AI’s power race is shifting leverage from chipmakers like NVIDIA to the grid
AI’s rapid growth is turning electricity into the main bottleneck. Demand from data centers is rising faster than grid expansion, pushing regulators and utilities to ration access, raise interconnection costs, and consider moratoriums on new large-scale facilities. Forecasts from Goldman Sachs, the IEA, and the EIA all point to steep increases in data center power use through 2030, but new generation, transformers, and grid connections are delayed or cancelled. This shifts leverage from AI companies to utilities, grid operators, and power producers, who control who gets power, when, and at what price. Texas is moving toward a “pay your own way” model for large users, while New York is weighing a pause on new data centers. Unlike bitcoin mining, which can curtail load and use cheap surplus power, AI needs constant, firm electricity, making it more expensive and politically contentious.
