Exactly How I Would Invest in Memory Stocks

Added: 2026-05-31

[00:00:00]Memory maker SanDisk is up 4,000% in one year, and investors are gobbling up every memory stock they can find.

[00:00:08]Nobody knew how to spell Micron 2 years ago, and now it's everybody's go-to stock for 2026. By the end of this presentation, you're going to understand the single most important challenge memory stocks face. And I'll also tell you how I would play this. Took me one decade on Wall Street, two decades analyzing technology stocks, and three degrees to realize this one simple fact.

[00:00:30]When everyone's talking about a stock, you're already too late to the game. But that hardly seems the case for memory stocks. Since our last piece on memory stocks titled, "Forget Nvidia, memory stocks are exploding in 2026." The largest memory makers in the world have continued trouncing the NASDAQ. But Joe, the constant stream of free, educational, insightful content you produce just isn't good enough. Why didn't you tell me about this 2 years ago? Well, we did. Back in June of 2024, we highlighted HBM or high bandwidth memory as an emerging theme. And since then, shares of the three stocks we've highlighted increased nearly 500% on average. Now, that was courtesy of the work done by Spear Research, which happened to have their finger on the pulse of the HBM trend. Always stand on the shoulders of giants, but then put up a big curtain over the giant so nobody can see it. That's what they taught me during my MBA. But Joe, shut the hell up about ways you helped me in the past and start predicting the future. You know what? You're absolutely right. You, Joe Retail, sitting there in your McMansion watching this on the telly, you fall into one of three camps. Camp one, you invested quite early and now you're heavily overweight. As a result, when it comes time to sleep, you're tweaking like a moheaken. Camp two, you invested recently and you're holding on to triple digit gains. Or camp three, you're sitting on the sidelines with a bad case of FOMO. Folks in all these camps want to know just one thing. Is this party going to persist? So, there's really three things I want to cover. First, is memory moving away from being cyclical to having persistent demand over time? I want to know more about this industry cyclicality we hear so much about because if we continue to see this cyclicality, then you're just playing a game of musical chairs. That never ends well for Joe Retail. Now, Micron sure seems to be getting a lot of attention.

[00:02:25]Why is it a better name than the other two Korean names, aside from the accessibility factor? And why did SanDisk see such explosive growth over the past year? Over 4,000% price appreciation. And note that SanDisk isn't part of the big three HBM manufacturers. So let's start there. In a recent piece we did ranking 18 of the hottest meme stocks. Probably the strongest company on that list surprisingly was SanDisk. How can SanDisk be a meme stock? Well, you need to ask Roundhill that question because we used their list of 18 meme stocks.

[00:03:03]And that's the second reason that you stand on the shoulders of giants. When someone questions your source, you don't have to do any research. You just point to the giant. Now, while Roundill's assets under management of around $8 billion would hardly be considered giant, their recent ETF debut sure put them in the news cycle. Round came out with their memory ETF, aptly titled with the ticker DRAM back in April, and it's already cleared $5 billion in assets under management. It's the hottest ETF since Bitcoin Mania, said CNBC, which means Round Hill is absolutely killing it. $1.1 billion last week raised in one day alone. That's because success on Wall Street for fund managers isn't about performance at all. It's how much money you attract. And memory investors almost have to use Round Hills ETF. And here's why. You cannot recreate this ETF yourself. That's what you're paying them 65 basis points for. Both Samsung Electronics and SKHEX are not easily available to US investors like Micron is. Hey everyone, Wyatt here, the head of operations at Nanalyze. Just jumping in to say that as of just this month, these stocks are accessible to US investors via Interactive Brokers. They just allowed access to the Korea exchange on their platform. So, if you are a US investor and you do have an interactive broker account, it is possible to buy shares of these companies directly on the Korea exchange.

[00:04:31]>> Now, investing in a broad country Korean ETF isn't viable because the weights that you get for those two names are around 50%. So, half your exposure is stuff you're not looking for. And here's what's interesting. Round actually backs out some of that country specific risk, it seems, by shorting Korea and Taiwan.

[00:04:50]You can see that under the ETF constituents. But Joe, I don't care about learning. Just tell me WHAT TO DO NOW. ALL RIGHT. All right. Fine. Here's how I would invest in memory stocks today if I chose to. I'd pay 65 bips to get a balanced basket of constituents managed by Roundill with 73% exposure to the world's three largest high bandwidth memory producers. Then I'd set some cell rules. For example, if demand drops across two or more players, that's a sign that the thesis is coming to an end and we're going to see some cyclicality.

[00:05:28]I'd also have some trim rules for overweights. It's never a series of binary investment decisions like most retail investors think, right? You can trim, you can slowly move in or move out of stocks. Time horizon here would be anywhere from 5 to 20 years. So 5 years if it's going to be cyclical or 20 if it's not. Eventually, this is going to become cyclical again because it's a commodity. Time until cyclicality is really the question here. Not true. who say many management consultants who still get their $750 an hour if they're wrong. IDC consultant Peter Gibbons says what started as a cyclical uplift in data center spending has become a selfreinforcing investment cycle. The super cycle is here says Bob Slidell of KPMG. AIdriven demand isn't a bubble, it's a fundamental market shift. And Bill Lumbberg of Deote says AI workloads could triple or quadruple annually through 2030 driving chiplets advanced packaging and of course high bandwidth memory. Now the names of those sources may be made up but the company's conclusions aren't. Experts are really saying this time is different. This is not your grandfather's semiconductor industry which has typically followed cycles that last 3 to 5 years on average. Now these cycles were driven by long fab lead times. So capacity additions take two to three years and they come online everything's changed right you have inventory buildups you have volatile endmarket demand for PCs mobiles data centers and of course AI and there's a correlation there with global economic growth. Now I think the first bullet point is the toughest to manage. So how much should you expand in today's maniacal world? It would be very difficult for these three companies to converge appropriately when it comes to supply without engaging in collusion, one would think. But regardless, let's assume the AI super cycle party continues forever. That ETF strategy pretty much has us sorted. But there's one important question I think we're overlooking here, and that is this. Are these memory stocks overvalued? Today, I'm going to use SanDisk as an example.

[00:07:37]As I said, over a single year's time, price appreciation of over 4,100%. Now, they're in fourth position in the DRAM ETF waited at about 5%. And even though shares soared 4,000%, here's the crazy part. They still command a reasonable valuation. I'm going to show you why in a second, but first, let's calculate valuation. We have our own simple valuation ratio, which is market cap divided by annualized revenues. That gives us 10. Well, our catalog average sits at around 6 or 7. So, it's really not that far above average when it comes to a revenue ratio. Now, when we look at forward price to earnings ratio, that's expectations of profits in the future based on the price today, we get a forward PE of just nine. That's about half of the NASDAQ average. Neither of these numbers are overvalued when compared to appropriate benchmarks. And here's why. Profits for this company over a single year jumped 8,000%.

[00:08:38]You can pause it and just look at the simple numbers right now. They had their Nvidia moment. And like Nvidia, they're not overpriced right now. Here's the crazy thing. You open up their last earnings deck. It's remarkable to see.

[00:08:51]You have 30% gross margins to 90% gross margins in under a year. What does that mean? It means they have incredibly strong pricing power. Similar to the increase we saw in gross margins for Nvidia. Look at the revenue trends by end market. Of course you see revenue coming from data center and edge surging to the moon there. Whilst consumer is actually dipping. There's an opportunity cost to be serving that segment that isn't growing so fast. Every hyperscaler wants to lock in this critically important supply chain component before everyone else does. No price is too high. You can recall the famous Jensen dinner when what was it? Musk and Ellison were saying, "Please take our money." They're locking in resources right now at any price. It doesn't matter. Will that always be the case?

[00:09:38]No, it won't. Everybody's signing on the dotted line these days. You see these press releases left and right. Yet, nobody knows what these contracts actually look like. Sure, they talk about how favorable the terms are in a press release, but that's not the actual contract. And guess what? Nobody cares.

[00:09:55]It's very easy to visualize these future growth opportunities today. Now, SanDisk is seeing fundamentals match their price appreciation, but what about the other HBM players? Now, I'm actually on my way to Korea right now as part of our international investing series. I just had to take my holiday first. This is my holiday face. Woo! Now, the last time we covered Micron, we noted the very cyclicality we were discussing earlier.

[00:10:23]Take a look at this chart. And I'm going to read you the headline that accompanied this. It said this. Micron's revenues plunge 40%. That's a massive draw down amid weak economy and China sanctions execs banking on Gen AI boost.

[00:10:41]Boy, did they bank right. So he pointed out back then the need to watch this compute and networking segment because ideally that's where all the growth was going to come from. Well, guess what?

[00:10:52]Growth came from literally everywhere.

[00:10:54]Look at their latest results. So if we break down these newly named four revenue segments, you see that cloud memory had so much revenue in that sigle segment that it was over the total amount the company brought in in Q2 2024. Same thing for mobile and client and core data center almost brought in as much last quarter as the entire company did in Q2 2024. And look at how strong these gross margins and operating margins are becoming. Again, they have massive pricing power right now. That was sure some excellent opium the management team was smoking back then because look at this Nvidia moment. So, they're expecting upwards of $30 billion next quarter. It's almost too incredible to believe. And you know what the cause of that is? It's real simple. All the hyperscaler spend money at all costs that's coming down the pipeline and companies are starting to receive a piece of that. That's why Micron's expecting 81% gross margins alongside that $30 billion or upwards of $30 billion in revenues. What you can then start to do is look forward based on analyst projections of future earnings and start to calculate the price target.

[00:12:08]In this case, you calculate a current PE ratio for Micron of around 35. But then you can take next year's forecasted earnings and you could assume a price to earnings ratio, an average one of 20, and you get a price target of $3,380.

[00:12:25]My 350% gain thesis. What it comes down to is simply this. Trying to bet on a winning horse is only a good idea if you're allowed to bet on all the horses.

[00:12:37]Some folks might be picking Micron because it's the only HBM stock they can purchase. Now, instead of cherrypicking, our aim should be to reduce company specific risk at an already volatile space. Just go long with an ETF on a super cycle assumption, add on dips where fundamentals aren't showing any weakness, and Bob's your uncle. Now, I know that some of you out there would prefer that I only find profitable trades, execute them for you in your brokerage account, and then help you figure out how to spend your newly found millions. I'm sure you can find plenty of 25-year-old life coaches out there offering these types of market prediction services. It's not what we do here. Our job is to help you think through the hype. Regular viewers already know the four most dangerous words in investing. This time it's different. Is this time different? Is memory suddenly not becoming cyclical?

[00:13:29]or is this just a temporary super cycle driven by both supply constraints and insane demand that isn't likely to persist. Remember, you don't have to make a binary decision here. Going back to the three profiles I mentioned earlier, Joe Retail with his fourdigit gains can start to do some light trimming. Ideally, rules-based trimming like we've done with the Nvidia shares we've been holding since they were a $40 billion company. Joe Retail with three-digit gain, same thing. and Joe Retail who has just entered a position or is thinking about entering one right now. Always establish before buying any stock the criteria that would cause you to sell it. Memory trade may have plenty of room to run, but hyperscalers aren't going to be spending trillions in capex forever. Watch for margin deterioration.

[00:14:16]Now, another way to play the AI trade without the potential cyclicality is by investing in the five most important semiconductor companies out there.

[00:14:25]That's covered in this next piece you ought to watch. Thanks so much for taking the time to watch this video today.