Also I’m fairly sure that backup2l doesn’t do deduplication, so if your server has many copies of the same file (for example mine has several WordPress installations on it) Borg’s backup uses barely more than the storage for just one copy. And the deduplication is also done at the block level, not the file level.
There’s also a problem with backup2l malfunctioning if filenames are a bit weird. Some of it is done by shell scripting and it’s subject to some of the pitfalls with file names containing spaces or quote characters. I don’t remember the exact details and I don’t know if it was fixed.