It might make even more sense to run a cutting-edge distributed OS on the actual Oxide hardware. With rack-scale platforms like this it could be feasible to do SSI with distributed memory across multiple "nodes". Current "cloud" platforms like Kubernetes are already planning on including support for automated checkpointing and migration, which is sort of the first step prior to going fully SSI.
Depends how deeply integrated it is; if nothing else, I suspect that bhyve and kvm have sufficiently different APIs that it would be at least quite annoying to paper over the differences.
