Google's Willow Quantum Processor Achieves Breakthrough: 5 Minutes vs 10 Septillion Years

📢 The announcement that changed everything

On December 9, 2024, Hartmut Neven — founder and head of Google Quantum AI — officially announced Willow, the company's latest quantum processor. With 105 superconducting transmon qubits, the new-generation chip achieved two milestones the quantum community had been awaiting for decades. First, it achieved exponential error reduction during scaling — known as “below threshold.” Second, it completed a standardized performance test in less than five minutes, while the most powerful classical supercomputer would need 10²⁵ years — that is, 10 septillion years.

The results were published in the journal Nature (volume 638, pp. 920–926, February 2025), titled “Quantum error correction below the surface code threshold.” Publication in one of the world's premier scientific journals confirmed that this was not marketing, but rigorously peer-reviewed research.

⚠️ The error problem in quantum computers

Qubits — the computational units of quantum computers — are extremely sensitive. They interact with their environment through phenomena such as decoherence, amplitude damping, and dephasing. Each time you add more qubits to a system, noise also increases — causing the system to become more “classical” rather than more quantum.

The idea of Quantum Error Correction (QEC) was introduced in 1995 by Peter Shor. The basic principle is analogous to classical error correction: you encode the information of one logical qubit across multiple physical qubits, so that errors can be detected and corrected without destroying the quantum information.

However, there exists a critical limit. The quantum threshold theorem proves that error correction works only if the rate of physical errors falls below a certain threshold — estimates from 2004 place it at 1–3%. Above this limit, the correction process introduces more errors than it eliminates. For nearly 30 years, no experimental system had managed to operate reliably below this threshold at practical scale.

📈 Willow's exponential improvement

The Google Quantum AI team tested progressively larger arrays of physical qubits on Willow: from a 3×3 grid of encoded qubits, to 5×5, and finally to 7×7. At each scaling step, the error rate was halved. This exponential reduction — each size increase brings proportionally greater improvement — is exactly what “below threshold” means.

Additionally, the team achieved a “beyond breakeven” result: the lifetime of logical qubits exceeded that of individual physical qubits. This constitutes unfakable evidence that error correction genuinely improves the overall system. Furthermore, it was one of the first real-time error correction experiments on a superconducting quantum system — crucial for any practical computation.

The code used was the surface code, a topological QEC code. First introduced theoretically by Alexei Kitaev in 1997 as the toric code, it was later adapted into a surface code with boundaries — ideal for two-dimensional implementation on superconducting chips. Today it is considered the most realistic path toward scalable quantum computers.

⏱️ 5 minutes versus 10 septillion years

Willow's second major demonstration involved the Random Circuit Sampling (RCS) test — a standardized benchmark first used by Google's own team. The RCS test checks whether a quantum computer can do something no classical computer can — it essentially serves as a “gateway” to quantum computation.

Willow completed the RCS test in less than five minutes. For comparison, Google estimated that the Frontier supercomputer — one of the most powerful classical systems in the world, located at Oak Ridge National Laboratory — would need 10²⁵ years. This number vastly exceeds the age of the Universe (approximately 13.8 billion years) and every known physical timescale.

It is worth noting that the estimate was conservative: Google granted Frontier full access to secondary storage (hard drives) with no bandwidth constraints — a generous and unrealistic concession. Even so, the gap remains astronomical.

🔄 From Sycamore to Willow — the evolution

Willow did not appear suddenly. Google Quantum AI was founded in 2012 by Hartmut Neven in Santa Barbara, California. The first major milestone came in October 2019, when the Sycamore processor — with 53 transmon qubits — completed a test in 200 seconds that, according to Google, would require 10,000 years on the Summit supercomputer. IBM disputed the claim, arguing that Summit could manage it in 2.5 days, but the research itself was published in Nature and became a landmark.

Compared to Sycamore, Willow brings enormous improvement — both in qubit count (53 → 105) and in quality. T1 times — which measure how long a qubit can maintain a quantum state — approach 100 microseconds (µs), a fivefold increase over the previous generation. Willow was fabricated at Google's nanofabrication facility in Santa Barbara — one of the few units in the world built from scratch exclusively for this purpose.

🎯 Why it matters now

The significance of Willow lies not just in the impressive numbers. It lies in the fact that for the first time, a quantum system practically demonstrated that errors can be reduced by increasing size — something that until now was only a theoretical promise.

This paves the way for the long-term goal: building large-scale fault-tolerant quantum computers capable of executing algorithms impossible for classical systems. According to Google's roadmap, the next step is demonstrating a “useful, beyond-classical” computation — meaning a quantum calculation that not only surpasses classical computers but solves real problems in fields such as pharmaceuticals, battery materials for electric vehicles, and nuclear fusion.

Willow is strong evidence that the path toward commercially relevant quantum applications is no longer theoretical — but engineering. It doesn't require new physics. It requires better engineering. And that is exactly what Google demonstrated.