Skip to content
RESEARCH

Google’s Quantum Chip Beat the Classical Limit in a Grid Game

A four-qubit game offers a clear illustration of quantum behavior, without proving that ordinary computers have been beaten at everyday work.

Published Source report Sources checked
Reading tools for text size, copying, printing and your reading list
My reading list
Michel Devoret wearing glasses and a headset microphone at a Copenhagen lecture.
Michel Devoret, a corresponding author of the Willow study, at a December 2025 lecture in Copenhagen. Archival photograph; it does not show the experiment.2025 archival author photoChristian Ursilva

Google’s Willow processor won a carefully defined mathematical game 98.3% of the time, exceeding the 88.9% ceiling for classical players who cannot communicate during a round. The study, published in Science on November 19, reports results first made public as a preprint in December 2025. The grid game used just four .[1][2]

A puzzle with incompatible demands

Two players receive different instructions: one fills a row of a three-by-three grid, the other a column. Entries are +1 or −1. Rows must multiply to +1; columns to −1. Their answers must agree where the two intersect. Under uniformly random assignments, even the best shared classical strategy succeeds only eight times in nine. Two pairs of qubits let the players do better by following a quantum measurement strategy.[3]

Here is the contradiction in arithmetic: multiplying all three required row totals gives +1, while multiplying all three column totals gives −1. Both calculations would include exactly the same nine entries. A single completed grid cannot satisfy both. The quantum game does not produce that impossible grid; it tests how two partial answers agree.

Google says the tasks exposed subtle errors missed by random-circuit benchmarks, suggesting a useful additional test for improving quantum hardware.[2]

Bigger experiments, different claims

The researchers separately tested a multiplayer parity game with up to 71 qubits, reporting performance above its classical bound. Another task used up to 105 qubits to find a hidden linear function. These are different experiments, not three sizes of the same magic-square game. Crucially, the preprint labels its finite-size classical circuit-depth comparison for the latter task a conjecture.[3]

The theoretical foundation explains why that distinction matters. Sergey Bravyi, David Gosset and Robert König proved that certain hidden-linear-function problems can be solved by quantum circuits whose number of sequential layers stays fixed as problems grow. Classical circuits built from gates taking a bounded number of inputs require increasing depth. Their construction uses local quantum gates on a two-dimensional grid. This is a precise separation between constrained models of computation, not a stopwatch contest between an unrestricted laptop and a laboratory processor.[4]

An advantage under one rulebook should not be read as a victory under every rulebook. The useful question is which resources were restricted and which outcome was measured, rather than simply which machine won.

The newly published paper gives researchers another way to judge how well a chip can coordinate its qubits. It does not show a faster way to solve an everyday computing problem.[5]

Keep your reading trail

Evidence & context

5 linked sources

Science research publication, Google Research announcement and cited preprints.

How to read the evidence and dates

An announcement records what an organization stated. An experiment reports an observation. A simulation or resource estimate depends on its assumptions. QubitWire’s analysis connects these sources; the labels are not quality scores.

Published is the article’s first QubitWire publication. Updated marks a substantive content change. Sources checked records a review of the underlying evidence. An older paper can support a new explanation, and another press release about the same work is not independent replication.

Sources

  1. Science on November 19
  2. Google Research
  3. Research paper
  4. Original theoretical research
  5. Preprint history
READ NEXT

Continue exploring

All Research stories
  1. More coverage of Google

    Willow test targets quantum errors that linger across time

    On Google’s Willow chip, a qubit that slips into a higher energy state can disrupt several rounds of error correction. Giving each qubit a turn to reset helped.

    Published · Source report · Sources checked · About 1 min read
  2. More coverage of Google

    Google’s AI finds a better way for quantum computers to find every match

    A revised study reports fewer search rounds for a specific mathematical task. The distinction between one answer and every answer explains both the promise and the limit.

    Published · Source report · Sources checked · About 3 min read