Google’s Willow chip advances quantum error correction

Google has introduced Willow, a 105-qubit quantum processor that improves the reliability of stored quantum information as more physical qubits are used to protect it.

Published online in Nature on 9 December 2024, the result demonstrates progress in quantum memory, not a finished general-purpose quantum computer.

More qubits, fewer logical errors

Quantum error correction spreads one logical qubit across physical qubits. Extra hardware helps only below an error-rate threshold. Otherwise, additional mistakes overwhelm the protection.

Google’s researchers reduced logical errors roughly twofold at each increase in surface-code distance, from three to five to seven. This below-threshold behaviour is essential for scaling.

The largest memory experiment used 101 physical qubits and recorded a logical error rate of 0.143% per correction cycle. Its logical memory lasted about 2.4 times as long as its best individual physical qubit.

A separate sampling benchmark

Google also reports random-circuit-sampling results. Its 9 December specification sheet lists separate Willow chips for the error-correction and sampling experiments. The sampling run used 103 qubits.

Random circuit sampling tests a processor using specially constructed quantum circuits. It does not demonstrate an advantage on a useful commercial workload, a distinction Google makes in its announcement.

Separate Willow chips: Chip 1 tests error correction, with fewer logical errors as codes grow; Chip 2 tests random circuit sampling with 103 qubits and depth 40. — Google’s launch specification sheet lists separate Willow chips for error correction and random circuit sampling. Graphic: Sulat.

What still needs work

Google’s technical explanation identifies remaining obstacles. Separate repetition-code tests, which protect against a narrower class of errors, hit an error floor whose cause remains under investigation. Decoding delays can also slow some error-corrected operations, even when the decoder keeps pace with incoming measurements.

Preserving this performance on much larger processors remains an engineering challenge. Willow strengthens the case for scalable error correction; running large, reliable algorithms is still the task ahead.

Featured illustration: Sulat. Official logo: Google. Decorative lattice, not a hardware diagram.