That looks like a template you could fill in with any emerging-but-not-yet available technology, and the resulting strategy will fail the vast, vast majority of the time, since most exciting new tech developments never come to fruition. Plus, what would "software for OAM" even mean? It's so vague as to not even constitute advice.
Given that OAM is a modulation technique, which is a physical-layer tech, nothing above firmware will even know about it. At most software will be able to read/set a flag enabling use, similar to FX/DX on Ethernet.
I think its more than that. When you exceed by an order of magnitude the bitrate, different APIs will be needed to take advantage. Sure you can just send on a socket, but it will max out at the rate you can call that API (including kernel switch, copying data, waiting for a completion interrupt etc).
Some kind of virtual hardware or mapped memory will be desirable to achieve great leaps in bandwidth usage.