You are right about the power efficiency but in terms of the...

You are right about the power efficiency but in terms of the 10-20 years away It doesn't really work like that - WBT haven't attempted to use the cell in large arrays yet. There is no point even attempting it until they have the selector technology working and optimized.

In reality this is the longest part of making much bigger arrays.

My understanding is the cell already stacks in the z axis which means they can do 3D stacking already, 3D stacking = big capacity.

They are focused on the low hanging fruit and getting revenue in the door via embedded.

Trying to do everything at once, almost always leads to failing at everything. I am happy they have sharpened focus on embedded but arnt losing sight of the huge price of discreate.