Recently, Zhao Haijun, the co-CEO of SMIC, cautioned during an earnings call that the unchecked expansion of global AI data centers, devoid of adequate planning, might culminate in unused capacity and a squandering of resources. This, he emphasized, underscores the need for vigilance against what could be the most significant bubble of the AI era. He highlighted that, at present, a multitude of companies are keen on constructing data center capacity sufficient for a decade's use within just one or two years, yet they are devoid of a clear plan for its actual utilization. Moody's Ratings forecasts that global spending on AI-related infrastructure will surpass $3 trillion over the forthcoming five years, with Alphabet, Amazon, Meta, and Microsoft alone projected to invest $650 billion in capital expenditures by 2026. Zhao Haijun posits that this construction frenzy, reminiscent of the premature development of high-speed rail and highways prior to a surge in traffic, could lead to a waste of resources. He draws parallels to historical precedents such as the dot-com bubble and the overabundance of fiber-optic cables in the telecommunications sector as cautionary examples.