Lately, I've been focusing on the memory sector and noticing that many people actually don't understand the logic behind it. Even though they are all memory concept stocks, their positions in the supply chain are completely different, and their ways of making money vary greatly.

In simple classification, the memory industry is divided into three layers. The top layer is the leading companies that directly produce chips—such as Nanya Technology, Winbond, and Macronix. They have the greatest profit elasticity, but also the most volatility, with the strongest cyclical impact. The middle layer consists of companies controlling ICs and modules, like Phison and Adata. Their moat comes from software integration, making their profits relatively more stable. The topmost layer includes the three global giants—Micron, Samsung, and SK Hynix. These giants control over 94% of the global DRAM market share, holding the pricing power.

Recently, I’ve noticed a phenomenon: when these international giants fully invest their capacity into AI-required HBM, Taiwanese manufacturers are instead able to benefit from a lot of order transfers. That’s why this year’s performance of DDR5 concept stocks is particularly worth paying attention to.

Speaking of volatility, memory stocks are truly caught in an endless cycle—shortage → capacity expansion → oversupply → price collapse → production cuts → shortage again. This cycle repeats every few years. Nomura’s latest forecast shows that in the second quarter of this year, DRAM and NAND prices increased by 51% and 50% quarter-over-quarter, respectively, much higher than the previous estimates of 6% and 20%.

Why is the volatility so high? There are three main reasons. First, this is a highly cyclical industry that cannot escape the cycle. Second, building a memory wafer factory costs hundreds of millions of dollars; if the investment timing is wrong, the market may have already reversed by the time capacity comes online. Third, the global memory market is highly oligopolistic, and decisions by a few companies can determine the entire price cycle.

If you want to pick U.S. stocks, Micron is my top choice. It is the only U.S. company with large-scale DRAM and NAND manufacturing capabilities. As HBM capacity continues to expand, overall profits are clearly recovering. SK Hynix is also good; it is the global leader in HBM shipments, with HBM3e and HBM4 already in mass production, directly benefiting from the explosive demand for AI computing power.

In Taiwan stocks, Nanya Technology is the purest DRAM concept stock. It focuses on DRAM manufacturing, with AI applications becoming a main growth driver. Customized AI memory products have already started contributing to revenue. Winbond adopts a niche strategy, focusing on special DRAM and NOR Flash, with stable layouts in consumer electronics, industrial, and automotive fields, resulting in relatively smaller profit fluctuations. Phison is one of the companies with the highest purity in NAND Flash; currently, the NAND supply gap is still close to 20%. Plus, AI inference brings nearly unlimited data storage needs, making it difficult to change the supply-demand imbalance in the short term.

The particularly noteworthy point about DDR5 concept stocks is the rapid increase in DDR5 penetration rate. Lianqi Technology specializes in DDR5 and HBM memory buffer chips, holding a certain degree of quasi-monopoly in this field, with clear growth momentum.

For trading strategies, I suggest two approaches. If you have a high risk tolerance, consider swing trading—gradually positioning at the cyclical bottom and gradually exiting when prices surge 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 stocks still has upward potential.

If your risk tolerance is low, wait until memory stocks have fallen sharply before entering. The cyclical nature of memory stocks means they tend to fall deeply during downturns, and the industry bottom is often the best entry point. Keep an eye on three things: the trend of DRAM contract prices, when supply chain inventory days turn from high to low, and whether major manufacturers are cutting capital expenditures.

Finally, I want to emphasize that memory stocks are not fundamentally stable growth stocks but cyclical trading assets. Your focus should not be on finding companies to hold forever but on understanding where the cycle currently stands. The memory stocks that fell sharply last cycle became big dark horses this round due to the AI supply gap. Memory stocks earn from rhythm, not just the companies themselves.

If you haven't started paying attention to this sector yet, now is a good time. Open a demo account, spend a few weeks observing DRAM contract price trends, tracking the financial reports and capital expenditure plans of major memory manufacturers, and practicing judging where the memory cycle is at. Once you have a clearer grasp of the cycle, consider using small capital for real trading.