I think the latest report about that repository was in [Solved] Stats, Stats, scanning - #4 by g33kphr33k . So even a 200+ TB repository seems to “work”. Based on the massive performance impact of that size, I’d say the absolute limits for reasonable performance are around 100TB / 100 million unique files for a single repository. If at all possible, try to stay one order of magnitude below that. What you’ll definitely want to do at that scale is increase the pack size to the current limit of 128 using --pack-size 128
Right now the only way to scale things further is to use multiple independent repositories. Restic is not optimized for Petabyte scale backups so far.