Makes me wonder... if x86 is such a bad thing, why has nobody yet produced an x86 chip where you can switch off the translation layer?
It seems like it should be possible to bypass the translation layer on Transmeta chips, but you'd probably need internal Transmeta documentation to write any code for it. More generally, I suspect the reasons for strictly supporting translation include:

- The target internal instruction set(s) being heavily optimized for the purpose of quickly running translated x86 programs on a specific microarchitecture, and thus not terribly suitable for direct use

- Intel/AMD/etc. not wanting to reveal those internal instruction set(s) to competitors

- Intel/AMD/etc. not wanting to provide the additional documentation, support, and testing that would be needed to make a "native mode" useful

- Intel/AMD/etc. not wanting to create a situation where they end up having to extend support for another entire instruction set in every chip because too many customers start to depend on its presence