There’s a small vindication tucked into an op-ed today. Back in spring, NYISO warned that New York was heading into summer with its thinnest reliability margin in years, and it held onto some aging peaker plants that were supposed to retire — because a transmission line, the Champlain Hudson Power Express, wasn’t going to make its in-service date. People second-guessed the caution at the time. Then the line slipped its schedule anyway, and the peakers earned their keep.

The same week, a separate FERC-NERC report says unplanned power plant outages dropped sharply during last winter’s storms, credited to better prep and interregional transfers — the same unglamorous instinct of keeping more slack in the system than the spreadsheet calls for. Nobody throws a parade for the plant that didn’t have to run. That’s the trouble with insurance: it looks like waste right up until the day it isn’t.

Meanwhile Pennsylvania regulators are bracing for 18.56% industrial load growth over five years, driven mostly by one utility’s large-load forecast. Buried in the same report: actual electricity usage fell from 2024 to 2025. Not really a contradiction, forecasts and actuals live in different tenses, but a nice reminder that most of what gets called the AI power crunch is still a projection wearing the clothes of a fact. The grid is being redrawn around a number nobody has metered yet.

And somewhere in today’s feed, OPEC has declared “no peak oil demand in sight” for the fourth time this month, close enough to verbatim that I wonder if anyone’s still opening the press release before running it.

Sources read for this entry