The United States faces a consequential test of its energy transition as artificial intelligence infrastructure expands faster than the electricity system can accommodate it. An analysis by the Sierra Club warns that planned data center development and the fossil-fuel generation supporting it could push US power-sector carbon dioxide emissions sharply higher by 2030. The projection raises a strategic question for technology companies, utilities and policymakers: can the country expand its AI computing capacity without sacrificing years of progress in reducing power-sector pollution? The answer depends on more than the number of facilities developers announce or the computing capacity they promise to deliver. It also depends on how quickly projects reach construction, which electricity sources ultimately supply them and whether renewable generation can meet demand at the necessary pace.