This is good news
With all the other x86 virtualisation solutions maturing to the point that they are potentially competing with VMware's offerings, it's good to see him return to academia to rethink the approach for virtualisation that is being taken today.
Processors containing binary translation instructions are already available or being taped out soon, so the need for virtualisation instructions will fade away and a thorough revision of the whole concept is in order.
I hope he will go back to work on SimOS or a new project from scratch as a potential solution for MIPS or VLIW based virtualisation that is becoming relevant again with the future release of the Loongson 3 processor. He's made his millions, but now it's time for computer science again.