The second part of your comment hints at the trouble with your first: Optane didn’t provide major benefits over existing storage architectures when used for a block-addressable file system. It had significant density benefits when used as RAM, but at an equally significant latency penalty.

To really see wins, you needed an application data model that could benefit from the high density/single node capacity, byte-addressability, and persistence. Basically an in-memory database, either general purpose or embedded in a specialized data processing application.

Those are premium, niche markets, and could never support the demand Intel needed to make commercial sense.