China's top chipmaker warns that rushed AI data center capacity could remain idle — SMIC chief says utilizing ballooning capacity 'has not been fully thought through'

Semiconductor Manufacturing International Co.'s co-chief executive, Zhao Haijun, cautions that the rapid construction of AI data centers globally may lead to idle capacity, similar to China's underutilized data centers from the early 2020s. The surge in data center construction is driven by a desire to build extensive capacity quickly, without fully considering how it will be utilized. The uncertainty surrounding the timeline and impact of AI adoption across industries has raised concerns about a potential AI bubble in the tech sector. Major tech companies like Alphabet, Meta, and Microsoft are projected to invest billions in AI infrastructure, with overall spending on AI-related infrastructure expected to exceed $3 trillion in the next five years. China is also heavily investing in AI infrastructure, though challenges remain in optimizing utilization and ensuring economic viability of large-scale data center projects.