Time Machine backups rely on hard links a lot, there are files in each snapshot pointing at the same data on disk. I imagine that could be what is going on here. This person could actually have had 60Tb worth of backup data sets on a 2Tb drive, only possible because there are very few differences between the data sets.
Until you have it write a github action that uses a bash script to process a directory structure with spaces in the names. Never seen ChatGPT so confused before.