It's a shame that microkernels and containers came to prominence around the same time. IMO, containers were easier to understand and start working with and seem to have "won out" as a result.
Microkernels are really interesting, though! I'm glad people are still working on and talking about them.
Microkernels are doing pretty much fine on real time embedded OSes, and QNX can hardly complain about their customer base.
Each of those containers gets monitored by a Minix instance running inside Intel CPUs.
Apple OSes are transitioning to userspace only drivers.
Google keeps pouring money into Fuchsia, and Trebelified Linux looks quite similar to a microkernel using user space processes with Android IPC, classical Linux drivers being left for the existing "legacy" drivers.
For MirageOS, you have OCaml and the Complete OS Change, that is much more radical than containers, which is 99.9% still C and Linux + 0.1% of added ingredients.
I would imagine something done in plain old C or Rust would get a little more traction. But then an entire OS ecosystem change is no small thing.
I think containers have been a great introduction to applications deployed as minimal, immutable images. I imagine operations teams will be more ready to deploy unikernels because of their experiences with containers.
Also as a note, microkernels are older and not necessarily related to unikernels.
Microkernels are really interesting, though! I'm glad people are still working on and talking about them.