Author: Wang Chuan
This article is a continuation of Wang Chuan: When the neighbor Lao Wang earned thirty times from investing in memory storage stocks, how can you still avoid anxiety (5) - The bullwhip effect.
1/ Stocks of memory storage companies—this is something that can easily bring inexperienced youths to a rapid climax. Over thirty years ago, the primary medium for removable storage in the computer industry was the 3.5-inch, 1.44 MB floppy disk. In late 1994, a company called Iomega introduced a removable hard drive with 100 MB capacity, known as the Zip drive, priced at $199. For consumers needing to back up and transfer large files, the Zip drive solved a major problem.
2/ Iomega's sales skyrocketed from $140 million in 1994 to $1.21 billion in 1996. Iomega's stock also soared from about $2 per share at the end of 1994 to the equivalent of $330 per share (accounting for stock splits) in May 1996—a return of over 160 times in a year and a half. I recall a sincere sigh from a netizen on a tech stock forum back then: "This is better than sex!"
3/ After May 1996, Iomega's stock declined continuously. By the end of 1999, its price had fallen more than 85% from its 1996 peak. Ultimately, in 2008, EMC acquired Iomega for $210 million—a 97% drop from its $7 billion peak market cap in 1996. What does a 97% drop mean? It means that after an 85% drop, some speculators think it's a bargain and jump in, only to see it fall another 80%.
4/ The main theories of the steadfast bulls in Iomega back then were: 1) Until 1996, potential competitors appeared weak and expensive. 2) There was a genuine possibility that the Zip drive could become a standard PC component, just like the floppy disk drive. If this could attract hundreds of millions of users, with each Zip disk generating over ten dollars in profit, the future would be limitless. Iomega's stock became the first meme stock of the internet era to gain widespread popularity among retail investors, attracting massive capital inflows and creating a self-reinforcing trend that buried many short sellers.
5/ Iomega's fall was more complex than outsiders imagined. The second half of 1996 saw only a simple price correction, with competitors still barely visible. Revenues in 1997 were $1.74 billion, but growth had noticeably slowed. CD-R, as a potential competitor, was already cheaper than Zip disks for high-end users. By 1998, CD burner prices were close to Zip drives, but a CD cost less than a dollar—much cheaper than a Zip disk—completely eroding Iomega's moat. In 1998, its revenue only fell 3% from 1997, but gross margins dropped from 31% to 25%, turning profits into losses. The story was over.
6/ Storage industry products, epitomized by Dynamic Random-Access Memory (DRAM), are among the most homogeneous products in the tech sector. Homogeneity means no brand premium; prices change rapidly with global supply fluctuations. Historically, DRAM chip prices have fallen by over 80% within short periods at least six times: 1985, 1998, 2001, 2009, 2012, 2023, with numerous other instances of 30% to 50% price drops. The stock price drops for memory companies are even more brutal than chip price drops, with 95% declines or bankruptcies being commonplace. Micron's stock price in May 2025 was equivalent to its price in June 2000, losing a full 25 years. Four major bankruptcy cases in the storage industry over the past thirty years include: 1) Mostek, 1986. 2) Qimonda, 2009. 3) Spansion, 2009. 4) Elpida, 2012. As for countless other bankrupted smaller companies, they are too numerous to count.
7/ The essence of the storage industry is elastic demand facing capital-intensive, long-cycle, inelastic supply. When storage prices are too high, elastic demand naturally retreats, and ways are found to circumvent it. But once rigid supply increases 18 months later, capacity must run at full throttle, and products must be sold immediately regardless of price to maximize benefits. Once rigid supply becomes even slightly oversupplied relative to elastic demand, price drops occur immediately, sometimes very violently.
8/ The phenomenon of soaring stocks across the entire storage industry starting September 2025, in hindsight, was essentially due to cloud service providers' demand for AI chips consuming various types of memory, especially High Bandwidth Memory (HBM), crossing a certain threshold. To lock in capacity for 2026 and 2027, cloud service providers were willing to accept significantly higher memory prices. As long as one or two buyers were sufficiently frantic, other competitors were forced to follow. The panic of shortages quickly spread to smaller buyers and the consumer electronics sector. Major memory manufacturers, having learned from previous sharp price drops, did not rush to increase capacity. Instead, they opted for steep price hikes, exploiting the fact that time was temporarily on their side to make a hefty profit.
9/ For the flash memory manufacturer SanDisk, production costs in Q1 2026 (non-fiscal year) were $1.288 billion, compared to $1.313 billion in the same period last year—meaning this year's production costs actually decreased by about 2%, with the volume of storage produced remaining largely unchanged. However, Q1 2026 revenue was $5.95 billion with a gross margin as high as 78.3%, whereas a year ago in the same period revenue was $1.695 billion with a gross margin of only 22.5%. Therefore, the 251% revenue growth mainly came from corresponding storage price increases, not from selling more goods.
10/ Why haven't more goods been sold, yet prices have more than doubled? Because demand suddenly surged, but supply is rigid. In the short term, there's only so much available, so everyone scrambles for the limited supply, driving prices up. This implies a potentially counterintuitive future phenomenon: When rigid supply finally increases and rebalances with demand, flash memory prices and gross margins will inevitably return to previous levels. At that point, although sales volume will increase, total sales revenue and net profit may actually decrease. The more they sell, the less they earn.
11/ Similarly, for Micron in the quarter from November 1, 2025, to the end of February 2026, operating costs were $6.1 billion, only up less than 20% from $5.09 billion a year ago. However, sales revenue was $23.86 billion—nearly triple the $8.05 billion from a year ago. Gross margin was 74.4%, compared to only 36.8% a year ago. Micron's product line includes HBM, DRAM, and flash memory. Capacity constraints and price changes are more complex than SanDisk's, but they cannot escape the same underlying logic.
12/ For homogeneous commodities, high margins themselves destroy high margins, and high prices themselves erode marginal demand. Looking at gross margins above 70%, major memory manufacturers are no longer idle. Starting in 2026, they will invest tens of billions of dollars to increase capacity, but significant new capacity is expected to come online only by the second half of 2027.
13/ Bulls on storage stocks might say, haven't storage companies started signing long-term agreements with customers to lock in capacity prices? Doesn't that mitigate the risk of price collapse? The truth is, the more unstable a relationship is, the more people love to sign long-term agreements. The reason for signing long-term agreements is that temporarily, everyone desperately needs them to avoid the worst-case scenario, creating a false but very fragile sense of security. But once circumstances change substantially, the stronger party under the new situation will generally find some excuse in the agreement to immediately renege. The so-called long-term agreements in the storage industry generally do not exceed five years. Once new storage capacity comes online and spot prices fall below long-term agreement prices, buyers can exploit various loopholes in the agreement to make memory companies immediately share the pain of falling spot prices. Even if the contract is legally airtight, buyers can threaten to shift more business to competitors who are more favorable to buyers once the agreement ends. At that point, memory companies generally make immediate concessions for long-term interests. The enforceability of long-term agreements signed by memory manufacturers and customers is probably similar to the effectiveness of the 1939 Molotov–Ribbentrop Pact (non-aggression treaty between the USSR and Nazi Germany). When everyone senses risk and rushes to use formal agreements to guard against it, don't naively assume the risk is mitigated. This is precisely a signal that risk is increasing.
14/ There is also an asymmetry here: When all players are making big money, it only takes one new entrant who doesn't care about short-term economic gains and can sustain long-term losses to change the supply-demand relationship. It only takes one new technological breakthrough to significantly reduce demand. You cannot predict beforehand exactly which factor will directly alter the supply-demand balance. You just need to know that the risk of storage price drops is now highly asymmetric compared to the possibility of continued sharp rises. These risk factors include but are not limited to: 1) Economic recession due to rising interest rates and inflation. 2) Cloud service providers cutting back on AI capital expenditures. 3) New storage capacity coming online faster than expected, especially from Chinese companies like CMXT and YMTC, who are exceptionally adept at scaling up capacity at any cost. 4) New AI chip designs, model architectures, and software algorithms that can significantly reduce memory requirements. In fact, precisely because of the soaring prices of storage, all the smart people in the world are racking their brains at various levels—chip design, model architecture, software algorithms—to actively reduce dependence on storage.
15/ The storage industry, like other industries with severe homogeneity, has another deadly trap: at the peak of the cycle, product profits are extremely high, but company P/E ratios are often very low, sometimes even single-digit. It looks like a good value investment, but this is actually when the risk is greatest. Because once commodity prices crash, original profits quickly shrink or even turn into losses, making the low P/E ratio meaningless. There are always many simple-minded investors who, tempted by low P/E ratios, invest their life savings into this enormous wealth incinerator during the industry's downturn without thinking.
Neighbor Lao Wang is still immersed in the sweet dream of getting rich effortlessly through memory storage stocks and being able to exit unscathed in the future. Don't disturb him. To find out what happens next, stay tuned for the next installment.
(To be continued)
All articles represent the author's personal opinions and are for reference only. They do not constitute investment advice for the mentioned assets. Investment involves risks; you should exercise caution when entering the market.
