Arguably, it came true with the Pentium Pro more than 20 years ago. Ever since then x86 processors have interpreted x86 instructions by translating them into multi-step sequences of microinstructions under software control. The microcode is basically a just-in-time translator with specialized hardware assists. (And it is very much under software control -- that's why the microcode needs periodic microcode patches to fix bugs.) The internal architecture today looks much more like a RISC/VLIW machine than a classical x86 processor.