How big of a dataset can Sideloader handle?
Tested to over 100 TB per run; pipeline parallelism scales linearly—your cloud bucket is the limit.