Pumping air is more "springy" than pumping water. But "springy" isn't exactly the worst thing when it comes to energy storage. Pressurized air storage exists, even though it comes with certain disadvantages, namely thermal losses. Ideally you would want to somehow harvest at least some of the "cold" created during decompression. At least with the "diving bell" variation of compressed air storage, you would store some of the energy not as gas pressure but as lifted water, avoiding the thermal penalty for this part of total capacity.

But you are right, FhG are working with a permanently contained gas bubble that is compressed/expanded as needed. Many of the experiments done with the prototype where about the effect of different amounts of air and about the efficiency difference between a local bubble and a passive air connection to the surface. I am sure that FhG have put more thought into the pneumatic link idea than I could ever do and they seem to have come to the conclusion that a pneumatic connection just isn't feasible, or maybe that losses along the connection would be too big. From what I could gather from other sites (e.g. http://forschung-energiespeicher.info/en/news/aktuelles-einz... ) they seem to have a strong focus on practicality, e.g. having all the electric and mechanic components concentrated in a module that is slotted into the sphere from the top, which might make it a candidate for removal/replacement.