Pumped storage ... needs specific geography, high and low reservoirs close to each other to reduce losses pumping water uphill over long distances. It also needs a guaranteed supply of water, lots of it and the sunny parts of the US where large amounts of solar power are being generated are distinctly lacking in water
One only needs a low reservoir (see the Taum Sauk). Furthermore, while pumped storage certainly isn't a good idea in the Southwest, it is ideal in the Great Lakes area, where there's tons of wind resources (see: Iowa, Minnesota, etc.). And, as it turns out, there is a (functionally) infinite supply of water in Lake Michigan and a functionally infinite amount of land with delta h on the West Coast of Michigan, which has hills immediately adjacent to the Lake due to thousands of years of wind blowing from Wisconsin to Michigan. A storage plant like this already exists, just south of Ludington MI. We could easily build 100 GW worth of pumped storage there, equal to the capacity of all nuclear power in the US.
Pumped storage is also lossy, typically about 65% efficient round-trip.
My experience is that the average is closer to 75%, and it can be as high as 90% with modern, well maintained pumped storage. Pumped storage also has extremely fast ramping capabilities, making it very useful for the minute-by-minute operation of the grid. Of course pumped storage, like all major power plants, requires transmission investment to be fully useful.
Grid gas, coal and nuclear generators don't need storage as they either run flat out to meet the instantaneous demand and they can throttle back in quieter times.
Nuclear, coal, and gas steam plants have very real operational limitations. Nuclear is almost never ramped back to follow load; it's cheaper in the long run to pay negative locational marginal prices (LMPs) if need be. Coal and gas steam can only ramp a few MW per minute, and have minimum outputs whereby they can't maintain power any lower -- and that's often at about 50% of capacity. At that point, any lower output requires a shut down, and then a 12-30 hour cool down whereby the unit can't be restarted. Nuclear, coal, and gas steam are extremely inflexible generators relative to hydro, gas/oil CT, and even gas CC.
At the moment intermittent wind and solar generators use the grid as free storage but the more intermittent power that is added to the generating mix the more that storage will be needed to deal with peak inputs and debits.
Free storage? Wind and solar fueled generators, like all generators, sell the energy instantaneously. Your metaphor makes no sense. All operating power plants sell in real-time. Same price for the same power. Eventually, substantially more storage will have economic value, but on the mainland US grid, not for a long time. California is poised to have 33% renewables by 2020, and they don't need additional storage. (There's an order for ~1.5 GW of storage to be procured, but it's not needed -- it's CA's way of pushing progress forward, seeing that eventually storage will be a less expensive resource (LCOE) than CTs.) Most other parts of the mainland won't have exceeded 10% non-dispatchable renewables by then.
Getting wind and solar farm operators to pay for this extra storage probably isn't going to happen, sadly.
Why should they? In most of tUSA, there's a day ahead and a real time market. Power has a price (LMP). Generators can sell into that market or not. When supply exceeds demand, the LMP goes negative, and all generators who are operating are equally responsible for the problem; all generators who are operating at those times pay the same financial penalty. That includes operating wind and solar and the nuclear and gas and coal that can't turn down.
In the mean time, the number of MWh that are curtailed is a tiny, tiny fraction of the total MWh consumed in America. Storage simply isn't very valuable on the American grid right now because we don't have very much in the way of inflexible generation -- about 20% of the GWh of nuclear, and under 10% of inflexible renewables. It will be many years (more than a decade) before the percent of electricity we have to "throw away" due to inflexibility exceeds 2%, and to the extent that coal plants continue to retire and load continues to grow, that year will be pushed farther and farther into the future.
Storage is interesting tech, but it's simply not necessary for the American grid to operate reliably or economically anytime soon.