On December 9, 2024, Google Quantum AI published landmark results in Nature: its 105-qubit Willow chip demonstrated below-threshold quantum error correction (QEC), a theoretical goal since the mid-1990s. The core question was whether adding more qubits actively reduces logical error rates — Willow answered definitively yes.

As the surface code distance scaled from d=3 to d=5 to d=7, the logical error rate per cycle dropped exponentially with error suppression factor Λ=2.14. At d=7, the logical error rate reached 0.143% per cycle. Most crucially, the logical qubit lifetime exceeded the best physical qubit lifetime by 2.4×, surpassing the 'breakeven point' for the first time in superconducting hardware — proof that a logical qubit can outperform the physical qubits it is built from.

On a random circuit sampling (RCS) benchmark, the equivalent computation would require Frontier (the world's fastest classical supercomputer) approximately 10²⁵ years — completed by Willow in under 5 minutes. This result serves as the technical foundation for Google's 2029 goal of 1,000 logical qubits.

Source: https://www.nature.com/articles/s41586-024-08449-y