Sufficient storage to handle local load + variation should be pretty well predictable in the future. The interesting difference to me is, since storage will probably cost substantially more than generation, what the ratio will mean for cost of energy in various locations. I have no doubt that at some point, the solar generation + storage formula will be broadly cheaper than fossil fuel.
Very large scale storage is fairly cheap as long as it's used every day. The issue is when your pumped storage power plant is only needed 1 month out of the year. However, with solar the daily demand for storage is fairly fixed as excess capacity is all generated at the same time of day.
That's just an overcapacity problem (in this case, overcapacity on generation, not storage). In another post here, I proposed that we build a smart grid that supports pricing arbitrage on generation/storage in order to maximize generation utilization, and then let the market sort out storage requirements with a more distributed, fine-grained storage model - car batteries, for example.
