lines of the supply chain. So all this requires money. And if for some reason —I don't believe it, but if it does—this AI thing isn't as crazy as I think it is, TSMC will be left with a bunch of production lines that won't do anything.
But this is where we are now, Rachel. I think TSMC's prospects as a potential long-term winner look pretty exciting. I think another extremely interesting point is how TSMC's production capacity utilization cycles have changed over the years. Once upon a time, and even until recently, their financial results were highly dependent on the refresh periods of consumer electronics, smartphones, and PCs.
So there was significant seasonal volatility for these factories, right? They are clearly a key player in the AI revolution. And when you look at Nvidia's recent announcement about how they're locking these multi-gigawatt facilities into a 20-year lifespan, computing hardware upgrades are becoming more standardized and predictable.
This could allow TSMC, for example, to much more confidently optimize its long-term deployments of extreme ultraviolet lithography systems from ASML. So these statements have many consequential effects. And I think you also know that it will be very important to monitor how TSMC's future earnings develop.
I mean, a lot will depend on their ability to implement overseas plant construction while keeping up with the complex requirements of next- generation packaging architecture. So, this deal is huge for them , but there are also a lot of internal rollout complexities for TSMC that investors should keep an eye on.
I agree, Rachel. That's why it's great that in these episodes we have the opportunity to learn both sides of the story. And technological innovation is a kind of risk, isn't it? So we have to make sure that this company continues to evolve in this complex chip architecture space. And how that happens has a huge impact on things like profits, output, and future demand .
And now, Rachel, I'll move on to action number two. So if I were to say that the number one stock was almost—again, nothing is ever obvious in the stock market, but the closest thing to obvious—I would say it was TSMC. The second one is a little bit more about there being an opportunity for growth here, right?
And with the opportunity for growth comes greater risk, because there are no free lunches in the market. So that would be IREN, ticker I R E N. By the way, it's interesting that they're also reporting earnings next week. So, if you're interested in this promotion, I'm pretty sure we'll talk about it on this channel.
The reason I put IREN in this video, where Nvidia provides the ground, energy, and shell, is because it's mostly "E" - energy. Energy is a competitive advantage in the current market we are in. And IREN has provided about 5 gigawatts of energy worldwide. Management said during the preliminary earnings call that demand for AI infrastructure far exceeds supply, and that capacity for 2027 and 2028 is already in short supply.
The fact that Nvidia is making such a large-scale deal with a third party, in my opinion, supports this statement. Another thing is that Nvidia is literally a counterparty of IREN. IREN recently signed a 5- year, $3.5 billion cloud AI services contract with Nvidia , plus a 5- gigawatt partnership, and Nvidia has the right to buy up to 30 million shares at $70, potentially a $2.1 billion investment, plus or minus a few hundred million, which is locked in as the GPU infrastructure is deployed.
So Nvidia is making deals, and that's exactly the deficit that I-REN is trying to monetize, right? This land, energy, and infrastructure are something that I-REN is building up quite well. Right now, one of the important moments for I-REN is the transformation from Bitcoin mining to AI cloud services , and that takes time , right?
Very little. The company, I think, is now worth about $12 billion, maybe even more, somewhere around $15 billion in market cap, but they still only get about $20 million in revenue from AI. The main reason is the time it takes to build such data centers. It takes time to provide them with nutrition.
It takes time to finally start making a profit from these GPUs , as all of this needs to be done in advance. But according to their contracts, they expect about $3.7 billion in annual revenue by the end of the year. They have a huge 5-year contract with Microsoft for 9.7 billion, and I'm sure as they expand their partnership with Nvidia, they'll add another 3.4 billion.
So, on just two players, they have almost $14 billion in contracts. And I'm sure these aren't the last deals. Overall, this is a good opportunity for the field of new cloud technologies. The downside is that there is competition. It takes time for income to start coming in. And, of course, buying GPUs and building data centers is a pretty expensive “hobby,” if you can call it a hobby at all.
Yes, one of the things that really strikes me when you look at what it takes to enable the AI revolution is the realization of the huge gap between standard data centers and the intense power requirements of modern AI computing clusters. We're talking about Nvidia's recent news, but it also highlights the shift towards facilities capable of supporting high loads.
They often exceed 100 kilowatts per rack. This essentially means squeezing the energy consumption of an entire suburban area into a space the size of a refrigerator, to use that analogy. So when you look at an infrastructure provider like I-REN, they are tasked with building or upgrading massive electrical substations that can handle this kind of continuous high-voltage power without creating a negative impact on local power grids.
And what's also interesting is that it actually rewrites some of the rules of corporate finance in the technology sector. You know, traditional cloud hosting relied on flat rates or simple lease agreements where tenants paid a predictable monthly fee. But now we see these huge energy needs forcing the industry to move to much more complex power purchase agreements with energy cost transfer mechanisms, which essentially means: tenants are taking on the variable costs of electricity consumption spikes that they themselves create .
And we are observing this. So Microsoft just signed a five-year contract with IREN for cloud AI services worth almost $10 billion, right? Microsoft has already officially commissioned its first 50-megawatt liquid- cooled cluster in Texas. And Nvidia also signed a multi-billion dollar deal with IREN for its internal workloads.
The last thing I will say about IREN: they operate at significant net losses, even with contracts worth billions, because the initial capital costs here are very high. Therefore, I believe the main test for management will be the ability to fund these initial costs of electrical grid and hardware before long-term rental income from prime tenants begins to show on the balance sheet.
Now, I think action number three, which you're going to talk about, will also touch on the " P" side of this LPS decision. So, Rachel, what's promotion number three? Well, I'll talk about Vertiv, and it's a pretty interesting business. It seems to combine the physical and chemical realities of the AI revolution together.
You know, if you think about placing one and a half million modern Nvidia GPUs in a site like Port Spike, which we're talking about, that's an unprecedented heat density, and traditional air cooling is not able to save the architecture of these data centers from melting. Managing these systems involves a great many profound complexities.
So Vertiv is a global leader in this segment of critical digital infrastructure. The company's management recently raised its 2026 revenue forecast to $14 billion at the high end. They expect over 30% organic sales growth, driven by demand for high-density AI data centers . And interestingly, Nvidia's commitment to ensuring these long-lasting data centers work means that even as the chips themselves are refreshed every few years, the cooling and power distribution systems built by Vertiv and similar companies remain in place.
This creates a very stable, high-margin maintenance and equipment replacement cycle for Vertiv. And what's interesting is that if you look at competitors that are now facing a capital expenditure squeeze, Vertiv actually generated over $925 million in free cash flow last quarter, even as it ramped up production.
Since we live in a time where Nvidia is the backbone of many of these multi-gigawatt projects around the world, Vertiv's order book is almost guaranteed to grow. So they are kind of the " shovel sellers" in this physical reality of the AI boom. You know, AI factories are essentially industrial thermodynamic puzzles, and that's why Vertiv solves the heating problem.
And without them, billions of dollars of computing power from companies like Nvidia would simply sit idle or overheat. So these high-margin infrastructure contracts that Vertiv is entering into make them, in my opinion, a very solid way to play this hardware cycle. Vertiv works side by side with Nvidia engineers to co-design power architectures and cooling systems directly into building envelopes from day one.
Therefore, there are also very high costs for changing suppliers. This means that when you have a multi-gigawatt facility optimized for Vertiv technology, replacing it with a competitor would mean completely dismantling the physical infrastructure. So, this effectively guarantees many years of maintenance, service and upgrade revenue for Vertiv.
This is a really interesting business, a little different from the other two stocks that Jose discussed today. A little different, Rachel, but as you mentioned, I want to emphasize again: Nvidia and Vertiv work together. I remember at the beginning of the video, Rachel, when you were talking about what Nvidia is doing, you mentioned a platform called DSX.
DSX, for those unfamiliar, is a platform created by Nvidia to demonstrate how to build a gigawatt data center. A gigawatt data center is a huge structure. It's just a gigantic, large-scale system. So, they created a kind of, I wouldn't say step-by-step, but an action plan. Here are the power solutions you need.
Here are the cooling solutions you need. And a great partner for them in this endeavor is Vertiv. And I think as we continue to see these energy-efficient building deals , one of the big winners is definitely going to be Vertiv. Yes. I think the main thing we're learning is that the physical reality of the AI boom is that you can have the best algorithms in the world, but if you don't have the chips, the power, and the cooling infrastructure, your AI factory will grind to a halt .
That's why today we've looked at these physical infrastructure companies, like TSMC, Eaton, and Vertiv, which we believe could be long-term winners for investors. Now, if you want more of these kinds of breakdowns, be sure to hit the subscribe button, leave a comment below about your favorite AI infrastructure company, and I'll see you next time.
Comments 0