Is it possible that they are working on batches, and not doing any hashing/compression in parallel with the uploading? It seems feasible from your screenshot that they are getting ~10GB of data at a time, compressing(?) and hashing, and then uploading, and then starting on the next ~10GB.
