AI data center power demand is projected to rise 165% by 2030, shifting the compute bottleneck to electricity

The article argues that the rapid expansion of AI compute is running into a power-infrastructure bottleneck. It says global data center electricity demand is projected to climb 165% by 2030 versus 2023, while delayed grid expansion has contributed to more than 70% of U.S. interconnection requests being withdrawn. It adds that ongoing AI capital spending continues to benefit hardware suppliers such as NVIDIA, Broadcom and Wolfspeed, and points to Bitzero (AIBZ) as an example of how scarce power assets are driving a new valuation framework.