英伟达的NVLink Fusion战略极为看涨，且该股被低估，因为市场最终将认识到其超过5万亿美元的增长潜力。

Nvidia pretty much said it's game over for its competitors. I want to take a closer look at Nvidia cuz Nvidia pretty much said it's game over for its competitors.

I want to take a closer look at four stocks that I believe can benefit dramatically from this and one stock that I believe needs to make some changes. And if they don't, then things might be a little bearish for this AI giant. today I want to look at Nvidia and this is Nvidia's earnings week.

So, we're going to get a lot of content and a lot of things coming out from Nvidia. with Nvidia earnings coming up, I'm actually pretty happy that the stock is not as excited, right? We were over $220 about a week ago. Now, we're at 208.

I'm a long-term investor of Nvidia, right? This is the company that has completely changed my life. Now, I don't know when Nvidia's going to have that quarter that's going to completely flip the switch and make it exciting again for the AI space.

And the reason I'm saying that is because year-to-date, the stock is pretty much flat, right? Up 10%. That's nothing, especially if you are the AI leader.

I don't know if it's this quarter or next quarter or the quarter after that, but one of these quarters, the market is going to wake up and say, "I was wrong with Nvidia and this, even though it is the biggest company at a $5 trillion, it can actually be bigger."

Now, in the intro of this episode, I mentioned that there's a market Nvidia is pretty much saying competition is dead and competition is so dead that that you need to come back to Nvidia. And that competition is the XPU market.

But Nvidia says that chip design is hard. And if you're building an XPU, you are wasting your time and this is a market that's dead unless unless you enter the Nvidia ecosystem.

Now, the reason that they mentioned this is because back then, right back then in the chip design process or in the kind of just the compute market, all you really had to do was sell a chip. And if you sold a chip, you got revenue, you got collected and that chip worked well.

Now, in this AI market that's completely changing because you're building factories, right? You're building full factories. And when you're building full factories, you need to be able to design you're no longer designing a chip, you're designing a system.

And most companies are still just designing a chip and that's not even half the battle, right? Probably not even a third of the battle because if you're building we're building these gigawatt data centers now, right?

And historically, a gigawatt data center, we can see it's collecting roughly $50 of revenue per year. Give or take plus uh plus or minus $10 billion. Regardless, these massive data centers are crucial and the number of revenues that they're collecting is extremely high.

So, if you are building an AI chip, you need to make sure that you're able to build an AI data center. And that's no longer just a chip itself.

You need to include the scale up, the scale out, all the networking types of solutions. You need to be able to understand how you can racks at your rack scale architecture production factory of software. You need to have kind of this crazy supply ecosystem that allows you to not just say, "Hey, I'm just building a chip, but I'm building this whole AI data center."

Because money is time or time is money, right? NVIDIA you can call them up and say, "Listen, I need a 1 kW data center." NVIDIA says, "No problem. I know how to handle the cooling. This is the blueprint that I have to do all this. You'll be able to set up your data center in no time. You know these chips are great, and you're going to be collecting revenue in no time."

If you're doing an ASIC with somebody internally, had they thought about all that already, right? And if they haven't, what is going to be the timeline of finally getting all that ready?

And then what if it doesn't work? You missed out on tens of billions of dollars on revenue on a 1 kW data center because you were trying to save a few margins on an ASIC solution.

Uh so, NVIDIA is saying, "If you're in this ASIC space, it's hard work. But, if you really want to be in this ASIC space, come with me, right? I'll help you out. I have this thing called NVLink Fusion, which allows you to bring your ASIC into the NVIDIA ecosystem."

So, while maybe you're no longer a 100% of my customer, you're still 60, 70% of my customer, and we both win because you feel like, 'Hey, you don't aren't not buying all my chips.' And I win because I'm still growing and growing with you as well.

Now, NVLink Fusion allows, like I mentioned, the connection of these XPUs to NVIDIA's infrastructure. And NVIDIA actually gives a lot of reasons why this is pretty important.

And like you kind of connect with an NVIDIA's ecosystem, you're connecting with their scale up, their scale out, their CPUs, and the list goes on. But, but the main reason NVIDIA does this is because or they're offering this and they're bringing it up is they're saying that this is a great way to manage risk, right?

A data center locked to one chip can become a schedule risk. And like I was talking about, imagine you're building a data center that only is meant for one type of ASIC chip that you are designing.

And what happens if there's a hiccup? That data center is not making any revenue. But, what if you are part of the N D I N D link fusion program? A customer can deploy direct level solution with N D I GPUs to start off with, right?

We're going to build we're building this data center my AI chip is not ready yet. Let me build with N D I's GPU and the whole ecosystem.

And then, they can decouple the development of their XPUs and put it at a different pace. So, maybe month one, month two, month three, as you're building this data center and you're putting things on, you're going to start with N D link and N D I's main solution.

But, now five months have happened and you've actually feel more comfortable with your AI chip and you are able to put it into this N D I ecosystem where now these two servers can live together.

So, it's the best of both worlds because you're able to start collecting revenue as soon as possible because N D I is the go at designing chips. But, at the same time, you're also going to be able to make your shareholders happy.

You may be able to increase your margins because as times progress, you're going to be able to start adding your own ASICs into these solutions as well, but still be part of the whole ecosystem.

And I think that is extremely extremely extremely bullish. Now, there are a few partners that I think can benefit from this. N D I has actually given us the N D I link fusion list of the stocks of the companies that are working.

In the CPU side, you have ARM, you have Intel, you have Fujitsu, you have Sci-Fi.

Now, one thing that I found pretty interesting was Qualcomm used to be a part of this list. And I don't know where they went. I don't know if that's bullish or bearish, but they used to be part of this list.

Now, in the custom silicon, you have Astera Labs, you have Alchip, you have Samsung, you have MediaTek, and you have Marvell. In the technology partners, you have Cadence and Synopsys. These are the ones that help you design these.

In the optical partners, you have Aear Labs, Light Matter, and Marvell. So, you can kind of already guess what are the markets that I believe can benefit from these, and it's going to be the parts within these ecosystems.

And the one one of the players I believe is very high likely of winning is going to be Marvell. It's not a guarantee, but very high likely. The reason I bring that is because in March 31st of 2026, Nvidia brought Marvell they they joined forces in the in the kind of expanded partnership.

And one of the partnerships was the optic side. The other things was the NVLink fusion. Nvidia ended up investing $2 billion in Marvell. Around the same time frame, we did hear that AWS, who is building their own chips to train them for is actually going to be using Nvidia's NVLink fusion.

And the one that designs and helps AWS with the training for is, you guessed it, Marvell.

Right, Marvell is the one winning here. And and now Nvidia I I believe this is really good for AWS understood. AWS understood that if we're going to have to be if we're going to be able to be the a good performer in our AI chip, good total cost of ownership, we have to join with Nvidia.

So, it's interesting. But then you have already announcements like this where Nvidia is actually still keeping that customer that designing their own AI chip, and they're going to be able to license a lot of products.

The other one that I thought was pretty interesting is Marvell and Google announced a partnership. Now, Marvell Marvell is not going to help Google design their own TPUs, but they're going to help specific AI accelerators.

Is this kind of the entry-level point for Marvell to somehow sleep slip the NVLink into Google? Because at the moment, there is a company that's very popular in the ASIC space that does not is not part of the NVLink fusion space, and that's Broadcom.

And that's the Broadcom company that I'm a little bit I want to say worried about, but I think they need a change certain things and we'll talk a little bit about there.

Microsoft is also working with them for the Maya 300, right? So, Marvell does have a lot of I wonder if we're going to get an announcement where one of the upcoming Mayas from Microsoft's A6 might end up being part of the NVLink Fusion world.

So, few stocks that I believe can benefit. First, Marvell, right? I think NVLink Fusion is going to be big, especially as we're getting into these sizes and this would be a good kind of revenue driver for Marvell.

Not only because it's going to be a partner for the NVLink, but most likely if you're trying to work on certain networking solutions or certain optic solution, Marvell is going to be that partner because they're working with like, "Hey, you're already doing the NVLink Fusion with us. Let us also work with you on some of the networking and optic solution as well."

So, Marvell is the first one. The next one is Astera Labs. This is another partner from their kind of list of of custom silicon. You have Astera Labs here and I do believe they can benefit dramatically.

Again, with all these types of approaches and Astera Labs right now down roughly 42 from its 52-week high or a year-to-date high, Astera Labs is another one that I believe can benefit in the long term of things.

Next the two are the kind of hyperscalers that are already working with Marvell. Amazon is the first and Microsoft is the second. Now, I'm a little hesitant on buying these because we did buy them a lot cheaper when we were kind of in the 220s below, we bought Amazon in the sub 400s, we were buying Microsoft.

So, bullish long-term, maybe not right now for individual stocks unless I'm dollar cost averaging.

Marvell and Astera Labs, I don't mind being a little bit more aggressive on these high beta names because they have pulled down a nice amount from their 52-week high and I am very bullish.

Again, this doesn't mean they're going to be winners. The market can always shift, but based on the data points that I have right now, I would be excited on these two stocks at the moment.

Now, like I mentioned, there is one stock that I I think needs to change how they're working with and Broadcom does not is not part of the NVLink Fusion.

Right, they're not part of the NVLink Fusion.

But, they have a huge customer base, right? They have Anthropic, they have OpenAI, they have Meta, they have Google and they have other customers.

What if those other customers start to leave Broadcom or start to work with other players because they're so hard-headed and don't want to be part of the NVLink Fusion world.

I For example, if I was OpenAI, I was if I was Nvidia, I'll go to OpenAI, I'll be like, "Are you sure jalapeno's going to do good? Are you sure you guys have the supply chain unlocked? Are you sure about this, that, and this, and that? Wouldn't it make sense for you to go talk to Broadcom to make sure you are still part of the ecosystem because you're going to make sure that everything's right, right? We have all this worked out."

So, if I'm a company like OpenAI and Anthropic, right? These are two players that unfortunately don't have the cash flow like the hyperscalers.

So, if a chip fails and it chip fails really bad, especially if they're building something at massive scale, that is going to be a huge hit for them.

For Google, if one of their AI chips fails, nothing bad happens, right? I mean, I think they they'll take a hit and maybe their stock might be punished, but it's not end game over.

If Open AI and their optics spent 500 MW on a chip that didn't scale up properly, that is bad news. And I'm pretty sure Nvidia is letting it be known to those players.

And I wonder if Broadcom is going to eventually be a part of the NVLink Fusion domain or partnership. If actually it does, I would actually be more bullish on Broadcom than I am right now.

Now, I'm not bearish Broadcom, and I don't think it's game over completely for them, but it is something that investors should keep a closer eye on. And I would rather right now focus on companies that are part of the NVLink Fusion partnership list.

So, I hope you guys enjoyed today's episode. Take care, have a good day, and see you all next time.