As an alternative of an AI bubble, US tech leaders are standing at a pivotal “inflection level” the place the arrival of economically productive agentic AI instruments has supplied early vindication of the trade’s unprecedented capital expenditure on core infrastructure comparable to knowledge centres and semiconductors, Sheridan mentioned on the sidelines of Goldman Sachs’ Asia Communacopia + Know-how Convention in Hong Kong on Monday.
“There’s a reasonably large disconnect between the demand and the provision of compute,” Sheridan instructed the South China Morning Publish. “We don’t assume that imbalance closes till effectively into the second half of 2027.”
His remarks come amid recurring questions on whether or not the historic ranges of AI infrastructure spending within the US – on monitor to prime US$700 billion this yr – will be capable of ship proportional funding returns.
Such scepticism has solely been amplified as Chinese language corporations have open-sourced cheaper AI fashions, prompting some analysts to foretell an erosion of margins for US mannequin suppliers.