

Sure, works fine for inference with tensor parallelism, USB4 / thunderbolt 4/5 is a better (40Gbit+ and already there) bet than ethernet (see distributed-llama). Trash for training / fine tuning, that needs higher inter GPU speed, or better a bigger GPU VRAM.
You can cycle the smaller drives to cold backup, that’s not a waste. You do have backups, which RAID is not, right?