Recently, I have been researching memory-related stocks and found that the logic behind this sector is actually much more complex than I imagined. Although they are all memory concept stocks, companies play completely different roles in the supply chain, with significant differences in profitability and risk characteristics.

The memory industry can be roughly divided into three levels. The first level consists of leading companies that directly produce chips, such as Nanya Technology, Winbond, and Macronix. When market prices rise, their profit margins are the most elastic, but they are also most affected by economic cycles, with sharp rises and falls. The second level includes companies that control ICs and modules, like Phison and Adata, which usually have more stable profit performance, with their moat coming from software integration. The third level comprises international giants like Micron, Samsung, and SK Hynix. Although Taiwanese manufacturers are strong, the global market is essentially monopolized by these three.

Speaking of the world's major memory manufacturers, the latest ranking as of April 2026 shows Samsung with a market capitalization of $897 billion, far ahead of SK Hynix and Micron. Samsung holds about 45.5% of the DRAM market share, SK Hynix is the leader in high-bandwidth memory (HBM), and Micron is the only U.S. company with large-scale manufacturing capabilities for DRAM and NAND. Kioxia, originally Toshiba’s memory division, is also interesting; its market value has jumped from 43rd globally to 10th within half a year, entirely benefiting from the explosive growth in AI data centers.

Why do memory stocks fluctuate so much? Essentially, this industry has an unavoidable cycle: shortage → expansion → oversupply → price collapse → reduction → shortage 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 higher than previous estimates of 6% and 20%. Plus, building a memory wafer factory often costs hundreds of millions of dollars; if the timing of investment is wrong, the market may have already reversed by the time capacity comes online.

Furthermore, the global memory market is highly oligopolistic, with Samsung, SK Hynix, and Micron controlling over 94% of the DRAM market worldwide, holding the pricing power. The NAND Flash market is also dominated by five major companies, with a market share exceeding 80%. This concentrated supply means that decisions by a few companies can determine the price cycle.

On the U.S. stock side, Micron Technology is the most pure-play memory stock, producing both DRAM and NAND, making it the most resilient storage stock in the U.S. market. As HBM capacity expands, overall profitability is clearly recovering. SK Hynix is the global leader in DRAM and HBM shipments; HBM3e and HBM4 have already entered mass production, directly benefiting from the explosive demand for AI computing power. Kioxia, as the fourth-largest NAND manufacturer globally, has a NAND cycle lagging behind DRAM, with slower price increases, making it suitable for late-cycle rotation strategies.

In Taiwan stocks, Nanya Technology is one of the few listed companies focused on DRAM manufacturing, with customized AI memory products already contributing to revenue. Winbond specializes in niche DRAM and NOR Flash, with stable layouts in consumer electronics, industrial, and automotive fields, avoiding the price competition of general-purpose DRAM. Phison is one of the purest NAND Flash companies in Taiwan, with a NAND supply gap still close to 20%. The surge in AI inference creates almost unlimited data storage demand. Macronix focuses on NOR Flash and ROM, with technological advantages in automotive and industrial sectors; these embedded storage needs are relatively stable.

Regarding trading strategies, memory stocks are not stable growth stocks but cyclical trading assets. The core logic is to gradually accumulate at the bottom of the cycle and gradually exit when prices rise sharply and market sentiment overheats. Currently, memory prices are still rising, and the tight supply situation is unlikely to ease in the short term. Holding manufacturing and module-related stocks still has upward momentum.

If you want to judge when memory stocks are most worth buying, you can watch for several signals. First, see if DRAM prices stop falling, which is the most critical indicator of an industry turning point. Second, observe whether leading companies start reducing production; Samsung, SK Hynix, and Micron are expected to see explosive growth in 2026, but they are all pulling back on capital expenditures to control capacity and prevent oversupply in 2027. Lastly, check if inventory days have fallen from their highs; global memory manufacturers’ inventories are at historic lows, with some major companies holding only about four weeks of stock, which is a direct reason prices are easy to rise but hard to fall.

The essence of memory stocks is about timing, not the companies themselves. The last deep decline in memory concept stocks has turned into a big dark horse this round due to the AI supply gap. Judging where the cycle currently stands—whether shortage, expansion, oversupply, or reduction—is more important than constantly holding a specific company.