Thematic Memory ETFs Give Investors a New Way to Play AI’s Hidden Bottleneck

Memory stock prices have had the wind knocked out of them this month as several negative catalysts have converged to rock this suddenly volatile group. The boom lifted many cyclical companies to unprecedented heights, driven by insatiable demand from hyperscalers. But for the first time in a while, cracks have appeared in the story. Valuations among memory chip makers are stretched, price hikes are facing pushback, and Chinese memory producer ChangXin Memory Technologies is threatening to upend the market with a massive $8.6 billion IPO in Shanghai. Is the memory trade about to be disrupted? Not so fast, my friend.

The Differentiating Aspects of the Memory Shortage

Hyperscalers are hungry for specific types of memory, but this demand has had a cascading effect on the industry. Companies like Micron Technology Inc. (NASDAQ: MU) and Samsung Electronics Co. Ltd. (OTC: SSNLF) have all capacity accounted for through 2026, and many analysts project memory shortages to last until 2027 or 2028 at the earliest. This supply shortfall is the central theme of the bullish thesis, and why these recent headwinds are likely a short-term blip.

Threats of hyperscaler spending slowdowns, Chinese competition, and oversupply are fears for the coming years; the reality is that memory suppliers are entrenched in lucrative multiyear contracts with deep-pocketed hyperscalers. Typically, memory chips fluctuate in price as supply ebbs and flows, but these long-term deals also lock in prices for extended periods, granting longer runways to high-margin sales.

It’s important to understand the different types of memory chips these companies produce, since the type of memory matters when selecting a fund. Here are the main components in the data center memory stack:

• Dynamic Random-Access Memory (DRAM) - The active working memory found in most computers, tablets, and mobile devices, which stores data on capacitors. DRAM is a volatile form of memory that must be connected to a power supply to retain information, but its high speeds and low latency have led it to dominate the memory chip market.

• High-Bandwidth Memory (HBM) - A specialized type of DRAM that’s increasingly crucial to the AI data center supply chain. HBM uses a 3D stack to vertically stack memory chips, reducing power consumption and creating the bandwidth that hyperscalers find attractive. HBM has emerged as a core component in the AI buildout, and only Micron, Samsung, and SK Hynix Inc. (NASDAQ: SKHY) are currently capable of producing it.

• NOT-AND Flash (NAND) - NAND Flash memory chips are non-volatile, meaning they retain data even after the power source is cut off. Users of NAND memory care more about efficient storage than high speeds, and it's primarily used in hard disk drives (HDDs) and solid-state drives (SSDs) sold by companies like Seagate Technology Holdings PLC (NASDAQ: STX). HBM might be the headline demand from data centers, but storing training data also requires non-volatile memory with high storage capacity.

3 ETFs Offering Unique Ways to Play the Memory Stock Surge

Your opinion on the next direction of the memory trade will decide which funds fit your portfolio. Each of these ETFs follows a different thesis, but the underlying holdings do have significant overlap. And remember, these memory ETFs still have spots, so you can expect volatility and high spreads no matter which fund you select.

Roundhill Memory ETF: High Liquidity Fund With Focused HBM Exposure

The group's veteran fund is still just a few months old, but the Roundhill Memory ETF (BATS: DRAM) has already amassed more than $20 billion in assets under management (AUM) and developed a healthy options market. More than 42 million shares trade daily on average, and its 0.65% expense ratio isn’t terrible for a fund this uniquely tailored.

The ETF holds 23 assets in total, with the highest concentration in the South Korean HBM producers SK Hynix and Samsung. It also allocates a smaller share to flash memory producers like Seagate and Western Digital Corp. (NASDAQ: WDC) and maintains liquidity through Treasury holdings. An investment in DRAM is a concentrated bet on HBM (with a sprinkling of NAND), but with less volatility and risk than our next fund.

Kurv Memory Select ETF: High Risk Through Highly Concentrated HBM Exposure

The Kurv Memory Select ETF (BATS: KMEM) began trading on June 30 and currently has less than $50 million in AUM. But if you have high conviction on the HBM supply shortage lasting for multiple years, KMEM may have the highest upside.

The concentration is heavy; SK Hynix, Samsung, and Micron account for nearly 80% of the holdings (with more than 41% in SK Hynix alone). Despite the low AUM, the fund matches DRAM’s 0.65% expense ratio, which is again fair for a thematic ETF. KMEM investors should prepare for high volatility and be comfortable with a new, risky vehicle.

Tema Memory ETF: Broader Exposure Across the Memory Ecosystem

The Tema Memory ETF (NYSEARCA: DISK) adopted its stock ticker because it's a much more diverse play on the memory chip industry. The fund has 22 stock holdings, none of which have a higher concentration than 18%. You’ll get NAND flash makers like Seagate, Western Digital, and SanDisk Corp. (NASDAQ: SNDK), along with HBM makers like Micron, Samsung, and SK Hynix.

The fund also offers exposure to companies that don’t trade on U.S. exchanges, like Kioxia Holdings and Nanya Technology. The expense ratio is high at 0.75%, but it has amassed more AUM ($76 million) than KMEM over the same period and offers much broader industry exposure than the other two funds. If you don’t want to be overexposed to hyperscalers, consider DISK over DRAM and KMEM.

Where Should You Invest $1,000 Right Now?

Before you make your next trade, you'll want to hear this.

MarketBeat keeps track of Wall Street's top-rated and best performing research analysts and the stocks they recommend to their clients on a daily basis.

Our team has identified the five stocks that top analysts are quietly whispering to their clients to buy now before the broader market catches on... and none of the big name stocks were on the list.

They believe these five stocks are the five best companies for investors to buy now...

See The Five Stocks Here

The article "Thematic Memory ETFs Give Investors a New Way to Play AI’s Hidden Bottleneck" first appeared on MarketBeat.