DeletedUser405 This decade people are excited about hypervisors, but there's no rule that says a large complicated hypervisor is more secure than a small microkernel (or that a large microkernel is more secure than a small hypervisor, etc.).
A couple times now a microkernel has been proven correct -- admittedly a small microkernel in the case of Verve, but research projects often start small. Meanwhile there are plenty of "hypervisor escape" bugs.
So "hypervisor means secure" is no more true than "JavaScript means secure" or any other magic-bullet claim.
It also has memory and performance costs.