When the Microscopic World Learns to Think in Parallel: A Journey into Quantum Computing and Zuchongzhi 3.0
> Imagine being able to traverse every path of a maze simultaneously instead of trial-and-error one path at a time. This is not science fiction — it is what quantum computers do every day. In early 2025, a lab in Hefei, China, unveiled a 105-qubit machine called Zuchongzhi 3.0, capable of completing in seconds tasks that would take a classical supercomputer quadrillions of years — far longer than the age of the universe.
---
Chapter 1: From Switches to Spirits — The Magic of Qubits
The classical computer's 'clumsy' approach
Classical computers are built on bits — switches that are either 0 or 1, never anything in between. Everything from your phone to the internet is, at its core, countless switches flickering on and off. But for extremely complex problems — simulating how a drug molecule binds to a virus, or forecasting a month of weather — a classical computer must try path after path like a diligent ant, slowing to despair as the possibilities multiply.
Qubits: superposition in the microscopic world
In the quantum world, particles don't obey everyday rules. An electron can be here *and* there until observed — a phenomenon physicists call superposition. Think of a coin: in the classical world it lands heads or tails; in the quantum world it can be both at once until you grab it.
A qubit can be 0 and 1 simultaneously. Two qubits represent 00, 01, 10, and 11 at once; three qubits, eight states. Each added qubit doubles the number of simultaneous states:
With 105 qubits, Zuchongzhi 3.0 can process \(2^{105}\) states — a 32-digit number far exceeding the count of every grain of sand on Earth.
Entanglement: 'spooky action at a distance'
Qubits also exhibit quantum entanglement, which Einstein called "spooky action at a distance." Two entangled qubits remain correlated no matter how far apart: measuring one instantly reveals the state of the other. Countless experiments have confirmed entanglement is real, and it forms the basis of quantum communication — theoretically absolutely secure communication.
---
Chapter 2: The Duel — Zuchongzhi 3.0 vs. Willow
December 2024: a historic month
Within a single week, two teams on opposite sides of the Pacific released results, both at the 105-qubit scale:
- December 9, 2024: Google published its Willow chip in *Nature*, solving the quantum error correction problem that had challenged the field for nearly 30 years.
- December 17, 2024: Pan Jianwei's team at the University of Science and Technology of China (USTC) released Zuchongzhi 3.0 results on arXiv, matching 105 qubits with world-class performance.
- Willow: error-correction first — proving logical error rates fall as scale grows, the necessary path toward universal quantum computing.
- Zuchongzhi 3.0: scale first — exploring the current ceiling of superconducting quantum advantage.
Willow's breakthrough: error correction 'below threshold'
Qubits are fragile. They operate near absolute zero (~ -273°C), and even a stray electromagnetic wave or temperature fluctuation can collapse their carefully prepared superposition. Worse, more qubits historically meant more errors.
The solution is quantum error correction: spreading one logical qubit across multiple physical qubits so errors can be recovered algorithmically. But this created a paradox — more qubits for correction historically introduced more errors than they fixed.
Willow achieved, for the first time, "below-threshold" error correction: expanding the qubit array from 3×3 to 5×5 to 7×7, the logical error rate *fell* exponentially — dropping by a factor of about 2.14 with each added layer. The logical qubit's lifetime reached 291 microseconds, 2.4× the 119-microsecond lifetime of the physical qubits. Corrected quantum information outlived its raw components — a key step from lab to practicality.
Zuchongzhi 3.0: pushing quantum advantage
| Metric | Zuchongzhi 2.0 (2021) | Zuchongzhi 3.0 (2025) | Improvement | |---|---|---|---| | Readable qubits | 66 | 105 | +59% | | Coupled qubits | — | 182 | New | | Single-qubit gate fidelity | 99.7% | 99.90% | +0.2% | | Two-qubit gate fidelity | 99.2% | 99.62% | +0.42% | | Readout fidelity | — | 99.13% | New | | Coherence time | — | 72 μs | — |
In the random circuit sampling benchmark, Zuchongzhi 3.0 completed 83-qubit, 32-layer sampling a quadrillion times faster than the world's fastest supercomputer — if a supercomputer needed 1,000 years, Zuchongzhi 3.0 needs under a second. It is also a million times faster (six orders of magnitude) than Google's 67-qubit Sycamore result from October 2024.
Two routes, one dream
---
Chapter 3: What Can Quantum Computing Do for Us?
Today's quantum computers resemble early electronic computers — room-sized, specialist-operated, solving only specific problems. (ENIAC in 1946 weighed 30 tons, filled 170 m², and did 5,000 additions per second; your phone now exceeds the Apollo guidance computer.) Quantum computing is undergoing a similar transformation.
Key applications:
1. Drug discovery — Simulating molecular interactions at the atomic level, where classical simulation is theoretically possible but practically impossible. Where AI tools like AlphaFold predict structures, quantum computers can simulate dynamic binding processes. (Average new drug: 10 years, $1 billion.) 2. Materials science — Unraveling the mechanism of high-temperature superconductivity, a holy grail that could revolutionize energy transmission, maglev transport, and fusion. 3. Financial modeling — Portfolio optimization, risk analysis, and derivatives pricing; JPMorgan and Goldman Sachs are already exploring quantum algorithms. 4. Artificial intelligence — Quantum machine learning may accelerate AI training by orders of magnitude; hybrid quantum-classical architectures may become mainstream. 5. Cryptography — Shor's algorithm threatens RSA encryption, while quantum key distribution offers theoretically unbreakable communication; China has built a nationwide quantum communication network.
---
Chapter 4: The Road Ahead
Three stages of quantum development:
1. Quantum supremacy — beating classical computers on specific tasks (achieved by Google's Sycamore, China's Jiuzhang and Zuchongzhi series). 2. Quantum simulators — hundreds to thousands of qubits for problems like quantum chemistry (we are at this threshold). 3. Universal quantum computers — programmable, error-corrected machines; possibly 10–15 years away.
China's roadmap: On the superconducting route, the team plans surface-code error correction experiments at code distances 7, 9, and 11. On the photonic route, Jiuzhang 3.0 manipulates 255 photons, achieving speeds 10 quadrillion times faster than supercomputers on specific problems.
Global landscape: The US leads with Google, IBM, and Microsoft; China's USTC team is world-class in both superconducting and photonic routes; Canada's D-Wave specializes in quantum annealing; Europe advances ion-trap approaches. In March 2025, NVIDIA held its first "Quantum Day" and launched the NVIDIA Accelerated Quantum Research Center (NVAQC). CEO Jensen Huang, who had once predicted 20 years to commercialization, revised his outlook at GTC 2025, saying quantum computing "is approaching an inflection point."
---
Epilogue: On the Eve of a Revolution
Quantum computers will not replace your laptop or phone — like supercomputers, they will be specialized tools for problems classical machines cannot touch. But their impact could be profound: drug development shortened from 10 years to 1, breakthroughs in materials and energy, leaps in AI capability, and absolutely secure communication. These changes may unfold over the next 10–20 years.
When Bell Labs invented the transistor in 1947, no one foresaw the information age it would open. Today, we stand at a similar turning point — and Zuchongzhi 3.0 and Willow are the opening chapter of this revolution.
---
References
1. Gao, D., Fan, D., Zha, C., et al. (2025). "Establishing a New Benchmark in Quantum Computational Advantage with 105-qubit Zuchongzhi 3.0 Processor." *Physical Review Letters*, 134, 090601. https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.134.090601 2. Google Quantum AI. (2024). "Quantum error correction below the surface code threshold." *Nature*, 638, 920-926. https://www.nature.com/articles/s41586-024-08449-y 3. Pan, J.W., et al. (2021). "Quantum computational advantage using photonic Gaussian boson sampling." *Physical Review Letters*, 127, 180502. 4. University of Science and Technology of China (2025). Announcement of the Zuchongzhi 3.0 superconducting quantum computing prototype. https://news.ustc.edu.cn/info/1056/90742.htm 5. Chinese Academy of Sciences. "Top 10 Science and Technology News of China, 2025." http://www.cas.cn/cm/202601/t20260130_5099165.shtml
*This article is written in a Feynman-style popular science format, aiming to explain complex quantum computing principles in accessible language.*