The 520 GB in your screen shot is just an estimate and since in some cases the estimate can be high somewhat substantially, you won't know if this is actually a problem until the backup completes. At that point you could use
tmutil compare or a tool like Backuploupe to see the actual payload size for that specific backup and feel better about whether this is an issue with directory corruption or just an estimate algorithm that isn't highly accurate for your storage details.
Due to how files are stored on drives, you could have a lot of files that fit efficiently on the SSD but end up wasting more space when they sit on a larger volume like the 2 TB Time Capsule so it could also be literally correct and no over estimate. HFS is fairly sophisticated on how it allocates blocks and stores files, but to simplify things - there is potential for huge inefficiency in file system storage.
For example, take this contact card I dragged onto my desktop:
The file size is 536 bytes, yet it takes well over 600% more space than needed since that one file takes up a 4 KB block of space on the filesystem. As the software is likely glossing over these details or making overly conservative estimates on how much slack space is going to be needed on the destination.
My experience is 20 GB of storage overhead is extremely unlikely so my hunch is it is an overly cautious estimate where the designers wanted to be "better safe than sorry" with making sure the actual backup used less space than estimated.