What happened
Google's Quantum AI team ran its Willow chip through the magic square game, a two-player non-local benchmark that caps classical strategies at a win rate of 88. 9%. Willow cleared 98.
3%, per the writeup CryptoBriefing published Wednesday. That gap matters. Magic square isn't a bespoke supremacy stunt where the goalpost moves every time a classical team optimises.
The ceiling is a mathematical fact about local hidden-variable strategies, proven decades ago by Mermin and Peres. A device that beats it is doing something a classical computer cannot, by construction. Willow is the same 105-qubit chip Google introduced in late 2024 to demonstrate below-threshold error correction.
The magic square result extends that line of work: not more qubits, but cleaner ones, run on a protocol with a hard reference line. Google's own researchers framed the headline carefully. The chip beat the classical bound.
It did not run Shor's algorithm. It did not factor a number anyone would pay to keep secret. The gap between this benchmark and a cryptographically relevant computation is still measured in orders of magnitude of logical qubits.
Why it matters
Crypto's exposure to quantum computing runs through two specific primitives. ECDSA, the signature scheme behind Bitcoin and Ethereum wallets, falls to Shor's algorithm given a sufficiently large, fault-tolerant quantum machine. SHA-256, the hash function underpinning Bitcoin mining and address derivation, weakens under Grover's algorithm, roughly halving its effective security.
