Yes, that seems to make sense to me.
However, there’s also larger contexts.
-
Newer PCIe generations (assuming motherboard and NVMe drive match) give faster throughput.
-
Some SSD and NVMe drives also allow system RAM to be used for caching, thus increasing their IO significantly - albeit at a risk associated with power failures during write operations. I use that trick to with Samsung SATA SSD (i.e. non NVMe) via their Samsung Magician utility. It effectively raises the throughput for cached content of a SATA SSD to NVMe levels. Using that trick with a drive that’s mostly used for reading, like Steinberg content or other sample libraries would make this trick rather low risk.
-
If NVMe slots are at a premium, then getting fewer, but larger NVMe devices might make sense (partitioned as needed to make the migration from multiple drives easier).