Elon Musk said in a post on X that AI compute production could outrun the industry’s ability to energize new data centers starting in 2027. According to Musk, roughly 15 gigawatts of planned IT load could sit delayed, concentrated in North America.
Musk attributed the gap not to chip supply but to site-level infrastructure, according to his post: grid interconnections, transformers, cooling systems, networking buildout, permitting, and turbine availability.
Grid interconnection queues and transformer lead times have been widely documented constraints on data center construction independent of Musk’s claim, with utilities across multiple U.S. regions flagging multiyear waits for new large-load connections. Musk did not cite a study or third-party analysis to support the 15GW figure.
Operators planning 2027 capacity should treat interconnection and transformer procurement, not chip availability, as the binding constraint on deployment timelines.
Elon Musk, in a post on X.
