You are probably going to move where the power is cheap though:
http://www.bloomberg.com/news/articles/2015-03-17/silicon-me...
In this case, Alcoa is performing aluminum smelting in Iceland precisely because of cheaper geothermal power generation available there.
Yes, as my pre-comment scan of the Wikipedia article on aluminum smelting (to double-check I had the right electricity-intensive industrial process) said in the second paragraph [1]. I didn't think it worth adding to the comment.
Of course aluminum plants move to where electricity is cheap, all else being equal... exactly what alternative do you think I was proposing?
Further, geothermal is significantly different for this use case, as geothermal can be used for base load applications where solar can not. Since such plants tend to run without regard for where the sun is in the sky, this sort of industry would be very difficult to run off of solar power without absurdly overprovisioning for power storage, which remains a significant problem at this scale. You are also not going to run your aluminum smelting plant for 18 hours out of the day on battery power!
Here's another fun thought: 6% of electricity is lost due to transmission costs. Will it be more efficient (in some cases, at least) to physically move charged batteries from solar fields to (nearer to) end users than to send it over the wire?
No, not in this universe, anyway.
The most basic power distribution in the world is people with Ox and cart distributing lead/acid batteries to remote villages. Transmitting power this way is the least efficient way possible.
It looks like a Tesla S battery pack is a little over 1,200 lbs. So for a rough estimate, consider the distance a Tesla S can tow 20,000 lbs. Probably not very far.
Unless I botched my reasoning somewhere, it looks like battery energy densities would have to increase by an order of magnitude or two for this to become a thing.
http://blogs.scientificamerican.com/plugged-in/2014/10/07/en...
Give it time.
In 2003, Alaska built the world's largest battery at 4.7MWh.
In 2012, China built the world's largest battery at 36MWh.
In 2014 Southern California Edison announced it was buying a 400MWh battery.
On that curve, we should be up to an 18GWh battery before all that long.
https://en.wikipedia.org/wiki/Bath_County_Pumped_Storage_Sta...
Pumped hydro electric storage ftw.
I would be shocked if Tesla Energy isn't assembling entire systems cargo container by cargo container in Reno and shipping them by rail across the country in the next 2-3 years.
Besides, an ability to increase production implies idle production capacity, which, as you state, is a poor use of capital allocation.
Lots of people see a few solar panels and tesla announcing sleek batteries, and think a new age is upon us. Not yet.
Take a look at Hawaii - the "rate of the change" is so big the utility pretty much stopped connecting people's PV panels because the grid couldn't deal with it.
As the percentage of renewables increases the fluctuations grow bigger. e.g. Germany already has 40GW of solar installed...so if its overcast then you need an entire fleet of nuclear plants to pick up the slack (sure it doesn't drop to 0GW but you get the point).
Those companies you mentioned that will still need power - not sure how that is relevant to the fluctuations? Or do you propose that they carry to cost for everyone?
That's the situation I'm familiar with, but that may be a peculiarity of the under-developed grid in my area.
Building smelters next to coal fields is a way of producing cheap aluminium. Coal is still the cheapest way to make electricity - it's only places where the government has explicitly made this not the case that coal is expensive.
Generally unfavourable government regulations simply moves production to places without the regulation - for no net change in production or coal consumption. Sometimes it's a net negative if the new location has lower local emissions standards and technology.
People are sometimes fooled by sunk capital costs of old power plants.
While change is happening and change in inevitable I think you're greatly overestimating the rate of the change and badly misunderstanding the quality.