Memory chip stocks 2026: why have they soared and can the rally last?

Memory chips have always been a boom-and-bust business, but this year's move looks different in scale. Samsung Electronics said this week that it expects second-quarter operating profit of approximately 89.4 trillion won (around $59 billion) — roughly 19 times higher than a year earlier. SK Hynix's American Depositary Shares (ADS) begin trading on Nasdaq on 10 July, in a deal expected to raise around $28 billion, one of the largest share sales on record after only SpaceX's $85.7 billion debut in June.

The question worth asking is not whether memory is a cyclical industry, but whether this particular cycle is structurally different.

Why memory and storage matter: DRAM, HBM, NAND and HDD explained

Memory chips store data and feed it to the processor as needed. They sit inside nearly every piece of modern electronics — phones, laptops, cars, gaming consoles and, increasingly, the servers that train and run artificial intelligence (AI) models. The major types of memory and storage devices are:

• Dynamic random-access memory (DRAM) – a computer's working memory. It temporarily holds the data a processor is actively using, then clears when the device powers off
• High-bandwidth memory (HBM) – multiple DRAM dies stacked vertically and connected by thousands of microscopic channels, positioned next to the processor. This moves far more data, far faster — exactly what AI accelerators need, given the huge data volumes they process
• Not and (NAND) – flash storage that keeps data long-term, used in products such as solid-state drives (SSDs) and USB sticks, with no moving mechanical parts
• Hard-disk drive (HDD) – also long-term storage, but uses mechanical magnetic platters to read and write data. It's slower than NAND but cheaper per unit of storage

How AI is reshaping the memory industry: HBM demand vs. DRAM supply squeeze

Data centres are expected to consume more than 70% of the high-end memory chips manufacturers produce in 2026, according to TrendForce. This surging demand has reshaped manufacturers' production strategies.

HBM and conventional DRAM share the same fabs and silicon wafers, so they compete for the same capacity. Because HBM commands significantly higher prices and margins, manufacturers have prioritised it wherever production lines allow — creating two different stories from the same cause: a demand expansion story for HBM, and a supply shortage story for conventional DRAM.

Manufacturers cannot simply ramp up HBM output to close that gap. The stacking process demands extreme precision: wafers are thinned to a fraction of normal thickness, and thousands of microscopic connections per stack must all work, since a single defect ruins the entire unit. That precision is difficult to replicate consistently at scale, so yields remain lower than for conventional DRAM — part of why supply hasn't caught up despite years of investment.

AI is reshaping demand for NAND and HDDs too. Although neither can match the speed HBM gives an AI accelerator, AI data centres still generate enormous demand for storage elsewhere: housing training data, model weights and the results of inference jobs. NAND captures much of that through fast enterprise SSDs, while HDDs pick up the cheaper, bulk 'cold' storage that doesn't need to be retrieved instantly.

Key memory players: global market share across DRAM, NAND, HBM and HDD

Samsung, SK Hynix and Micron cut across DRAM, NAND and HBM, though each has a different focus. Everyone else in the table specialises in a single memory technology.

• Source 1: Counterpoint – DRAM and HBM market shares based on revenue, as of Q1 2026
• Source 2: Counterpoint – NAND market shares based on revenue, as of Q1 2026
• Source 3: Forbes – HDD market shares based on revenue, as of 2025

Isn't this industry always cyclical? What's different this time

Over the past two decades, the memory chip sector has followed a familiar pattern: prices spike when demand outruns supply, manufacturers rush to add capacity, that capacity arrives all at once, and prices collapse. Most recently, in 2022 – 2023, Micron and SK Hynix lost billions of dollars after overestimating how long pandemic-era demand would last.

The bull case for this cycle behaving differently rests on two structural points. First, Samsung, SK Hynix and Micron — the three companies that have long dominated DRAM — are showing more capital discipline than in past cycles, resisting the urge to race each other into speculative overbuilding. Second, AI-optimised memory is simply harder to produce: Bank of America (BofA) estimates it requires three to four times the production capacity of conventional memory per unit, and the new capacity needed to meet that demand faces physical limits — cleanroom space, power and water supply — that cannot be solved simply by spending more.

Following Micron's latest results, BofA pushed its estimated timeline for the super-cycle out to the end of 2027, with a scenario extending as far as 2030. Multiple data points support this view: Micron's HBM capacity is sold out through 2027; Kioxia confirmed in January that its entire 2026 NAND output had already been committed, with some hyperscale clients requesting supply agreements stretching into 2027 and 2028; and SK Hynix's chairman has warned that global memory supply is likely to remain roughly 20% below demand through 2030.

We do not think the industry's underlying cyclicality has disappeared, but the mechanism behind it has changed. Long-term agreements now lock in pricing and allocation years in advance rather than quarter to quarter, giving buyers and sellers more visibility than in past cycles, when a shortage could reverse within a few quarters. That doesn't make memory a one-way trade — it makes the eventual turn slower to arrive, and potentially harder to time.

Memory and storage stocks dwarf broader tech market in 2026

Memory and storage stocks have significantly outpaced both the semiconductor sector and the wider tech market this year. Even after the recent pullback, Kioxia and SanDisk are still up close to 600% year-to-date.

Valuations still look inexpensive on forward price-to-earnings (P/E) for most memory chip makers, but that is conditional on the market's conviction that the sold-out capacity and multi-year contracts described above convert into delivered revenue. Importantly, this year's stellar performance has been driven largely by earnings growth rather than multiple expansion, which is why forward valuations remain low despite such large share-price gains.

High profit margins reflect genuine pricing power: TrendForce recorded conventional DRAM contract prices rising 93 – 98% quarter-on-quarter in the first quarter of 2026, followed by a further 58 – 63% in the second quarter, while NAND flash contract prices climbed 85 – 90% and then 55 – 60% over the same two quarters.

Hard-disk drive names tell a different story. Western Digital's forward P/E ratio has more than doubled over the past year, from around 12x to about 29x, and Seagate shows a similar picture. Unlike the memory chip makers, this less-cyclical corner of the industry has re-rated as investors price in adjacent AI-driven demand for bulk data-centre storage.

Stock performance and financial metrics

What to watch: capex spends, Apple's China chip talks, ETF risk

Manufacturers are responding with a historic expansion in capital spending. South Korea's government has announced a ten-year plan to 'secure the core elements of AI faster than any other country.' Key projects include 800 trillion won (around $530 billion) of investment from SK Hynix and Samsung to build two new chip fabrication sites in the country's south-west. Micron has also launched a $200 billion programme covering fabrication plants and research. None of this new capacity reaches volume production before 2027 — supporting pricing in the near term, but also meaning several large expansions land in the market at roughly the same time in 2028 and beyond, the point at which the sector's current supply discipline will face its real test.

Apple's supply chain shows how severe the squeeze has become. The company has been in talks to source memory from CXMT and YMTC — both flagged by the Pentagon for alleged ties to the Chinese military. If the deal goes ahead, it would inject fresh competition against the incumbent players, while handing CXMT and YMTC a marquee customer just as both prepare public listings later this year.

For traders, there is also a risk worth flagging: leveraged and single-stock exchange-traded funds (ETF) tracking South Korean memory names have amplified this year's rally and would just as readily amplify a reversal. Even if the structural story holds, the path there is unlikely to be smooth.

The figures stated in this article are based on a snapshot taken on 7 July 2026 unless otherwise stated. Past performance is not indicative of future results.