Why American data centers can't plug in
The article examines why new data centers in the US face years-long delays connecting to the electric grid due to a combination of outdated interconnection rules, utility monopolies, and a fragmented approval process. It argues that the current system, designed for a different era of energy demand, is unable to keep pace with the rapid expansion needed for AI and cloud computing.
Background
- The article looks at why U.S. data centers—the warehouses of computers that power cloud computing, AI, and streaming—face months or years of delays connecting to the electrical grid, even when they have the land and permits.
- Key bottleneck: the "interconnection queue"—the process by which a new power customer requests a connection to the transmission grid. The queue is managed by regional grid operators (like PJM, MISO, CAISO) and is overwhelmed with requests from solar, wind, battery, and data center projects.
- A core problem is "serial" study: each new interconnection request triggers a full system study, and later requests must wait for earlier ones to be studied. This creates backlogs of thousands of projects. The article proposes moving to "cluster" studies (grouping requests) and reforming cost allocation so that grid upgrades are paid for differently.
- Why it matters: AI and cloud computing are growing fast, but the grid can't connect new data centers quickly. If interconnection reform doesn't happen, the U.S. could see data center construction stall, driving up costs for tech companies and slowing AI deployment.