> The big downside to decentralising residential power is that's variable demand–precisely the sort of demand you can net out against other parts of the grid.

Residential is variable but for the most part not all that amenable to time shifting, at least at present. Isn't a grid most efficient with a constant load, with the next best being something that varies only slowly over time and is highly predictable?

Then there's EVs. I'm under the impression that replacing a notable fraction of the ICE cars on the road with EVs would at present place the grid over capacity most places in the US.

When it comes to unit cost doesn't the ultimate benefit here lie with the consumer's pocketbook? I don't see why residential considerations should make much (or even any) difference to dense commercial or extremely high capacity industrial users. Given that solar plus battery is reasonably affordable for a large chunk of the US population it doesn't really seem like much of a downside when framed as a voluntary expense. I still see people building it out where I'm at (suburbs) despite (AFAIK) the subsidies ending.