The main reason why this hasn't been tried is because the microprocessor itself is not very heat tolerant.
There are some very heat-tolerant semiconductor materials. The only one I can think of offhand is silicon carbide. That is used in some electronics for automobiles.
The trouble is that silicon carbide crystals have a great many defects. So to get any yield at all, chips have to be very small. Goodbye microprocessors. Hello small-scale integration. So back to million-dollar mainframes with lots of tiny chips in them, and long wires between parts... which, of course, limits speed.
Someday, perhaps what you propose will be possible, as techniques for growing silicon carbide crystals improve. But not yet.
John Savard