22 Oct 2025

Google reports a superconducting chip run estimated to take 13,000 times longer classically

Quantum technologies
AI illustration · Illustrative scene, not a photograph of the specific device or experiment.

What was demonstrated

On its Willow superconducting processor, Google Quantum AI ran a benchmark circuit set measuring second-order out-of-time-order correlators (OTOC(2)). The run took about 2 hours. Google estimates the same task would take about 13,000 times longer on a classical supercomputer, compared with the best classical algorithm it could find. A set of second-order OTOC values measured on Willow is estimated to take about 3.2 years per data point to simulate on the Frontier supercomputer, according to a figure caption. The classical estimate rests on red-teaming by the authors, which they describe as about 10 person-years of work across nine classical simulation algorithms.

Why it matters

If the classical estimate holds, it suggests a superconducting processor can run a task that known classical methods would struggle to simulate. The authors also argue that the output is a verifiable expectation value, which another quantum computer or a natural quantum system could reproduce.

Caveats

  • The 13,000-times figure is an estimate against the best classical methods the authors tried, not a measured classical run. It could fall if better classical algorithms are found.
  • The task is a set of benchmarking circuits. The authors describe it as a route toward practical quantum advantage, not as a practical application that has been demonstrated.
  • The qubit count differs between parts of the work. The Google blog gives 103 qubits for the Quantum Echoes run and 65 qubits for the second-order OTOC analysis, and a figure caption cites 105 available qubits. The Nature article could not be opened for checking, so the published title and DOI are taken from the Google blog.

Technology profiles

Sources

  1. company announcement
    A verifiable quantum advantage
    Google Quantum AI. Google Research, 22 Oct 2025.
  2. preprint
    Constructive interference at the edge of quantum ergodic dynamics
    Dmitry A. Abanin, et al.. arXiv, 11 Jun 2025. DOI: 10.48550/arXiv.2506.10191.

Last reviewed 21 Oct 2026.