Operating below threshold means a quantum processor has achieved physical qubit error rates sufficiently low that quantum error correction works as theoretically predicted — adding more physical qubits to the error correction code exponentially suppresses the logical error rate. This is the crucial transition point that separates noisy intermediate-scale quantum (NISQ) devices, where errors accumulate faster than they can be corrected, from the path toward fault-tolerant quantum computing.
Google's Willow chip provided the landmark demonstration of below-threshold operation in December 2024. Using surface codes of increasing distance (d=3, d=5, d=7), Google showed that each increase in code distance cut the logical error rate roughly in half — the hallmark of below-threshold scaling. This was possible because Willow's physical two-qubit error rates (approximately 0.3%) were well below the surface code's ~1% threshold. Previous attempts with higher error rates had shown the opposite trend: larger codes performing worse, confirming the system was above threshold.
Being below threshold is necessary but not sufficient for practical fault-tolerant quantum computing. The logical error rate suppression factor per unit of code distance increase depends on how far below threshold the physical error rates are. At 0.3% physical error rate, the suppression is modest (roughly 2x per code distance step). At 0.1%, the suppression is much stronger, meaning fewer physical qubits are needed to reach a target logical error rate. Driving physical error rates further below threshold is therefore essential for making fault-tolerant quantum computing resource-efficient enough to be practical.