Memory module prices plummeting, memory stock prices falling across the board, has the memory super cycle peaked?

Is the memory price crash a sign of a repeat of the traditional cycle?

Recently, the memory prices that had been rising for several months suddenly turned downward, sparking market concerns that the memory cycle has peaked.

According to market tracking data, many US retailers are offering widespread discounts on DDR5 memory, with the highest price drops of up to $100 per kit. For example, Corsair’s VENGEANCE series, with 32GB capacity and a maximum 6400MHz frequency model, is now priced around $379.99, a significant drop from the recent peak of $490, with a single kit falling more than $110.

The domestic market is also affected, with some wholesalers telling China Business Journal that mainstream 16GB memory sticks are “dropping over a hundred yuan a day,” and large stockpilers are frantically selling off inventory.

“Since last Saturday, prices have collapsed,” said Wang, a wholesaler who has been operating in storage devices at Bai Nao Hui for years. He showed a drastic price curve for a mainstream 16GB 3200MHz memory stick: “Last May, it was only over 130 yuan, then skyrocketed to a peak of 980 yuan in December, but after months of high fluctuation, the current spot price has fallen back to around 700 yuan.”

Wang expressed helplessness, saying that the price surge had overdrawn consumer expectations: “If it’s not a necessity, people won’t buy. Compared to before November last year, our sales have dropped by more than 60%.”

Meanwhile, Google released a paper titled “TurboQuant,” describing a new compression algorithm. The study states that this technology can reduce the runtime memory footprint (KV Cache) of large language models by at least 60%. Investors quickly priced this in: the AI hardware shortage problem will be fundamentally alleviated, and memory demand will be greatly reduced.

The cold snap in the spot market has rapidly spread to the capital markets. Micron’s stock price has retreated over 24% from recent highs, and Western Digital has fallen nearly 21% from its peak of $777.60. At the same time, last week, the market capitalization of the US memory chip sector evaporated nearly $100 billion.

Faced with plunging prices and collapsing stock prices, market participants are deeply divided on the outlook for the memory industry. Some investors believe that the traditional “pig cycle” of memory has peaked, while HSBC analysts argue that market fears are overblown, and that we are currently in the middle of an AI-driven super cycle for memory, with strong demand for high-end products like HBM, and that memory shortages could last one to two years.

Buyers say “No”: Is the traditional “pig cycle” repeating?

For traders following the traditional cycle, the market plunge is not so simple. Former journalist and well-known semiconductor analyst Dan Nystedt pointed out that many bullish investors attribute the recent crash to Google’s paper, but this is only superficial. Dan believes the real reason is that some smartphone memory chip prices have stopped rising.

“The real reason is much simpler: the prices of certain smartphone memory chips have stopped increasing. Buyers ultimately said ‘no’—this is the first top signal that experienced memory cycle investors look for before selling.”

Dan Nystedt said. Due to high DRAM and NAND prices, some smartphone manufacturers plan to cut or even cancel mid- and low-end phone production in 2026. He revealed that two weeks ago, a buyer refused higher DDR4 prices.

Dan Nystedt compares the memory industry to the “pig cycle” in agriculture: high prices encourage companies to expand capacity, but building factories takes time, and when new capacity is released simultaneously, prices collapse. He believes investors following this script have already exited quickly, causing Micron and SanDisk’s stock prices to plummet.

Over the past 50 years, memory chips have experienced more than a dozen major boom/bust cycles. Since 2010 alone, there have been three: 2012-2015 with 3G/4G and cloud computing booms; 2016-2019 with 5G and cloud service expansion; and 2020-2023 driven by pandemic-fueled PC/server surges. The cycle starting in 2024 is driven by AI servers (HBM and SRAM).

“Whenever someone writes ‘this time is different,’ it’s usually a classic sign of bullish sentiment going into madness,” Nystedt quoted legendary trader Jesse Livermore: “The market is always right, and opinions are often wrong.” He warned investors that when chip buyers stop panic buying and rebounds are repeatedly met with sustained selling, seasoned funds will quickly retreat according to the script.

Structural change: Are memory companies no longer “cyclical stocks”?

However, Jukan, an independent analyst, offered a different view on Dan Nystedt’s analysis.

He pointed out that buyers’ resistance to price increases mainly focuses on traditional memories like DDR4, not the entire memory market. Previously, the abnormal surge in DDR4 prices was partly due to stockpiling in China, which gave smartphone manufacturers room to adjust low-end device specifications.

“But DDR5 is a whole different story,” Jukan said. Smartphone and PC manufacturers have, in the first and even second quarters of this year, quietly accepted significant price increases for DDR5. In the current AI and high-end device ecosystem, DDR5 is not a bargaining chip but a core input that buyers must pay a premium for. Flagship products built around DDR5 cannot simply lower specifications.

Second, the market is ignoring the fundamental transformation in the business models of memory giants. Jukan scoffs at the so-called “seasoned investors” who blindly sell off when spot prices fall.

“Memory companies’ operating models are no longer the old reckless expansion,” Jukan pointed out. Samsung, SK Hynix, and Micron are moving toward the TSMC business model—building capacity only after securing prepayments (advance payments) and long-term demand visibility from key customers.

Recent reports from Korean media indicate that Samsung is exploring prepayment-based cooperation agreements with giants like Microsoft. Memory giants are well aware of how destructive overcapacity can be to cycles, so they now pursue extreme restraint in capacity expansion rather than reflexively overbuilding.

Investment banks support: The super cycle for memory is only halfway through, and the market’s five major concerns are overblown

Contrary to the panic in the spot market, investment banks remain confident in the long-term prospects of the memory industry. HSBC’s March 30th research report explicitly states, “In our view, current concerns are exaggerated; we are in the middle of a super cycle driven by AI.”

The bank lists five specific concerns that it considers overreactions:

1. Negative impact from raw material and electricity price increases due to Middle East conflicts;

2. Slowing growth in memory prices in the second half of 2026;

3. Industry technologies like Google’s “TurboQuant” and Nvidia’s “KVTC” that reduce memory usage in AI systems;

4. Gradual increase in capital expenditure plans by major memory manufacturers;

5. Intensified competition from Chinese memory producers.

The report states that the Middle East conflict has not materially affected raw material procurement for memory manufacturers. The absolute profit growth will have a much greater impact on stock prices than the slowdown in DRAM price increases. Meanwhile, memory manufacturers remain highly disciplined and cautious in executing capital expenditures.

Regarding Google’s TurboQuant technology, which has triggered market sell-offs, the bank believes it is premature to worry now. Commercialization will take about a year, and its reference parameters are smaller than the current AI environment. More importantly, the report notes that TurboQuant will alleviate memory bandwidth bottlenecks, improve system efficiency, and reduce token costs, thereby accelerating AI commercialization and adoption. It states:

“The net effect is that we believe efficiency improvements will accelerate AI development—this is a positive event that should trigger a sharp increase in AI adoption.”

The bank also forecasts that AI server shipments will surge 28% year-over-year by 2026. From 2026 to 2027, average DRAM capacity per server will grow strongly by 17%. As AI inference demand explodes, enterprise SSDs (eSSD) are entering a golden era. The report estimates that by 2027, eSSD will account for 40% of NAND demand, up from 18% in 2023, with AI servers consuming 62% of that.

The bank believes that the current market is in the middle of an AI-driven super cycle, comparable in scale to the six-year DRAM shortage triggered by office automation from 1990-1995. Historically, from 1990 to 1995, the proliferation of Windows 3.0 and subsequent OS led to a long six-year structural DRAM shortage, causing the market size to surge from $7 billion in 1990 to $41 billion in 1995—a sixfold increase.

The bank argues that the infrastructure built today by large models, intelligent agents (Agentic AI), and physical AI (like autonomous driving) will cause a memory shortage that will last at least one to two years.

Based on these assessments, the report remains confident in their beneficiaries during the memory super cycle. Regarding the recent plunge, it states: “We believe any correction offers an additional buying opportunity.”