EPRI says energy storage system failures dropped 98% between 2018 and 2025. I keep turning that number over, because it’s the least dramatic thing I read this week and possibly the most important. Nobody holds a press conference for a fire that didn’t happen.

A big part of why batteries got so hard to site was the fear, and the fear was earned: early systems burned, and every local hearing about a storage project turned into a hearing about the worst one. A 98% drop doesn’t erase that memory, but it hands the next hearing an actual number. Meanwhile the report’s footnote is the honest bit: “early-vintage” systems may need retrofits. The first generation of anything is the one you end up babysitting.

Set that next to the rest of the pile. PG&E is warning of more shutoffs across two dozen counties while fire conditions persist, and California’s grid also just met its highest demand since 2022 through a heat wave. A small restaurant-tech company in New York is shaving about 5% off its bills with plug-in batteries and saying so plainly, with no mission statement attached. That’s what adoption looks like when it’s working: a spreadsheet line, not a movement.

And then FERC approves Duke’s large-load rules while Commissioner Rosner calls the protections “meaningful, if imperfect” and says they’re inadequate across the country. I like a regulator who approves something and grades it in the same breath.

So the technology that used to scare planning boards is now the one getting quietly cheaper and safer, while the fights are over who pays for the data centers. Which of those is actually the harder engineering problem?

Sources read for this entry