Text data like CSV compresses extremely well. These tools shrink files for storage and transfer, and unpack them when you need to read them again.
CSV is repetitive text: the same column names, delimiters and similar values repeat on every row. Gzip exploits that repetition and often reduces a file to a small fraction of its size, which speeds up uploads and cuts storage costs.
Compression is lossless, so the decompressed file is identical to the original. Keep the uncompressed source until you have confirmed the archive opens correctly in the system that will receive it.
Compress before emailing a large export, before uploading it to a system with a strict size limit, or before archiving old data you rarely touch. Decompress with the matching tool the moment you need to open, edit or re-import the file.