Battery science is hard, capacities are limited by chemistry, and the easy-to-find electrolytes (also the cheapest and most abundant) are fairly well known. Barring breakthroughs in allotropes (such as graphene, everyone's current favorite carbon molecule), novel inorganic molecules, or biotech-derived approaches, I suspect chemistry doesn't have a whole lot extra to give us. Even reasonably abundant minerals such as lithium portent supply crunches when scaled to, say, a century or more worth of battery EV production (though Elon Musk speaks confidently of addressing this via recycling -- though no recycling process is 100% efficient).
The more interesting approaches I've seen for grid-scale battery storage involve possibly less efficient, and certainly less convenient, but cheap and highly abundant materials such as the molten salt or liquid metal (see Donald Sadoway's work) concepts. The chemistry works, the materials abundance is there.
I've played with some back-of-the-envelope calculations suggesting that for electrical energy storage, very large scale thermal storage, say enough for two weeks of supply at the scale of the US, is plausible. The heat-substrate facility would be approximately the size of an existing very large petroleum storage tank farm (these exist in Oklahoma at the terminus of several oil pipelines). See: http://redd.it/1viied
But the really useful forms of energy storage humans have turned to are long-chain hydrocarbons, whether created recently in nature as olive oil, beeswax, tallow, or other biogenic oils, or in the form of petroleum, and to a lesser extent gas and coal.
The prospect of the US Naval Research Lab's electricity-to-fuel project utilizing seawater as a source of both hydrogen and CO2 to feed a Fischer-Tropsch synthesis process strikes me as among the most promising prospects I've read in a long time. It requires an external energy supply (the Navy proposes existing nuclear reactors aboard aircraft carriers, or renewable energy in the form of ocean thermal energy conversion (OTEC). Proposed at a scale of 100,000 gallons of aviation fuel (essentially kerosene or diesel) daily, fed by 240MW of electrical energy (roughly the scale of a carrier's reactors), at a cost of $3-$6 per gallon of fuel produced, it's more expensive than present fossil fuel energy sources, but, when tied to a renewable energy source, offers the prospect of a constant-cost liquid fuel source effectively forever.
It also seems to me that scaling this to roughly national levels of production is at least plausible. The NRL's proposal is 1/8400th of present US petroleum consumption, but scaling up costs and plant estimates, $8 trillion in capital, 2 TWe in energy, and a total of 10 m x 4.5 km x 4.5 km processing facility volume, would replace the 20 million barrels of oil presently consumed daily in the US. This compares against some $4 trillion in capital expenditures on new conventional petroleum exploration from 2005 - 2012 with a net reduction in oil supply of 2 million barrels/day.[1] And yes, the supplied electrical energy is an additional cost: this won't be cheap. But it's forever.
Round-trip efficiency isn't great: no better than the 60% efficiency of hydrogen electrolysis, with additional energy input requirements for CO2 extraction and a lot of seawater handling (8.8 billion liters/day for the Navy's proposal, 74 trillion at national scale). But once produced, oil doesn't suffer from the relatively rapid storage losses of most alternatives: a few percent per hour or day for many battery, thermal, or kinetic flywheel systems. If used for electrical generation there's another 30% generating loss experienced, for a maximum net efficiency of 42%, but if used to provide liquid fuels to critical uses, this could well prove highly attractive -- there are somethings it's really hard to substitute for.
The process should also be carbon neutral, though it does involve a net transfer from present ocean to present atmospheric carbon reservoirs.
I discuss this more here: http://redd.it/22k71x
It's not that I don't expect some improvement in battery technology, but there's a 5x advantage in energy storage densities (MJ/kg) between the best present battery storage technologies (lithium-air) and liquid hydrocarbons. That's closer to 25x for LiON vs. oil[2]. While incremental efficiency improvements (and better energy scavenging through, say pervasive PV surfaces) might be useful for electronics and their Moore's Law dynamics, for devices operating in the physical world and its constraints, batteries will continue to be at a disadvantage in cost, capacity, complexity, and/or abundance.
________________________________
Notes:
1. Steven Kopits presentation, February, 2014. http://energypolicy.columbia.edu/events-calendar/global-oil-... Also discussed and analyzed by Gail Tverberg: http://ourfiniteworld.com/2011/03/03/steven-kopits-oil-the-e...