>Based on reading this discussion it would seem plausible and even likely that the
>Nvidia x86 processor rumors that surface from time to time are likely to have some
>basis in fact.

>Assuming the hypothetical x86, the question is could they execute on this?

Probably yes, they still have cash. Only problem is that their integrated processor is going on worse process than the competitors.

>Can they build the x86? How do they work around a license?

Instruction sets are not patentable, however lots of implementation details are. 20 years from filing or 17 from issue is quite a long time, but those patent expire at some time. I'm pretty certain that *ALL* patents on 80486 are expired, also patents on Pentium would be expiring soon.
Pentium PRO isn't far behind.
I'm pretty certain that they could work around the patent minefield of things done later than that, but it wouldn't be optimum design since optimum designs are probably patented.
Or Intel sues and nVidia counter sues, and Intel decides that its not worth to stop shipping 50% of their volume to see this through, and they work out the cross licensing deal.