More importantly I think is the differential 'problem'. It stands to reason that the technology will advance to the point where the propulsion motor in an EV could be direct drive, but you'd then need a motor per wheel to manage the differential wheel speeds. Very feasible of course, but a second motor on the same axle increases the overall complexity that one is otherwise trying to drive down by implementing direct drive.
An interesting potential solution is an integrated dual-ish DD motor, where you have only one unit that has mostly one set of components and is thus mostly one motor (as opposed to two motors bolted together), but it doesn't push its power through a traditional-ish mechanical differential like the model S does now.