On December 9, 2024, Google Quantum AI published results in Nature showing that their Willow superconducting processor had crossed a critical milestone in quantum error correction: operating a surface code below the error threshold for the first time.

The central achievement is 1 logical qubit realized as a distance-7 surface code memory on a 105-qubit processor. The code uses 101 physical qubits (49 data qubits, 48 measure qubits, and 4 leakage removal qubits), with each step up in code distance suppressing the logical error rate by more than a factor of two.

Key results from the Willow experiment:

• 1 logical qubit — distance-7 surface code on 105 physical qubits
• Λ = 2.14 ± 0.02 — logical error is halved each time code distance increases by 2 (below-threshold confirmed)
• Logical error per cycle: 0.143% ± 0.003% at distance 7
• Logical lifetime: 291 ± 6 μs — 2.4× the lifetime of the best physical qubit (119 μs)
• Real-time decoder at distance 5: average latency 63 μs, stable over 10⁶ error-correction cycles
• High-distance repetition codes up to distance 29 run for 5.5 hours, revealing a rare correlated error floor at ~10⁻¹⁰ per cycle

Below-threshold operation means that adding more physical qubits to the logical qubit always reduces logical error — a necessary condition for practical fault-tolerant quantum computing. Prior to Willow, no quantum processor had definitively demonstrated this property in a surface code.

The distance-7 logical qubit also exceeded break-even: its coherence lifetime surpassed that of every one of its 101 constituent physical qubits, demonstrating that error correction is already providing a net benefit rather than merely catching up to added noise.