When Google revealed its Willow quantum chip in late 2024, the headlines were filled with claims about a processor that could complete in minutes a task classical supercomputers would not finish in the lifetime of the universe. Behind the dramatic comparisons lies a genuinely significant technical step forward. Willow’s most important achievement is its evidence that error rates can actually decrease as systems get larger—an essential requirement for building practical, fault-tolerant quantum computers.
This article explains what Willow is, what Google has actually demonstrated, the limitations of the current results, and why the work could be a stepping stone toward practical quantum applications.
What Willow Is
Willow is a superconducting quantum processor built by Google’s Quantum AI team in Santa Barbara, California. It contains about 105 physical qubits arranged on a chip a few centimetres across, operating at temperatures close to absolute zero inside a dilution refrigerator. The qubits themselves are tiny electrical circuits that behave according to quantum mechanics and are controlled using carefully tuned microwave pulses.
Although qubit count is often used as a headline number, it is not the primary reason Willow is interesting. In most existing quantum devices, adding more qubits introduces more noise, making them less reliable. Willow’s difference lies in its ability to combine those physical qubits into logical qubits using error correction, and to show that the logical error rate improves—rather than worsens—when the size of the error-correcting code is increased. This is a key sign that the hardware is operating in the so-called “below-threshold” regime, where scaling can, in theory, lead to exponentially better reliability.
The Two Big Claims
Google’s announcement rested on two central results.
First, Willow demonstrated that larger error-correcting code sizes led to lower logical error rates. This is the practical confirmation of a theory that has underpinned quantum computing for decades: if your hardware is good enough, you can use more qubits to make your logical qubits much more stable.
Second, the team ran a benchmarking experiment called random circuit sampling (RCS), completing it in under five minutes. The claim was that a classical supercomputer would take an impossibly long time—measured in absurdly large spans, far beyond the age of the universe—to produce the same output with the same level of accuracy. This is not a practical task but rather a stress test to show how quickly quantum circuits can reach a complexity that defies brute-force classical simulation.
Why Error Correction Is So Hard
Qubits store information in delicate quantum states. They can exist in superpositions and become entangled, properties that make quantum computing a powerful tool. But this same delicacy makes them extremely vulnerable to disturbance—whether from heat, stray electromagnetic fields, or imperfections in control signals.
Unlike classical bits, quantum states cannot be copied directly. Quantum error correction works by spreading one logical qubit’s information across many physical qubits, and by using repeated measurements to detect errors without destroying the quantum information itself. The “threshold theorem” says that if your physical qubits are good enough, you can suppress logical errors exponentially by increasing the code size. Achieving this in real hardware is one of the main barriers to fault-tolerant quantum computing—Willow’s demonstration suggests that it is finally possible on at least a small scale.
The Five-Minute Benchmark in Context
Random circuit sampling is not a useful application on its own. It is designed to be hard for classical computers to simulate, making it a good test of quantum hardware’s raw capability. Google ran a similar demonstration with its earlier Sycamore processor in 2019, but Willow’s version is larger and more robust, and it was run under stricter conditions.
The famous “five minutes versus septillions of years” comparison is meant to illustrate the vastness of the quantum state space the chip can explore. However, it is not evidence that Willow can perform everyday commercial tasks faster than classical computers—it is simply a sign of progress on synthetic benchmarks.
Hardware Advances
Willow’s success is partly due to improvements in manufacturing and design. Google has been working on refining the materials used for superconducting circuits, reducing unwanted energy loss, and enhancing coherence times. The layout of the chip and the control electronics have been optimised to reduce interference between qubits and to allow larger code blocks to function reliably.
These manufacturing improvements are important because future error-corrected quantum computers will need to be built at scale, with consistent performance across thousands—or eventually millions—of physical qubits. Willow is not just a one-off experiment; it is a sign that Google can produce chips that meet the demanding specifications of quantum error correction.
How Far This Is from Practical Use
A machine with 105 physical qubits, even when used for error correction, is still far from the thousands of logical qubits thought necessary for truly useful applications, such as simulating large molecules or breaking widely used cryptography. The step from a handful of stable logical qubits to thousands will require significant progress in fabrication yields, control systems, and cryogenic infrastructure.
That said, Willow’s “below-threshold” behaviour changes the outlook. It suggests that each generation of hardware can improve logical error rates simply by increasing the size of the error-correcting code, as long as physical performance continues to improve.
What Changes for Researchers
For researchers, Willow makes it worthwhile to start experimenting with error-corrected logical circuits rather than focusing purely on noise-tolerant tricks. It enables small demonstrations of logical memory and primitive logical gates, providing a testing ground for the software tools, compilers, and algorithms that will eventually run on large-scale quantum computers.
This also highlights engineering priorities, including better fabrication, cleaner materials, faster and more accurate measurements, and classical control systems that can handle the massive data throughput required for real-time error correction.
Potential Applications
There are three primary areas where error-corrected quantum computers can make a significant difference.
Quantum simulation could allow chemists and materials scientists to model complex molecules and materials in ways that are impossible for classical computers, potentially accelerating discoveries in energy storage, catalysis, and drug design.
Optimisation problems in logistics, finance, and network design might benefit from quantum algorithms that can explore vast search spaces more efficiently under certain conditions. These would not replace all classical approaches but could give significant advantages in specific, well-structured problems.
Machine learning may also be enhanced, with quantum subroutines for tasks such as sampling and linear algebra potentially speeding up certain training and inference steps, or enabling new types of generative models.
Security Implications
Willow’s debut has prompted questions about cryptography. Current encryption systems, such as RSA and elliptic-curve schemes, could be broken by a large enough quantum computer running Shor’s algorithm. However, Willow is still far from having the number of logical qubits and low enough error rates required for that. The key takeaway for security is that work on post-quantum cryptography—encryption methods resistant to quantum attacks—should continue without delay.
Position in the Global Race
Google is one of several major players in the race to build a fault-tolerant quantum computer. IBM is scaling up its superconducting qubit systems, while Microsoft is pursuing a different type of qubit entirely. Meanwhile, various start-ups are working on alternative architectures, such as photonic or neutral-atom systems. Willow’s results have been welcomed across the field as an encouraging sign, even though all agree there is still a long road ahead.
What to Watch for Next
Key future milestones include:
- Demonstrations of logical gates that improve as code size increases, not just logical memory.
- Sustained operation of error-corrected qubits through full sequences of gates in small algorithms.
- Scaling up to chips with more qubits while maintaining performance and fabrication reliability.
These results would indicate that the progress observed in Willow is translating into the capabilities required for real-world algorithms.
Bottom Line
Willow is not a general-purpose quantum computer, and its five-minute benchmark is not a helpful application in its own right. However, it is strong evidence that a real device can operate in a regime where error correction works as theory predicts. This makes the goal of fault-tolerant quantum computing more tangible.
The next steps are to leverage that success by implementing robust logical gates, increasing the number of logical qubits, and ultimately running algorithms that outperform classical computers on significant real-world problems. That will not happen overnight—but with Willow, it is easier than ever to believe it is possible.
You are here: home »