Researchers at Google Quantum AI have successfully demonstrated contextuality, a fundamental quantum phenomenon, operating across systems containing up to 105 qubits. The findings, published Thursday as the cover story in Science, employed Google's superconducting quantum processors to test this property at an unprecedented scale.
Contextuality refers to a quantum behavior where measurement outcomes are not predetermined but instead depend on which other measurements are performed in conjunction with them. Prior demonstrations of this phenomenon had been largely confined to systems containing only a handful of particles.
In a practical demonstration, Google's quantum system achieved a 98.3% success rate in playing the Mermin-Peres magic square, a two-player coordination game. This substantially surpassed the maximum win rate achievable through classical strategies, which tops out at 88.9%. A separate experiment involving 105 qubits similarly outperformed classical approaches operating under comparable resource constraints.
Beyond validating quantum properties, Google characterizes these tests as a refined diagnostic tool for identifying subtle hardware defects that conventional benchmarking methods overlook. The company suggests this capability could facilitate progress toward practical applications in chemistry and materials science, though the current experiments did not involve simulations of either field.
The researchers acknowledge significant caveats in their work. The paper makes no claim of unconditional proof that quantum computers outperform classical systems broadly, and notably, the quantum advantage demonstrated in the coordination game diminished as additional qubits were incorporated into the system. The research initially circulated as a preprint in December 2025.
Achieving hardware free from errors remains the central challenge facing the quantum computing industry. Google has publicly committed to delivering a functional, error-corrected quantum computer by 2029—a timeline Microsoft matched following its June unveiling of the Majorana 2 chip. IBM's acquisition of HRL coincided with the company's demonstration of a self-correcting chip, while Nvidia has conducted simulations of a 104-qubit error-corrected system. Capital continues flowing into the sector, exemplified by Quantinuum's June IPO that raised $1.68 billion.
Source: The Next Web



