First: Friends of mine that should know, say that there is not yet any real effort even to figure out what the hardware for a next console would be, much less actually pulling together all the resources to design, implement, manufacture and release one. I think it'll be many years before a truly new console comes out. (I'd be surprised if it was as early as 2012). Perhaps we'd see something like an "Xbox Plus" by then, that might come with different peripherals (Blu-ray? Natal?), and/or additional hardware (Natal acceleration, etc).
Second: The nice thing with XNA is that all you need to do is port the Compact CLR/.NET Framework, and the XNA runtime assemblies, and you will have a new platform supported. Even better is the fact that no developer can go "outside the box" and "bang the hardware." With the original Xbox, some developers wrote pretty much to the metal, which meant that those games were hard to create "compatibility handler" for for the new Xbox. (My guess: A "compatibility handler" is really more of a "port").
XNA has none of those problems -- all the code use the managed API, that Microsoft has full control over.
If I was in Microsoft's shoes, it would look very tempting to keep supporting XNA Game Studio on any new platforms (game consoles, music players, smart phones, hand-helds, dining room tables, etc). Thus, personally, I would be comfortable making the bet that, even if I started today and my game took three years to make, I could use XNA to deliver it. And, if it turned out that Microsoft decided to get out of the console business three years from now, and we all are playing games on our Samsung Em-Gages or something, then most of the effort in a game goes into the design and the art -- re-writing the code for another platform, once you have a working game, is a lot quicker than the first time around. A lot quicker.
Jon Watte, Direct3D MVP
Tweets, occasionallykW X-port 3ds Max .X exporter
kW Animation source code