Every time I see a story like this, I wonder how much we are being held back by the fact that it's hard to get several such projects to combine their efforts. Isn't that an awful waste?
The current open source managed OSes are really very different from each other, with only small bits of overlap. You really only have a few contenders at the moment: SharpOS, Cosmos, and Renraku. Of those, SharpOS (GPLv3) and Cosmos (BSD) are C#, and Renraku (CDDL) is Boo.
SharpOS's goals were never particularly clear (they seemed to be building a full OS), but the project seems to be largely dead; they're now a part of MOSA (I'll get to that part in a moment) though, so I'm curious to see where that goes. Cosmos is making progress, but it's not intended as an OS, but rather a set of components you can connect together to make your own OS. Renraku is still fairly new (about 7 months old now, although it's fairly far along) and is intended to be a full OS, when all is said and done.
MOSA is the Managed Operating Systems Alliance, and originally was intended to be a compiler (JIT and AOTC) and a set of specifications for managed OSes to work together. However, at this point they're building their own default kernel into it. Interestingly, they're diverging greatly from the existing managed OSes by using separate address spaces.
At this point, the field is young enough that I think the more people trying things in different directions, the better. There are a lot of unanswered questions, but the more approaches people try to take, the better off we'll all be.
I will say that the primary reason that Renraku is not MOSA-based (despite me being involved in the very early days there (although I didn't end up doing much)) is that they're solely C#-focused, like SharpOS and Cosmos. Renraku is pure Boo, which gives us a flexibility that's we believe to be impossible using C#. Even though we have to do many things ourselves, we think this gives us an advantage that will carry us through to v1 in a few months, and a usable managed OS in 2011.