Recently, I’ve been looking into the memory sector, and I found that many people still don’t understand which memory stocks are available and how to classify them. I’ve put together this industry chain, and I’ll also talk about why this kind of stock is so volatile.

First, it’s important to understand that the memory industry is divided into three tiers. The top tier is chip manufacturers, such as Nanya Technology and Winbond—companies that directly produce chips. When the business cycle is good, their gains are the biggest, but when things turn, their losses are also the most severe. The middle tier consists of companies that control ICs and modules, such as Phison and Adata. Because they handle software integration, their profits are comparatively more stable. At the very top are global giants such as Micron, Samsung, and SK Hynix. They control more than 94% of the global DRAM market, so the pricing power is in their hands.

When it comes to which memory stocks are worth paying attention to, I think you should look at it from two angles. In the U.S. market, Micron Technology is the most “pure-play” target: it does both DRAM and NAND. At present, HBM capacity is expanding, and overall profitability is in a recovery phase. SK Hynix is leading in the HBM space—HBM3e and HBM4 have already entered mass production, allowing it to directly capture the benefits of AI computing demand. Although Samsung has the largest market capitalization, it is relatively more complex as an investment target.

In the Taiwan stock market, Nanya Technology is the most pure-play DRAM concept. Its customized AI memory has already started contributing to revenue. Winbond follows a niche strategy, avoiding the price-cutting competition faced by general-purpose DRAM. Phison is the purest in NAND Flash: the current supply gap is still close to 20%, so in the short term it’s hard to change the situation of supply being tighter than demand. Macron mainly makes NOR Flash and ROM, with advantages in automotive and industrial applications, making it suitable for balancing the fluctuations of the memory cycle.

Why are memory stocks so volatile? The core reason is that this industry has a cycle that it can’t escape: shortage → capacity expansion → oversupply → price collapse → production cuts → and then back to shortage again. Nomura Securities’ latest forecast shows that in Q2 2026, DRAM and NAND prices will rise quarter over quarter by 51% and 50%, respectively. This increase far exceeds earlier expectations. In addition, memory manufacturers need to invest hundreds of billions of dollars to build wafer fabs. If the investment timing is wrong, by the time the capacity comes online, the market may have already reversed.

What’s happening now is that inventories at global memory original equipment manufacturers are at historical lows. Some large manufacturers have total inventories down to only about 4 weeks. That’s exactly why prices are prone to going up but difficult to fall. Although Samsung, SK Hynix, and Micron are expected to see a surge in 2026 earnings, they are all “pressing the brakes,” controlling capital expenditures in an attempt to prevent a supply glut around 2027.

The logic of trading memory stocks is actually very simple. If you can figure out which phase of the cycle the industry is in right now, you can capture the rhythm. When DRAM contract prices stop falling, leading manufacturers begin cutting production, and inventory days decline from their peak, it’s usually a good time to set up positions. Memory stocks don’t make money only from a company’s own growth—they profit from the timing rhythm of the business cycle.

My advice is to spend time observing the trend of DRAM contract prices, track the financial reports and capital expenditure directions of major memory manufacturers, and practice judging which phase the cycle is currently in. After you have a clearer grasp of the cycle, consider operating with small capital. Currently, memory prices are still trending upward, and the supply-side tightness is unlikely to ease in the short term. At this stage, holding related targets on the manufacturing and module side still has upward momentum.