Whatever it turns into it had better have massive dynamicly adaptive redudency (fault detection and correction).

When things are this small things like alpha particals and simple continuus erosion defects will make individual parts fail. With some defects being there from the very start. Depending on the functionality/complexity of the logic it might get very hard to have an inbuilt mechanism that checks correct functioning so that correction can be applied.

Our brains have something like a 1000% redundancy built into them as well as each component (cell) having some self repair capability internally and externally.

It will be interesting to see how even simple 'self-fixing' mechanisms will be done.

Now Microsoft will be able to blame the CPU for random crashes.

could someone give TSMC a ring?