Many data centers use spinning disk hard drives in a write-once configuration. This is fine for some use patterns and lets them take advantage of SMR and higher density. Dropbox has blogged about this: https://dropbox.tech/infrastructure/extending-magic-pocket-i...

My understanding is that every drive is write-once and once full it doesn’t change. Eventually, if most of the file data/blocks stored on a drive are deleted by users, the whole drive is erased and re-written.

That said, while being write-once only isn’t a blocker, my guess is seek latency will still prevent this tech from replacing spinning disks.