> You're comparing figures on generation to storage. Yes, a big nuclear plant that generates 2.5 GW of electricity often costs $15 billion … buying 2 gigawatt hours grants you the ability to store enough energy to output 2 GW for a duration of 1 hour
I'm clear on the distinction between power and storage capacity; that's why I said, "Li-ion was about US$111/kWh which is probably around US$300/kW on th[sic] current grid, depending on how many hours a day you need battery" and linked to my note from 02019 about precisely how the price of batteries interacts with the price of energy, depending on how many charge/discharge cycles you get per day and how long they are.
The thing I probably should have been more explicit about is that, on the current storage grid, batteries are needed about 2–3 hours a day; US$111/kWh times two or three hours gives "around US$300/kW": to bridge those 2–3 hours of shortfall, you need 2–3 kWh/kW of batteries. That's what makes Li-ion batteries profitable to install today.
It is of course true that as the amount of solar energy in the grid increases, this duration also increases, especially if demand response doesn't materialize as we might hope. Figures from the CAISO report I linked elsethread http://www.caiso.com/Documents/Final-Root-Cause-Analysis-Mid... show that California's solar generation in the summer is roughly 12 hours a day and fairly constant during about 10 of those hours, ending just after peak demand, and that peak demand is about twice "valley demand". On the fateful day, the demand ramped down from about 45000 MW to about 25000 MW nearly linearly over the other 12 hours, so if you wanted to feed the entire California demand from expanded solar generation, you'd need to store up 420 gigawatt hours (1.5 PJ). You'd need 12 kWh of batteries per (average) kilowatt, which would bring the price of storage to above US$13000 per kW at the battery prices I found two years ago, which is slightly higher than the price of thermal power generation. Hopefully those battery prices will go lower and/or people will shift to using more power when the sun is up, for example because it's free, but those prices are already low enough to invest the first US$50 billion into ramping up battery production and new extraction technologies.
Your point about service life is well taken, but I think you probably would benefit from doing real NPV calculations, with discount rates. I'll make an effort tomorrow if I have time.