Recently researching investment opportunities related to memory, I found that many people have a somewhat confused understanding of this field, especially not knowing how to select stocks. Today, I want to talk about the concept stocks of memory modules and the logic of the entire supply chain.

Honestly, the reason why memory stocks are so volatile is mainly because this industry is highly cyclical. Shortage → Capacity expansion → Oversupply → Price collapse → Production cuts → Shortage again, this cycle repeats every few years. Nomura Securities' latest forecast shows that in Q2 2026, DRAM and NAND prices will increase by 51% and 50% quarter-over-quarter, respectively. This upward revision is significantly higher than previous predictions, indicating that the shortage is indeed more severe than expected.

The memory supply chain is actually divided into three layers. The first layer is the leading companies directly producing chips, like Nanya Technology, Winbond, and Macronix. These companies have the greatest profit flexibility but also the highest risk, as they are most affected by the economic cycle. The second layer includes companies controlling ICs and modules, such as Phison, Adata, Innodisk, and Transcend. These memory module concept stocks tend to have more stable profits because they control software integration and system protocols, creating a deeper moat. The third layer consists of global giants like Micron, Samsung, and SK Hynix, which dominate over 94% of the global DRAM market and have the strongest pricing power.

I think the key is to understand your own risk tolerance. If you want to profit from price differences, focus on chip manufacturers; if you prefer more stability, memory module concept stocks are safer; if you're looking at long-term trends, invest in U.S. giants.

The current global market landscape is quite clear. Samsung leads with a market cap of $897 billion, far ahead. SK Hynix has the largest shipment volume in the HBM (High Bandwidth Memory) market. Micron is the only U.S. company with large-scale DRAM and NAND manufacturing capabilities. Kioxia (formerly Toshiba Memory) has performed extremely well in the past six months, jumping from the 43rd to the 10th largest market cap, mainly benefiting from the explosive demand for high-end SSDs in AI data centers.

There are also many opportunities in Taiwan stocks. Nanya Technology is Taiwan’s purest DRAM stock, with AI applications becoming a major growth driver. Winbond focuses on niche DRAM and NOR Flash, avoiding the price wars common in general-purpose DRAM, resulting in relatively stable profits. Phison is the most specialized NAND Flash company; currently, the NAND supply gap is close to 20%. Plus, AI inference drives unlimited data storage needs, making the short-term supply-demand imbalance unlikely to change. Macronix specializes in NOR Flash and ROM, with technological advantages in automotive and industrial fields, suitable for balancing the volatility of the memory cycle.

When is the best time to buy memory stocks? I think it depends on three signals. First, DRAM prices stop falling, which is the most important signal of an industry turning point. Second, leading manufacturers start reducing production, indicating supply is tightening. Third, inventory days decline from high levels, showing downstream demand is warming up. Currently, global memory manufacturers' inventories are at historic lows, with some major companies holding only about four weeks of stock, which is the direct reason why prices are easy to rise but hard to fall.

Memory stocks are essentially cyclical trading assets, not stable growth stocks. You need to judge not which company to hold forever, but where the cycle currently stands. The memory stocks that plunged last cycle became big black horses this round because of the AI supply gap. In simple terms, you profit from the rhythm, not the company itself.

I recommend opening a simulated account first, observing the upcoming DRAM contract price trends over the next few weeks, tracking the financial reports and capital expenditure trends of major manufacturers, and practicing judging which stage of the memory cycle we are in. Once you have a clearer grasp of the cycle, consider real trading with small capital. Memory module concept stocks and the entire industry chain are worth paying attention to now, but be sure to understand what you are betting on.