A new ETF for memory chips has emerged: DISK and KMEM. How to choose?

The U.S. stock market has launched two ETFs specifically betting on memory chips: DISK and KMEM, both listed around June 30, and their strategies are completely different.

Why are two more memory ETFs suddenly appearing?

The answer is just one word: shortage.

The larger AI models get, no matter how powerful the GPU, it’s useless without fast enough memory to feed data. HBM, DRAM, and NAND flash have all become the tightest links in AI infrastructure. Industry forecasts predict this market will grow from $216 billion in 2025 to $758 billion by 2027—more than tripling in less than two years.

Wall Street’s reaction is always straightforward: once a structural gap appears in the market, capital will find a way to create a tool for "one-click buying."

DISK and KMEM are products born from this logic.

What exactly is the difference between DISK and KMEM?

Let’s start with the basics: DISK (Tema Memory ETF)

An actively managed memory chip fund, listed on the NYSE, with an expense ratio of 0.75% and 19 holdings.

The strategy leans toward diversification: covering HBM, DRAM, and NAND, but the top two holdings are SanDisk and Kioxia, at 23.04% and 20.76% respectively, totaling nearly half the portfolio—clearly tilted toward NAND and storage. Samsung and SK Hynix are ranked lower, each with a weight below 8%.

KMEM (Kurv Memory Select ETF)

Expense ratio 0.65%, listed on Cboe BZX, opened for trading on July 1. The holdings are highly concentrated: SK Hynix 41.53%, Micron 19.85%, Samsung 18.81%—the top three companies account for over 80% of the portfolio.

The fund requires that its holdings derive at least 50% of their revenue or assets from memory chip-related businesses. The prospectus states that derivatives can be used to track the stock performance of these memory companies, without necessarily holding the stocks directly—it can even buy other memory-themed ETFs. This gives it more flexibility than similar products. Even the previous DRAM ETF only used swaps for part of its positions in Micron and Samsung; KMEM has a wider range of tools at its disposal.

There is some overlap in holdings between the two—for example, SK Hynix, Micron, and Samsung appear in both—but the weight structures are completely different.

KMEM is essentially a concentrated bet on the "three major memory giants," and its volatility will closely follow the stock prices of these three companies.

DISK, on the other hand, places its bets on NAND and storage, with SanDisk and Kioxia having a greater impact on net asset value, while traditional memory giants like Samsung and SK Hynix are merely supporting players.

How should ordinary retail investors think about this?

First, an easily overlooked fact: both funds only started trading a couple of days ago, with no historical performance to reference. Their size, liquidity, and bid-ask spreads are all unknown at this point.

Newly issued sector-themed ETFs often have higher trading costs in the first few months—they are not something you can blindly jump into.

Second, both funds are essentially highly concentrated sector bets, completely different from broad-based index funds.

Many people already directly hold stocks of Micron, Samsung, or SK Hynix, or have already bought the DRAM memory ETF. Adding another DISK or KMEM with 80% concentration means stacking additional positions in the same group of companies, effectively taking on more risk than they might realize.

Position sizing should be recalculated based on this logic, rather than simply buying a bit more.

When a sector’s wind comes, supporting financial products always pop up first to let retail investors jump in with one click.

The tool itself is neither right nor wrong—it solves the "how to buy" question. How much to buy and whether to buy at all still depend on one’s own clear calculation of risk.