Quantum sims run 13,000x faster
- Google Quantum AI said its Willow processor ran the new Quantum Echoes algorithm 13,000 times faster than the best known classical method. - The result, published in Nature on October 22, 2025, measured a many-body physics signal linked to molecules, magnets and black holes. - Outside researchers said the benchmark is narrow and not yet a useful chemistry calculation. (nature.com)
Quantum computers are machines built to mimic quantum behavior directly, and Google says one of them just beat a supercomputer on a checkable task. (blog.google) (nature.com) The company said its 105-qubit Willow chip ran a new method called Quantum Echoes 13,000 times faster than the best known classical algorithm on Frontier, one of the world’s fastest supercomputers. Google and collaborators published the result in Nature on October 22, 2025. (blog.google) (nature.com) The task was not drug screening or fusion design itself. It was a physics simulation that reconstructs a hard-to-measure signal in quantum many-body systems from correlations spread across space and time. (nature.com) In plainer terms, the experiment tracks how a disturbance ripples through a crowded quantum system, like watching one bump on a pool table spread through every ball at once. Those patterns can help researchers study molecules, magnetic materials and other systems where quantum effects dominate. (blog.google) (nature.com) Google called the result the first “verifiable quantum advantage,” meaning the answer can be reproduced and checked rather than taken on faith. The company said that matters because earlier quantum speed claims often relied on benchmarks with less obvious scientific use. (blog.google 1) (blog.google 2) Nature’s news coverage said outside researchers were more cautious. Several said the benchmark is specialized, and some questioned how directly it maps onto practical chemistry or materials calculations people actually want solved. (nature.com) Google itself tied the technique to future applications rather than present products. In company posts, it said the same kind of simulation could help learn the structure of systems in nature, from molecules to magnets to black holes. (blog.google) That distinction matters for the “drug discovery and fusion” framing around the result. The published paper demonstrates a faster way to compute a specific quantum observable, not a finished workflow for designing medicines or modeling a fusion reactor. (nature.com 1) (nature.com 2) The larger race is over whether quantum machines can handle simulations that overwhelm ordinary computers because nature itself follows quantum rules at the atomic scale. Reviews in Nature and drug-discovery literature have long pointed to molecular simulation as one of the clearest possible uses, but they also note that hardware and error rates remain limiting. (nature.com 1) (nature.com 2) So the cleanest way to read Google’s 13,000-times figure is as a benchmark on one verified physics problem, published with peer review and greeted with skepticism about scope. It is a concrete speed claim, but not yet a claim that quantum computers are routinely doing chemistry better than classical ones. (nature.com) (nature.com)