Back feeding and metering all that power isn't exactly simple and it's also a big economic cost to just shove onto people in the form of increased battery wear and effectively reduced range since the most common trips begin early in the morning before renewables like solar come back online to produce the excess needed to recharge all the cars you've borrowed power from over night. Because of that cost that kind of distributed battery power is going to come out quite expensive unless you're just vastly underpaying for the depreciation of the battery like Uber does for it's drivers' cars wear and tear.
Another big thing is your 21 TWh number has huge flaws. That's fully draining everyone's batteries to provide the power and no one is going to accept that we want the cars because we need to go places. Next we're decades away from even getting close to full EV penetration. Even if we stop selling new ICE vehicles they're durable goods that people hold on to for a long time. It's the kind of 'Company/technology X takes over everything' assumption that so vastly inflates the valuations of tech companies all over the sector.
It was simply an order of magnitude comparison for manufacturing capacity. Slowing EV adoption at say 50% penetration would be beneficial for the grid as that excess capacity in mining and manufacturing could be used for grid batteries.
I doubt the the cheapest grid has anywhere close to 21 tWh as having more generation is more useful. Hydro is extremely flexible and reliable can fill in for a modest and predictable multi day shortfall. On the other hand if a power plant is down then batteries get extremely expensive.
That said, the cost of batteries isn’t simply a function of total capacity as unused capacity extends battery lifespan. The actual grid will take this into account which then impacts spot prices etc. Nobody operating grid scale batteries is going to 0% for wholesale prices of 5c/kWh but you bet your ass they’re going to 0% for 1$/kWh.
Another big thing is your 21 TWh number has huge flaws. That's fully draining everyone's batteries to provide the power and no one is going to accept that we want the cars because we need to go places. Next we're decades away from even getting close to full EV penetration. Even if we stop selling new ICE vehicles they're durable goods that people hold on to for a long time. It's the kind of 'Company/technology X takes over everything' assumption that so vastly inflates the valuations of tech companies all over the sector.