After AI stocks saw their prices soar alongside their performance, market concerns about a potential bubble have surfaced, including potential risks such as overvaluation, technological bottlenecks, and capital manipulation. Nvidia and OpenAI, meanwhile, announced a groundbreaking investment plan: a $100 billion joint venture to build a 10,000-megawatt AI super data center, deploying millions of GPUs to support the training of next-generation large language models. This significant move has further fueled market expectations for the future of AI computing power. Recently, several mutual fund managers told Securities Times reporters that the global AI race is entering a new phase, with the competition for AI infrastructure characterized by ultra-large scale, ultra-high energy consumption, and ultra-high investment officially underway. From chips to liquid cooling technology, from computing clusters to energy support, a global AI computing power industry chain is being fully activated. At the same time, the focus of international competition is quietly shifting from "model competition" to the more fundamental and core "computing power competition."