The article that gp linked mentions that it's stored in non-volatile memory that supposedly is "programmable" only once. Obviously, it depends on the chipset, but how is non-reversibility guaranteed in this case?
The bit cell is programmed by applying a high-voltage pulse not encountered during a normal operation across the gate and substrate of the thin oxide transistor (around 6 V for a 2 nm thick oxide, or 30 MV/cm) to break down the oxide between gate and substrate. The positive voltage on the transistor's gate forms an inversion channel in the substrate below the gate, causing a tunneling current to flow through the oxide. The current produces additional traps in the oxide, increasing the current through the oxide and ultimately melting the oxide and forming a conductive channel from gate to substrate.
So, basically, they intentionally apply an out-of-spec voltage on the cell's output port, overloading the gate and causing a permanent short to ground. The cell always reads as 0 afterwards.
I don't see the "non-volatile" part at first, sorry about that. I guess non-volatile just means the data persists across resets, not necessarily that the fuses are stored in flash or something that can be modified.
