A well recorded, uncompressed, Steinway-D needs a fairly large, efficient, and expen$ive speaker to reproduce its full fire power at a realistic volume ...
Indeed, an important aspect here is sheer radiation area coupled with high efficiency to give way to the physicality and full dynamic range needed for the more authentic reproduction of a piano. Add to that coherency, relative timbral accuracy, not least (transient) speed/dynamic prowess, and a low degree of smearing through the whole of the (wide) frequency span - as rightly pointed out by poster @james_w514 into the sub-region as well, all of which is no small task.
Listening to a grand piano at a relatively close distance can be an extremely physical, dense and dynamically (and emotionally) startling experience, just as it can be a very liquid/floating and gentle instrument to take in. The trick is to incorporate the whole of the envelope of these traits mentioned into a speaker, the "gentleness" as well as the sheer force of the instrument, for it to be a more convincing reproduction - not something realistically attained with a <$4,000/pair speaker. This is not strictly what the OP is inquiring on, though; for that to be accommodated several other posters above have chimed in.