The logic of AI infrastructure is spreading, from GPU capital flows to storage.

If you're still wondering why Micron can rise 20% in a day, you might need to look at AI infrastructure from a different perspective. Over the past two years, everyone's focus has been on NVIDIA's GPUs, but as the parameters of large AI models grow exponentially, data storage and transmission bottlenecks are becoming more and more prominent. Simply put, no matter how powerful your computing ability is, if data can't be stored or transmitted in time, the entire system still gets stuck.

Last week, NVIDIA signaled an acceleration in demand, providing strong support for GPU-related stocks, while AI startups are lining up for IPOs, prompting investors to allocate more money into the entire AI ecosystem rather than just focusing on the few largest chip stocks. Micron happens to be at the forefront of this diffusion effect.

UBS analysts also said that the market will gradually give Micron a "more normal valuation multiple," no longer treating it as purely a cyclical stock. Looking at the broader environment, the three major cloud service providers are expected to increase capital expenditures by over 30% next year, much of which is directly or indirectly related to storage. Reports indicate that to meet the training needs of GPT-6 level models, data center storage bandwidth is expected to increase 5 to 8 times over the next three years.

So yesterday, Micron surged nearly 20%, which is not just good news for one company, but also represents the entire AI investment logic spreading from GPUs to all related areas. In the past, I didn't pay much attention to storage when prices were cheap, but this time, Micron's actions reminded me—this wave of technological advancement is far from over; it is just becoming broader and deeper. #美光市值突破1万亿美元