Google's Willow chip just pulled a magic trick quantum computers have needed for decades 🪄

Google's Quantum AI team has demonstrated that a technique called magic state cultivation can slash errors by roughly 40-fold on its Willow processor, according to a paper published in Science on Thursday. The company said the method achieved a record logical fidelity of 99.99% for T-states, a key resource for running the harder operations a useful quantum program needs but that standard error correction cannot directly protect. Only about 8% of attempts passed every check, with the chip discarding more than nine of every 10 tries to keep one clean state.

Magic states function as single-use fuel for the non-Clifford operations quantum algorithms require, and producing them has historically consumed up to 95% of a quantum computer's total capacity, according to Google. The conventional remedy, magic state distillation, was laid out by physicists Sergei Bravyi and Alexei Kitaev in 2004 and works by boiling many noisy magic states down into a few clean ones, but it demands sprawling qubit "factories" that do nothing else. Cultivation, first proposed by Google researchers in 2024, instead behaves like a filter, injecting a rough magic state into a tiny patch of the chip, running a chain of checks, discarding flawed attempts, and growing the survivors. That earlier paper suggested further distillation "may never be needed in practice."

Until this experiment, cultivation had only been tested in theory and computer simulations. The advance is one of the inputs behind Craig Gidney's 2025 estimate that RSA-2048 encryption could be broken with under a million noisy qubits, a 20-fold drop from the 20 million he estimated in 2019, a result that also frames the long-running debate over the encryption protecting bank logins and Bitcoin wallets. Google tightened its public posture on quantum in March with a Quantum AI whitepaper co-authored with Ethereum Foundation researcher Justin Drake and Stanford cryp, outlining how the method fits into broader plans for fault-tolerant hardware.

Quantum computers run on qubits, quantum bits that can hold a blend of 0 and 1 at once and explore many answers in parallel, but they are exceptionally fragile: a little heat or electrical noise can flip a qubit and wreck a calculation, which is why error correction, spreading one piece of information across many qubits so mistakes can be caught and fixed, dominates the field. Google's Willow result suggests a path to cheaper, more useful machines that can spend less capacity producing fuel and more running actual programs, including the day they could be used to design drugs and new materials.