A week or two is the longest almost anybody should need to bank energy. Mostly you just need a few hours of high-efficiency storage to maintain local resiliency. Beyond, say, the 4 hours' worth cycled every night, round trip efficiency matters little.

Beyond a week, burning liquified ammonia imported from tropical solar farms wins. Probably you keep a week's worth of that on hand to burn while you wait.

There is no need for storage (beyond use for load-leveling and peak shaving) until after renewable generating capacity is overbuilt enough to charge it from. (The alternative would be to recharge storage by burning NG: You would better burn that NG and deliver the power to users instead of drawing down and then recharging storage from it.) Until then, capital is better spent building more renewable generating capacity itself, displacing more coal, than on storage.

But building the factories that will build the storage that will someday be needed should happen now, because we will need a lot, and making automated factories takes a long time. And that is happening. Some of those may end up idle as cheaper methods replace theirs, but for at least a few decades, demand will be insatiable.