Google's new Willow quantum chip has broken new ground in an important random circuit sampling ... [+] benchmark, an important development in Google's roadmap for fault-tolerant quantum computing.PixabayGoogle has chalked up several amazing quantum computing records with its newest quantum 105-qubit superconducting chip called Willow. This performance is no surprise, considering Googles heritage of record-setting quantum chips, reaching back to Foxtail in 2017, Bristlecone in 2018 and Sycamore in 2019.Google announced Willow last month, and I think it is necessary to reemphasize the importance of this research after Jensen Huang, CEO of Nvidia, recently remarked that quantum computing likely wont be useful for another 20 years. Granted, there remains a lot of ground to cover to reach fault tolerance, which will be critical for many practical applications, but there has also been a lot accomplished in quantum in just the past 12 months. Marketplace evidence, research results (including qubit fidelity close to what is needed for fault tolerance) and the roadmaps of many quantum computing companies indicate that useful quantum technology is much closer than Huang believes.Read on for more on how the new Willow chip performed on the random circuit sampling benchmark. I also discuss what may be the most important piece of this development for future quantum fault tolerance, the results of applying a new error-corrected surface code. To provide more context, Ill also share historical perspective from Professor John Martinis, who led some of the most important work on earlier generations of Googles quantum chips, and how his work has now paid off just as he predicted with Willow.Willow Hardware And Software ImprovementsWillows performance across key metricsGoogleWillow has improved on earlier generations of Googles quantum chips in several ways. For starters, the use of tunable qubits and couplers in Willow has provided it with much faster gates and operations that help achieve lower error rates. This speed also allows hardware to be optimized or adjusted during operation. Variances in superconducting qubits can sometimes create high error rates, but tuners allow nonconforming qubits to be reconfigured and aligned with other qubits to eliminate errors.MORE FOR YOUNext up is the duration of quantum states. A major limitation of quantum computing has been the length of time qubits can maintain their quantum states. Willow has increased that time by 5x, from 20 microseconds to 100 microseconds. This allows more complex problems to be run.A third advantage of Willow is that Googles logical qubits can now function below the critical quantum error correction threshold. The QEC threshold arises from a theory developed in the 1990s, and until now it has been a barrier to efficient quantum computing. In the Willow chip, however, error rates are reduced by one-half as physical qubits are added in scale. Thanks to this, as Google increases the size of its surface code from 3x3 to 5x5 to 7x7 the encoded logical qubits maintain their coherence for longer times. Increasing grid size allows for more complex error patterns to be corrected, similar to more redundancy in classical error correction. It also means that logical qubits can maintain their quantum states longer than the underlying physical qubits.This leads me to the single most important part of Googles Willow announcement: Willow is the first quantum processor to demonstrate an exponential reduction in error rates as the number of qubits is increased. Traditionally, adding qubits causes the error rate to increase.Other factors necessary for fault-tolerant quantum computing have also been demonstrated by Google researchers. For one thing, having a repeatable performance over several hours without degradation is needed to run large-scale fault-tolerant algorithms and Willow has now demonstrated that capability.Benchmarking Quantum ProcessorsGoogle uses random circuit sampling as an ongoing benchmark to compare new experimental quantum processors against supercomputers running classical algorithms. It is important to point out that random circuit sampling is not useful as an application in itself; it is only a threshold test. But if a system fails to pass RCS, there is no need for further testing.Five years ago, the Google quantum research group claimed that the 53 superconducting qubits of its 54-qubit Sycamore chip (one qubit was faulty) had achieved quantum supremacy meaning that it outperformed comparable classical computing. Back then, Google researchers said they were able to complete a RCS benchmark computation in 200 seconds that theoretically would take a classical supercomputer 10,000 years to complete. IBM disputed the claim using calculations indicating it was possible for a classical computer to achieve the same results. However, it was eventually accepted by the quantum community that if Google had used all 54 qubits, it would have taken a classical supercomputer much longer than 10,000 years to equal Sycamores achievement.This year, in another quantum supremacy test, Google pitted the new 105-qubit Willow chip against the same RCS benchmark experiment that the Sycamore chip ran in 2019. Willow ran the RCS benchmark in under five minutes; it has been determined that todays best classical supercomputer would need 10 septillion years to run the same benchmark (thats a 1 followed by 25 zeros). In short, because Willow performs below the error correction threshold, it is able to conduct random circuit sampling far beyond what is possible with classical computers.If youre not familiar with quantum computing, these comparisons may seem confusing at first. But they are directly attributable to the number of qubits involved. The Willow chip has 105 qubits compared to Sycamore's 53. Each additional qubit results in an exponential increase in computing power, not a linear increase. The difference in the execution time between the tests in 2019 and the ones conducted in recent months today becomes understandable in this context. Because Willow has 52 more qubits than Sycamore, it has 2^52 (4.5 quadrillion) more computational states.Besides the increase in qubits, many other improvements have been made to quantum systems since 2019. Algorithms are a billion times better because of extensive experimentation by the large community of computer scientists in the ecosystem. Plus, quantum processors have improved significantly in various ways, including in the quality of qubits.Google's roadmap to fault-tolerant quantum computingGoogleFollowing its 2019 benchmark results, Google published a road map with a 10-year timeline for developing a large error-corrected quantum computer with 1,000 logical qubits using 1,000,000 physical qubits. As shown in the diagram above, the roadmap has six milestones; after its latest achievement with Willow, Google is now approaching the third milestone.For another perspective on the Willow chip, I recently discussed Googles achievement with Prof. John Martinis, who led the Google team that designed and tested the Sycamore chip. Prof. Martinis is currently working on a quantum startup called Qoloab with his cofounders Alan Ho (another Google veteran) and Prof. Robert McDermott.During that conversation, I recalled remarks that Prof. Martinis made about a yet-to-be-developed quantum computer chip for a Forbes article I published nearly five years ago. Googles plan is roughly to build a million-qubit system in about 10 years, with sufficiently low errors to do error correction, he said. Then at that point you will have enough error-corrected logical qubits that you can run useful, powerful algorithms that you now cant solve on a classical supercomputer. And maybe even at a few hundred qubits, with lower errors, it may be possible to do something special-purpose.Those remarks are very close to describing how Googles Willow chip has actually played out.How Long Until We See Commercial Quantum Applications?Google currently believes that it will be able to produce useful commercial quantum applications in the next five years or less. Many quantum scientists believe it will take at least another decade before quantum computers are able to handle world-affecting computations in areas such as climate change, drug discovery, materials science and financial modeling.Of course, Google is not the only company on this path. There is a great deal of experimentation and collaboration being done with logical qubits. One notable example is Microsoft, which has done exciting work with both Quantinuums H-2 trapped-ion processor and Atom Computings neutral-atom processor.Google acknowledges there are many challenges remaining. While the maximum code distance used in the Willow research was 7, to obtain the necessary error rate for fault tolerance would require a distance-27 logical qubit, which would need almost 1,500 physical qubits to create it. For quantum error correction, a higher distance means that an error code can handle more errors before it fails. A larger distance means the code has more layers of checks and balances that can detect and repair errors before they cause problems.That is just one of the many challenges that must be overcome to achieve fault tolerance. While some might believe Googles timeline is overly optimistic, I believe the company is on track. In another five years, fault tolerance will be a lot closer. And useful commercial quantum applications in some form or another should be quite doable.