Recently, I’ve been looking into investment opportunities related to memory stocks and noticed a very interesting phenomenon. Many people treat memory concept stocks as a broad category, but in reality, their roles in the industry chain are vastly different, and their profit logic is completely distinct.

In simple terms, memory stocks are divided into three levels. The first layer consists of companies that directly manufacture chips, like Nanya Technology and Winbond. They have the greatest profit elasticity but also the most volatility. The second layer includes companies that control ICs and modules, such as Phison and Adata. Their profits are usually more stable because they handle software integration. The third layer comprises international giants like Micron, Samsung, and SK Hynix. These three companies monopolize over 94% of the global DRAM market, holding the pricing power.

Recently, I’ve noticed an important shift. The current ranking of the world’s major memory manufacturers is as follows: Samsung leads with a market cap of $897 billion, far ahead of SK Hynix and Micron. Among them, SK Hynix is particularly strong in HBM (High Bandwidth Memory) because of its close ties with NVIDIA. Micron is the only US company capable of large-scale production of DRAM and NAND, and its recent financial reports have repeatedly hit new highs.

Why do memory stocks fluctuate so much? There are three core reasons. First, it’s a highly cyclical industry—shortages → expansion → oversupply → price crashes → cutbacks → shortages again—this cycle repeats every few years. Nomura Securities’ latest forecast predicts that in Q2 2026, DRAM and NAND prices will increase by 51% and 50% quarter-over-quarter, respectively, significantly upwardly revised from previous estimates of 6% and 20%. Second, building a memory wafer factory costs hundreds of millions of dollars, and if the timing of investment is wrong, the market may have already reversed by the time capacity comes online. Third, due to market concentration in a few companies, their decisions can determine the entire price cycle.

Talking about core US storage stocks, Micron is the most pure-play, producing both DRAM and NAND, making it the most resilient storage stock in the US market. As HBM capacity continues to expand, overall profitability is clearly recovering. SK Hynix is the global leader in DRAM and HBM shipments; HBM3e and HBM4 are already mass-produced, directly benefiting from AI high-computing demands. Kioxia, originally Toshiba Memory, is a major NAND Flash manufacturer, and its market cap has skyrocketed from 43rd to 10th place globally within half a year—an astonishing surge.

In Taiwan stocks, Nanya Technology is the purest DRAM concept stock, with revenue already contributed by customized AI memory products. Winbond focuses on niche DRAM and NOR Flash, avoiding the price competition of general-purpose DRAM, resulting in relatively smaller profit fluctuations. Phison is the most specialized NAND Flash company in Taiwan; current NAND supply shortages are still close to 20%, and AI inference is driving nearly unlimited data storage needs. Macron is deeply involved in NOR Flash and ROM, with technological advantages in automotive and industrial fields, suitable for balancing the volatility of the memory cycle.

Here’s a crucial distinction: the logic of memory stocks and AI stocks is completely different. Memory stocks are cyclical trading assets, while AI stocks are trend investments. Memory stocks profit from the rhythm of the economic cycle, whereas AI stocks profit from structural growth dividends. You need to judge which stage of the cycle memory stocks are in now, rather than looking for companies to hold forever.

If you want to trade memory stocks, the current environment is still favorable. Memory prices are still rising, and the tight supply situation is unlikely to ease in the short term. Related manufacturing and module stocks still have upward momentum. But if you have a lower risk tolerance, you can also wait for memory stocks to fall sharply before deploying. The industry bottom is often the best entry point.

What are the three most important signals to watch? First, a halt in DRAM price declines, which is the most critical indicator of an industry turning point. Second, companies start to cut production; although Samsung, SK Hynix, and Micron are experiencing booming earnings, they are all slowing capital expenditures, indicating attempts to control capacity to prevent oversupply around 2027. Third, inventory levels are decreasing; currently, global memory manufacturers’ inventories are at historic lows, with some major companies holding only about four weeks’ worth of stock, which is a direct reason prices are more likely to rise than fall.

The essence of memory stocks is cyclical trading assets. Your job isn’t to find companies to hold forever but to judge which phase of the cycle we are in. The memory stocks that plunged deeply last cycle have become big winners this time due to the AI supply gap. Essentially, you’re making money from the rhythm, not the company itself.