Are DRAM Stocks Too Hot To Handle Yet or Is This Just the Start of a Bigger Move?

DRAM makers have arguably been the in the hottest corner of the semi trade so far this year, and while a wave of increased volatility has made the names tough to buy on weakness, there is a great debate surrounding whether or not DRAM is experiencing structural demand as AI compute demands continues going off the charts with new applications (many of which may be highly monetizable) coming online in the coming months and quarters, all while firms look to get their orders in way in advance.

Indeed, increasing production feels like a great move, but it’s one that won’t alleviate the supply-demand imbalance for quite some time. And who knows if the extra supply coming online will be scooped up, pushing the short supply further out into the future.

Structural tailwinds facing memory chip makers?

It’s hard to say, but it feels like higher prices of high-bandwidth memory are the “new normal,” so to speak, even as firms, like Apple (NASDAQ:AAPL | AAPL Price Prediction), look to procure a bit of DRAM supply beyond the likes of the Big Three, consisting of Micron (NASDAQ:MU) and the South Korean makers in SK Hynix and Samsung. In any case, this is clear for the DRAM stocks: they’re making big profits right now, and they’re poised to keep raking it in for some time with great earnings visibility into the next few years.

Indeed, it’s hard to deem that things really are different as the AI revolution shows no signs of slowing down. At the same time, though, the so-called “RAM-pocalypse” or “RAM-mageddon” is really pressuring consumers where it hurts. With suits against the big RAM makers thrown into the equation and continued pessimism regarding the hyperscaler’s big CapEx plans, maybe a scenario does exist where CapEx is scaled back, and the DRAM stocks lead the charge lower, perhaps giving back much of the gains enjoyed this year.

Looking at the hyperscaler CapEx

At the same time, though, it’s hard to imagine that one hyperscaler wants to put itself out there first by announcing plans to pull back on AI-related CapEx, especially considering what could be on the line. In any case, the global DRAM supply is concentrated within the hands of just three firms (The Big Three, as they’re called), and there’s a good chance that the shortage just keeps worsening before it gets better.

Apple doesn’t just raise prices that aggressively on existing hardware. It truly is an unprecedented time or a 100-year flood, as Apple CEO Tim Cook referred to rising RAM costs.

In my humble opinion, there are risks on both sides of the coin. If CapEx just keeps going higher and hyperscalers do whatever it takes to gain an upper hand in the AI race, price hikes on everyday electronics as a result of higher RAM prices could keep on coming. For the Big Three, that means massive multiple compression and mouth-watering earnings growth, perhaps for some while longer.

At the same time, if hyperscalers decide to pause (I think higher interest rates might do it), the DRAM stocks could be at risk of serious downside. Even if the structural demand narrative proves true, a breather period could set in, perhaps as part of a broader tech-driven market sell-off.

The easy money has already been made

At this juncture, it feels like investors are already well over the long-lived RAM shortage narrative when it comes to Micron. While I have no idea what the next move will be, I do think that as an investor, it just makes sense to be ready for volatility and bigger corrections. Shares of Micron might be down 20%, but this is a mere blip considering the scale of the past-year rally, which has taken shares up close to 700%.

With Michael Burry shorting the firm, perhaps now represents a great time to take profits if you’re already in the name, while nibbling into weakness could make sense for those who are underexposed to DRAM. Personally, I think Micron stock is too risky to buy while the market cap is at north of $1.1 trillion.

My main fear is what could happen if a rate hike does happen in the second half, and what could happen if hyperscalers can find creative ways to advance AI with less DRAM.

Contact [email protected] for any questions or corrections.