Google's Willow chip runs 13,000x faster

Quantum computing is usually where big speed claims go to get messy. A machine does one weird benchmark fast, everyone argues about the comparison, and the practical payoff stays fuzzy. Google is trying to move that conversation. On October 22, 2025, its Quantum AI team said the Willow chip ran a new algorithm called Quantum Echoes with what it calls the first *verifiable* quantum advantage — about 13,000 times faster than the best known classical approach on a top supercomputer. (blog.google) ### What actually changed? The new part is not just “quantum was faster.” Google says Willow ran a specific algorithm on real hardware and produced an answer that can be checked by repeating the experiment on quantum machines of similar quality. That is the “verifiable” piece, and Google is treating it as the step beyond earlier supremacy-style demos that were harder to independently validate. (blog.google) ### What is Quantum Echoes? It is Google’s name for an algorithm built around out-of-time-order correlators, or OTOCs. Those are tools physicists use to track how information spreads through a quantum system. In plain English, you disturb part of a system, reverse the evolution, and look for an echo-like signal that reveals hidden structure in the dynamics. The Nature paper says these measurements stay sensitive to many-body quantum behavior at long timescales. (blog.google) ### Why is “verifiable” such a big deal? Because quantum advantage claims usually hit the same wall — if a quantum computer solves something classical machines cannot, how do you know the answer is right? Google’s pitch is that this task can be checked by rerunning it and comparing the resulting signal, rather than asking everyone to simply trust a black-box output. That does not end debate, but it does make the claim sturdier than a one-off benchmark no one can audit. (blog.google) ### Where does the 13,000x number come from? Google compared Willow’s performance on Quantum Echoes with the best classical algorithm it could identify running on Frontier-class hardware. The company says Willow executed the task about 13,000 times faster. That is a big number, but the catch is that it applies to this exact algorithm and comparison setup — not to “all computing,” and not even to most useful scientific workloads yet. (blog.google) ### What does Willow itself add? Willow is Google’s 105-qubit superconducting processor, and this result leans heavily on hardware quality. Google says the chip can run very fast gates, high-fidelity readout, and huge numbers of measurements quickly enough to pull weak signals out of noise. The team says the project involved one trillion measurements, with millions of Quantum Echoes measurements completed in tens of seconds. (blog.google) ### Is this useful, or just another benchmark? Google is trying hard to connect it to something real. The company says Quantum Echoes can help learn the structure of systems ranging from molecules to magnets, and it paired the announcement with a proof-of-principle molecular geometry experiment tied to NMR-style data. That is more concrete than random circuit sampling. But it(blog.google)ine. (blog.google) ### Why are researchers still skeptical? Basically, because quantum history has trained them to be. Nature’s coverage framed the result as another quantum advantage claim that still leaves open the hard question of realistic utility. A speedup on a carefully chosen physics task is not the same thing as a broadly useful machine, and classical algorithms often impr(blog.google)this time. (nature.com) ### So what’s the bottom line? This is probably the strongest version yet of Google’s old quantum promise: not just faster on a toy task, but faster on a checkable algorithm with a plausible scientific use case. But “plausible” is the key word. Willow did not suddenly make quantum computing commercially ready. What changed is narrower and still important — Google gave the field a better argument that quantum advantage can be both real and testable. (blog.google)