Post

Actually, I increasingly feel that when investing in a company, what you really need to study is not the product, but the business model. Because technology can be caught up with, the business model often determines how much money a company can make in the long term. Why do I always believe that most storage companies have obvious cyclical characteristics? The reason is not only that DRAM, HBM, or NAND are affected by supply and demand, but more importantly, most of them adopt the IDM (Integrated Device Manufacturer) model. From chip design, wafer manufacturing to packaging and testing, almost the entire production process must be completed by themselves. The advantage of this model is that it can control technology and capacity, but the cost is also very high. A wafer fab, an EUV lithography machine, an advanced packaging production line—all are huge investments. Even if the market enters a downturn cycle, fixed expenses such as equipment depreciation, factory maintenance, personnel costs, and continuous capacity expansion still exist and will not decrease because of chip price drops. So when storage prices enter a downward cycle, profits are often quickly compressed, and financial report fluctuations are very obvious. Looking at Nvidia from the opposite perspective. It does not build its own wafer fabs but focuses on chip design and outsources manufacturing to TSMC. Although this fabless model also bears foundry costs, it does not carry hundreds of billions of dollars in manufacturing assets, allowing more funds to be invested in CUDA, NVLink, software ecosystems, and next-generation product R&D. Therefore, I have always believed that a company's true moat is not just technological leadership but whether the business model can continuously generate high returns. Investment is often not about comparing whose technology is the best, but about comparing whose business model is more capable of weathering cycles.