There was a startup created based on the ideas that came out of the exokernel research in the mid-90’s called ExoTech. They tried to make high-performance video-on-demand server appliances. I’m not sure what happened, but they didn’t take off.
As for exokernels themselves, the dramatic performance gains that these systems yielded back then were gradually dampened by evolutionary optimizations to mainstream operating systems. That, along with the established use of mainstream systems, and with the myriad real and complex application workloads that don’t yield the same idealized performance graphs as those found in various papers, has largely cast exokernels into the realm of academia, despite other arguable advantages including flexibility, reliability, security, and so forth. Their fate is one that is mostly shared by microkernels.
Thanks to Russ for this discussion.