1/SK Hynix saw a sharp drop in the Korean stock market opening today, which is sure to make many people panic. Especially since the US stock market only opened tonight, the market will naturally associate it with: Will Micron follow suit and fall? Will SanDisk be smashed? Has the entire storage reached its limit? I don't think we can draw such a simple conclusion. It is easy to misjudge the trend of a single day. What really needs to be seen is: Is this emotion killing valuation, or have the fundamentals really changed. The biggest problem with this round of storage is that it rose too smoothly earlier. AI training requires HBM. AI inference requires DRAM. Data center expansion requires SSDs. NVIDIA's new chips continue to consume high bandwidth memory. This logic is correct. But once the stock price rises to a certain level, the market will definitely start asking: How long can HBM be tight? How much longer can DRAM prices continue to rise? Will cloud vendor CapEx slow down? When will AI computing power be converted into profits? So the sharp decline of Hynix does not necessarily mean that the market is rejecting AI. It's more like the market is re examining the pricing of storage. The storage industry has a very cruel characteristic: When it rises, everyone treats it as a growth stock. When it falls, everyone remembers that it is a cyclical stock. The previous discussion was about HBM shortages, price increases, and Nvidia's supply chain. Once the market starts to worry about supply and demand loosening, customer order cuts, and capacity release, valuation will be immediately suppressed. This is the difference between storage and pure software companies. No matter how advanced it is, it cannot fundamentally bypass the cycle. So this decline is not entirely unreasonable. The market is squeezing out the overheated part of the "AI premium" first. 4/But I don't think we can say storage to the end now. The reason is simple: At present, we have not seen the HBM requirement truly falsified. As long as Nvidia, cloud vendors, and big model companies continue to expand their AI data centers, HBM、DRAM、 The long-term demand for enterprise grade SSDs is still there. What truly proves the peak of storage is not a one-day drop in stock prices. But these signals: HBM orders are starting to loosen. The DRAM contract price is starting to fall. Cloud vendor CapEx is undergoing a major overhaul. Stack the inventory again. Micron, Hynix, and Samsung have begun to provide conservative guidelines. If these do not appear, it is more like a loosening of high-level chips. 5/Tonight's US stock market opens, focus on three directions. Firstly, Micron. Micron is the most direct DRAM/HBM target in the US stock market. If Micron only follows the low opening but does not break through the level with high volume, it indicates that the market is still digesting emotions. Secondly, storage chains such as SanDisk and Western Digital. They are more susceptible to the influence of sector sentiment. Thirdly, semiconductor ETFs and Nvidia. If it's just a drop in storage and Nvidia and SMH hold on, it means the problem is focused on storage valuation. If the entire AI hardware falls together, it is a greater risk appetite pullback. My current view on storage is: The long-term main trend is still present, and short-term fluctuations will be significant. This line is no longer the time for low levels to go unnoticed. There are too many AI expectations, institutional positions, momentum funds, and valuation premiums in current storage stocks. So it will rise sharply and fall sharply. But to truly determine whether it has come to an end, we cannot just look at the Korean stock market for a day. We need to look at the financial report, orders, prices, and CapEx later. The sharp decline of Hynix does not mean the end of its storage cycle. It's more like reminding the market: AI storage has moved from the 'storytelling stage' to the 'order verification stage'. Next, whoever has real needs, real prices, and real profits can stay.