An opinion article from Amazon Web Services argues that AI can help U.S. utilities clear a growing backlog of power projects waiting to connect to the grid. The author, AWS general manager for energy and utilities Joseph Santamaria, notes that more than 2,000 GW of potential projects are awaiting review and that average interconnection wait times have risen from about two years in the early 2000s to roughly five years today.
Santamaria says AI-based planning tools can compress interconnection studies from months to hours by simulating thousands of what-if scenarios at once, while human engineers still make final decisions. He cites Duke Energy using AI to double the number of projects evaluated in a year, and Pacific Gas & Electric consolidating data from over 80 sources to prioritize more than 17,000 grid upgrades, with a claimed 20% cost cut in a $100 million pilot.
The piece acknowledges critics who question whether new data centers and other large loads will raise costs for existing customers or worsen bottlenecks. It points to an Amazon-NIPSCO agreement in Indiana as an example where the large user covers infrastructure costs, and notes FERC held discussions this summer on AI for grid planning. Because the article is written by an AWS executive, its examples and projections are not independent evidence, but they reflect a growing push by cloud providers to position AI as a fix for grid delays.