„…what if there was instead a tremendous expansion of neighborhood grids based on neighborhood PV, Wind or Geothermal generation?...“ 

Bill, under the scenario taken from the article, we are talking about thousands of GWh of electricity to be needed in about 10 years. The local “microgrids” are not going to supply that. Not sure, how may locations can use geothermal energy, but out of those mentioned, only PV can be relatively easily deployed. However even in SoCal, one can get maximum of 1.362 kW/m², but that figure assumes, that the Sun stays all the time straight above you head and that you have ultra-mega efficient solar panel. Since the efficiency of solar panels is around  20% and you can effectively get maybe 4 hours equivalent of “sun above your head” a day, assuming that it is not raining, the likelihood of having PV system able to supply just your EV quite limited. With sufficient supply of land not impossible in Arizona, Nevada but overall limited. 

So we are back to conventional sources, where the best available is the one nobody wants around – nuclear power stations. One block of 2 000 MW could supply maybe 40 GWh/day. Typical power station has 4 such generators, so we talk about 160 GWh/days. No matter how you slice it, I still arrive to the conclusion, that there is not enough primary energy sources to support more than maybe 25% EV penetration. I know, that the excel projection can do more, but unless I am wrong, Matts article is just an utterly rubbish extrapolation for EV cheersladders. And now, I have to work :-)