The water extraction logic applied to the AI industry is actually the same approach. Large models like Claude and ChatGPT are upgraded, trained, and infer every time, all relying on chips, storage, energy, data centers, and network infrastructure behind the scenes. Breaking down the industry chain further, find those most important, indispensable, and everyone must go through links. Just like on BNB Chain where meme transactions require BNB as Gas fees, the AI industry is the same.

Large model companies sell memberships, users call models wildly, application layers keep expanding, and ultimately all turn into continuous consumption of computing power, storage, and energy. These underlying companies are like the Gas fee entry points in the AI industry. The Claude update also mentioned that storage manufacturers like Micron, Samsung, and SK Hynix have become strategic infrastructure partners.
So, when you buy underlying assets like storage, chips, energy, and data centers, you are essentially riding the long-term water extraction behind the growth of AI applications.