So I've posted this before, but aluminum smelting is just plain old too hard for this to be as cheap as batteries. I mean, it's great if you're Alcoa, an aluminium smelter, but let's take a look at bulk prices:

Aluminium costs $1.80/kg: http://www.indexmundi.com/commodities/?commodity=aluminum

Alumina, the product of the aluminium-air reaction, costs ~$0.45/kg: http://www.indexmundi.com/en/commodities/minerals/bauxite_an...

The cost of refining aluminum from ore, in bulk, is therefore >$1.35/kg (otherwise, Alcoa would be cleaning up on the commodities market). The cost of a 3000km battery with 100kg of aluminium is therefore $135, or roughly 4.4 cents per kilometer, or 7 cents per mile.

That's nearly as expensive as gasoline, going off of commodity metal prices and assuming absolutely no overhead or profit margin. That's not to mention the environmental consequences of aluminium refining w.r.t. the massive amount of energy therefor required.

It's impressive tech, but it's not feasible. The Hall-Héroult process has been the subject of intense scientific scrutiny for over 100 years; everyone and their brother has tried to make it more efficient: we use aluminium for everything. Aluminium is not going to get any easier, barring massive breakthroughs in fusion energy and/or the Second Coming.

EDIT: Originally my calculated cents/mile was off by a factor of 10. However, the latter point, "nearly as expensive as gasoline", stands: it costs $250 for a gas car to drive 3000 km, at 30 mpg and $4/gallon. I didn't notice the error because I used this latter comparison rather than figuring out gasoline's cents/mile.