Episode Transcript
The Finance Ghost: Welcome to episode 273 of Magic Markets. I'm coming to you from the Cape of Storms. It has been quite a day, so we record this on the Monday each week and yeah, I've gotten off lightly compared to many others in the province.
Moe, no such issues for you up in Canada for once. Although we've had many stories from you before of shovelling snow. So for once the problems are down here, not up there.
Mohammed Nalla: Indeed, Ghost. Luckily I'm not inundated with the storm. You are down in Cape Town. When you said shovelling snow, I literally still had to look out the window because even though we're sitting almost halfway through May at this point in time, it's still cool. We're down in single digit temperatures, it hasn't really kicked in. So interesting weather patterns globally, I guess.
But that's not the topic of what we're discussing today. I'm just glad I'm not sitting in those Cape Town storms and I'm glad that at least you're not swimming while we're recording this podcast.
The Finance Ghost: It is called the Cape of Storms, some of it is just to be expected, I guess. But yes, this week we are talking about something else entirely, which is actually just the top of the value chain, I suppose, for want of a better description, because there's a big old world beyond Nvidia.
And obviously here I'm talking about the AI value chain, and the market seems to be looking at these names - stuff like memory, AI server CPUs, the foundries themselves, edge computing devices. All of this is very relevant right now to investors in global stocks.
It's why you are seeing these massive moves in names like Micron, AMD, Intel. Some of them are just about meme stocks now, like SanDisk.
I'll give you a few examples. Micron's up 137% year to date. Intel up 217% - I'm going to have to try not to feel bad about selling it recently. I did get a big chunk of that rally, but not all of it, because it has carried on after I sold.
AMD, that's up 104% year to date. So essentially these names have doubled (and in some cases have tripled) in the space of just four and a half months, which is insane.
Nvidia, in case you're wondering and not keeping track, that's up 14% year to date. Actually really strong, and yet it manages to feel so tame in comparison to some of those movements.
Moe, it's pretty clear that the market is looking to play this AI theme at the top of the value chain, but to do it in names that go beyond Nvidia as the default choice of the last few years.
Mohammed Nalla: Yeah, I think that's really important. You've mentioned those year to date performances, but the real reason I think we're doing the show is just if you look at last week's performance, some of those names were up 35%, 30%, 25%. And so that's absolutely stellar.
If you took profit on some of those names just last week alone, you would have made the equivalent of two years' worth of an average equity markets return just in a single week on some of these names. It just blows the mind.
Now I think the key point we're trying to land on here is that there's a lot going on in the AI space. You've mentioned edge AI, you've mentioned devices. In fact, we've had to splice the show simply because we can't cover the entire ecosystem in a single show.
But I think the key point that really needs to be made is that the market has moved on from that very simple version of the AI trade, which was “buy Nvidia” because they are the ones that sell the GPUs, to a much broader question of who actually benefits from building the entire AI factory.
What does that whole value chain look like? And if we look at that, the AI factory, if you want to call it that, is not just the GPUs or the graphics processing units. It really hinges on things like high bandwidth memory. That's important.
In fact, as we move away from the era of just AI training into an era of AI inference, CPUs become a lot more important. And then there's storage that's hard drives, you need servers, there's advanced packaging, and these are just all elements of the entire whole chip ecosystem.
So it's more than that. It's also: who are the semiconductor equipment players? You need foundry capacity, you've got devices. We're not even going to cover that. But what we're seeing is this broadening of the AI trade across that entire value chain.
And the market is saying, if Nvidia is the obvious winner, then who are the second order or the third order winners? What does this entire cycle look like as it matures and as it plays out?
It's why something like Micron is suddenly being treated like this strategic AI infrastructure stock, rather than just a cyclical memory stock. Historically it was just a memory stock. It was just the cycles. Now there's something different to that investment thesis.
It's why a stock like AMD is being rewarded not only for GPUs, but for their CPUs as well, because that's being used in AI servers. It's why Intel's got that whole story around the foundry optionality. You're going to touch on that. Also the other names like Dell, Applied Materials, SanDisk, they're all getting dragged into the same trade.
Now the important thing is that these are not all the same story. It's the same theme, it's the AI theme, but it's not the same story because Nvidia is the dominant compounder with this really strong moat.
But Micron still has a memory cycle and AMD is still trying to prove that it can become the credible number two AI compute platform. Intel, there's a story around massive optionality there, but big execution risk.
And so for investors, the question is not simply: is this AI trend real? Is AI demand real? Yes, I think it is. I think the harder question is how much of the future's already been pulled into today's share prices. What's it actually reflecting?
And that's where stock selection becomes so much more important now than it was earlier on in the AI trade.
This week we're going to have a quick look at some of the names in this space. Ghost, why don't you kick us off?
The Finance Ghost: Sure thing, Moe. Let's start with Micron then, which we covered in Magic Markets Premium in April. It's one of the three scaled memory manufacturers in the world. Your other two are Samsung and SK Hynix.
These are not necessarily - well, I mean, everyone knows Samsung obviously, but Micron and SK Hynix are not necessarily names that people would have had on the tip of their tongue before this cycle.
And these companies are benefiting from explosive demand for memory. At Micron it's a combination of DRAM, it's NAND memory, it's high bandwidth memory.
And what we are essentially seeing, just for the non-techies (like myself to be honest), is a cyclical supply bottleneck. And that's a language that financial people, investors, will understand. It basically means demand exceeds supply. Supply cannot keep up - we've seen it in GPUs.
But the real question is: is it actually cyclical, or is it structural? Because here you've got data centres as the obvious underpin of demand, but you've also got devices which have increasing memory needs in this AI era.
And the risk here, aside from just the immense capex required in all these sorts of businesses, is that any normalisation of supply versus demand would cause a lot of pain across Micron's entire business.
So this is a good example of the kind of risk that the market just glosses over because it would make memory cheaper for everything. Not just data centres, but also for laptops, for phones. We're seeing that inflation come through everywhere right now and it would pretty much all dissipate at once.
Although revenue has essentially tripled in the past year, this is where you have to be very careful of that cyclical rug pull. You've got incremental supply that's expected to come on board from 2027.
And sure, between now and then the company is going to make serious profits - it already is, with gross margin at 75% in the quarter that we recently looked at in Magic Markets Premium, which is obviously insane.
Operating margin of 69% - most of the costs are in gross margin - even more nuts. When these cyclical names make money, they make an absolute fortune.
But it is a tough investment because if you go back far enough, you will find that the memory cycle has done this kind of thing before. So you have to be in the classic “this time it's different” camp to actually get involved in some of these names.
Moe, that's a theme across a few of these things, actually.
Mohammed Nalla: There's so much hope that's really trading in the market. You see it reflected in the valuations, and I guess that's where the risk comes from. Because if you're too bearish, the momentum continues. These companies are delivering stellar results, as you've mentioned, strong margins, and it keeps on going. If you're getting in at the wrong point of the cycle because of that hope premium that's effectively priced into a lot of these stocks, that introduces a lot of downside risk.
And then the other thing you mentioned, names like SK Hynix and Samsung. If you just go and have a look at the Korean Stock Exchange, those are two very large components of what's been happening up there. The index overall has rallied, almost as strongly as some of these names. And that introduces wider risk if you're just playing this on a passive basis with some sort of geographical exposure.
I'm going to pivot from that.
Let's look at AMD. You've mentioned the cyclicality, we've covered Micron - with AMD, you've probably got a cleaner, broader compute story coming through there because it's giving you exposure to both sides of this data centre story, that is both CPUs as well as accelerators.
Now, the market obviously wants to believe that AMD can take share from Nvidia in the AI GPU space, and that's really the exciting part of the story. But I actually think that the more balanced way to think of AMD is that AMD is trying to become this credible number two AI compute platform. It's not just a GPU challenger, but it's a broader play on the whole AI infrastructure theme.
The latest results certainly support that narrative. If you look at AMD's data centre business, it's been very strong. It's been helped by the EPYC server CPUs, and that's spelled E-P-Y-C. So again, you know, fancy spelling there for a lot of these big tech names. You've got the Instinct GPU shipments, and this really matters because AI servers, as I mentioned, they don't just need accelerators, they also need these very powerful CPUs.
As we move into this era of inference, to help orchestrate those workloads, it feeds the accelerators and it helps them basically support the broader system, as well as the use case for AI in general.
Now the bull case is easy to understand. If the AI capex cycle keeps going, AMD has multiple ways to participate. It's the GPUs, it's the CPUs, and that server CPU business, that's where they've already taken a meaningful share from Intel over time.
And then they've got that GPU accelerator opportunity where even if they get a modest share gain in what is a huge market, it can move the needle on a player like this. Then they also have credibility with the hyperscalers, which is absolutely critical because those giants are the ones writing the big cheques. So I think that's the investment thesis.
But then if you look at the risks, they're also meaningful because the first risk is that Nvidia's moat is not just hardware. Nvidia also has software. They've got developer lock-ins, their CUDA ecosystem. We've spoken about that when we covered Nvidia. That's really giving Nvidia exposure to the entire stack.
And so while AMD can produce competitive chips, the bigger question is whether customers will actually build the architecture and their processes around AMD the same way they've actually built it around Nvidia.
The other point is valuation. We've touched on that. When a stock has moved by more than 100% on a year-to-date basis, that's a lot that's already in the price. And so you've got to see continued earnings upgrades to justify that kind of momentum and for it to continue.
And then lastly, as I wrap up, the third point here is execution risk, because this is a market where supply - you've got software support, I mentioned Nvidia's ecosystem, as well as customer adoption with those hyperscalers, they all matter. So it's not enough to just have a chip that benchmarks well, that competes with your competitors out there. You actually need the ecosystem as a whole to scale.
And so my one line on this is that AMD is not just cheap Nvidia, if you want to call it that. It's the market's attempt to try and find who's the second major AI compute platform, who's the number two spot on that podium effectively. And that's a powerful story. But after the kind of move we've seen now, it really needs to keep delivering.
Now we've covered Intel, for example, in April. I think that's the one you're probably going to go into next. And I think you've made money, but again, that's another player in the space that's grappling with being a contender and a competitor, but also with some of those execution risks possibly coming through as well.
The Finance Ghost: Moe, I did make some good money here. I wish I could have made all of the money, but I mean that's true for everything in the markets, right? That's how these things go. When stocks go parabolic and if you don't have the discipline to sell when you believe that a valuation has gone too far, then you're really going to hurt yourself over time. You're going to wear a lot of sell-offs as well. So it is what it is.
We did the work on this in Magic Markets Premium in April. That was what triggered my decision to move along. Despite the share price having some strong momentum in it, the extent to which the market is just running ahead of the story here is actually incredible.
Obviously you've got stuff like the US government having a stake, good news around partnerships with Musk, with Apple. The US market is absolutely cheering for some kind of homegrown hero in this space. And all of the optionality really sits in their foundry business, which makes huge losses every single quarter.
Operating losses of between $2.3 and $3.2 billion over the past five quarters, and that is per quarter. It completely offsets the profit that they make in CCG, which is the Client Computing Group, which means group profit is essentially just the DCAI segment, which is Data Centre and AI.
As you could guess, DCAI is doing really, really well right now. That goes up, it means the group's numbers go up. But it's not because there's broad-based improvement. It's because two things are offsetting each other and then one is going up. So that can very quickly change.
And that's a big part of the problem with Intel. That story from all of these actually is the best example of how the market is just running on hopium right now. These guys are all selling hope. You've got some big IPOs coming in the AI space and it's just filtering through into all of these names at the moment.
Mohammed Nalla: So that's a nice place for me to actually just ask you a direct question. You've mentioned hope and hopium effectively, right, that these stocks are running on. Before we actually wrap the show, why don't you run us through what you see some of the key risks are?
We've discussed some of the key players here. We haven't even gone into storage, where you look at other players. You haven't even gone into servers. We can save that for another show. But when we're looking at specifically the chip manufacturers, what are the biggest risks on your radar?
The Finance Ghost: There are a few, obviously. I think valuation's got to be top of mind, right? You've got huge re-ratings, big assumptions about the future, you've got parabolic share price charts, you've got names that have never really done well suddenly looking like meme stocks, which is a little bit out of nowhere really.
And one of the catalysts for a correction here would surely be the rug pull if supply actually catches up to demand. Historically, it always has. So again, it's the “this time it's different” trade. You have to believe that to be true. And if we then get to an oversupply situation, it gets really nasty.
You've also got the geopolitics to think about, the strained trade relations.
But on the plus side, capex at the hyperscalers, that's the likes of Meta, for example, and Microsoft and all of those, that's the revenue for these players. And those hyperscalers make a lot of cash, so that cash can just keep flowing into the capex that these players are only too happy to provide. And they're doing so at such good margins.
So that's the mitigating factor here that is getting people very excited about this part of the value chain. But yes, there's a lot of risk. And if a rug pull does come, as per the name, it's going to come suddenly and it's going to be nasty.
Mohammed Nalla: I think you always have to balance those positive and negative points. I think a neat way to close this is that the AI trade is clearly not over, but it's definitely changing. The first phase was easy to understand. Nvidia was the obvious winner. They had the best accelerators, the best ecosystem, the clearest kind of earnings momentum.
But as the cycle matures, the next phase is a lot more complicated, and the market's now moving into names where that AI linkage is real, but the business models are very different.
I think the key point is these are not interchangeable AI stocks. Some are structural compounders, some are cyclical beneficiaries, some are turnaround stories, some are optionality trades.
I think the key message for our listeners is not to confuse exposure to the AI theme with the quality of your exposure to the AI theme.
The share price moves have been absolutely enormous, so the market's paying upfront for a lot of that future growth. And so I would say the opportunity is still there, but the margin for error is a lot smaller.
And that's why we do a deep dive every week into a new global stock in Magic Markets Premium. We've kept that at R99/ month, which we think is phenomenal value for anyone looking for deep insights into quality companies. So if you're not a subscriber, go and check it out. It's magic-markets.com.
To our listeners, let us know what you thought of the show. Hit us up on social media. It's @MagicMarketsPod, one word, @FinanceGhost and @MohammedNalla, all on X, or you can find us on LinkedIn. Pop us a note on there, let us know what you thought of the show.
Until next week, same time, same place. Thanks and cheers.
The Finance Ghost: Ciao.
This podcast is for informational purposes only and is not financial or investment advice. Please speak to your personal financial advisor.
