In August 2026, a team at the University of Science and Technology of China (USTC) achieved below-threshold quantum error correction on the superconducting quantum processor "Zuchongzhi 3.2," using a distance-7 surface code — logical error rates dropped significantly as code distance increased. To understand why this matters, one has to start with the counterintuitive phenomenon of "the more you correct, the worse it gets."
1. What "error correction makes things worse" means
Quantum computing is like an extremely powerful but extremely fragile race car: huge potential, but the slightest temperature fluctuation or electromagnetic interference can knock qubits off course. Quantum error correction (QEC) installs an automatic steering system — but that system itself has weight and vibration. If the foundation isn't stable enough, correcting more aggressively makes things worse.
QEC has a breakeven line: correction only truly works when the corrected logical qubit is more reliable and longer-lived than the raw physical qubits. Once past this line, larger code distance and more resources mean lower logical error rates — "the more you correct, the better" — the precondition for quantum computing to move from theoretical feasibility to scalable engineering systems.
Few teams worldwide have reached "below threshold." Google's 2024 "Willow" processor provided the first hardware-level proof on distance-3, 5, and 7 surface-code logical qubits, with error rates falling exponentially with distance. What Zuchongzhi 3.2 achieved is independently crossing the same threshold on a domestically developed processor, via an all-microwave route with greater scaling potential.
2. The two key tricks of Zuchongzhi 3.2
To turn "correction makes things worse" into "correction makes things better," the core is making correction outrun noise. The researchers used two techniques:
Trick one: leakage recovery. Ordinary qubits live only on the "0" and "1" floors, but real devices are multi-story buildings — a qubit can accidentally jump to floors outside the computational subspace. Once that happens, standard correction can neither see it clearly nor clean it up, and the leaked qubit can drag neighbors down in later operations. Leakage recovery uses specific microwave drives to guide escaped qubits back into the computational space, carrying away excess energy and preventing leakage from accumulating across correction rounds.
Trick two: fast reset. The ancilla qubits responsible for "reporting errors" undergo a fast, unconditional reset at the end of every round, ensuring they start each round clean and don't propagate residual errors.
Together with the independently developed "all-microwave quantum state leakage suppression architecture," these form the foundation of stable operation. Its elegance: the suppression process is seamlessly embedded in the normal QEC workflow, with no additional hardware or time overhead.
As Professor Xue Peng (Beijing Computational Science Research Center) put it, QEC used to be like "repairing while water keeps leaking"; this work plugged the stubbornest leaks with recovery plus reset, producing the effect of "larger scale, higher reliability."
3. What an independent route means
With Zuchongzhi 3.0 (March 2025 cover paper in *Physical Review Letters*; 105-qubit random circuit sampling 15 orders of magnitude faster than classical supercomputers), China already held a leading position in superconducting quantum computational advantage. Zuchongzhi 3.2 pushes one step further: competing at the frontier in error correction.
Notably, the route is independent:
- Google's Willow uses traditional "frequency tuning + multiple control lines" coupling
- China's approach is "all-microwave leakage suppression" — all control done with microwave pulses
- Cooling: every additional 100 qubits roughly doubles cooling power demand
- Control cabling: microwave crosstalk causes spurious excitations; calibration, once done manually every 2-3 days, is now compressed to 4 hours with AI agents
- Packaging yield: currently ~80%
- Algorithm-hardware co-design: the "valuable problem set" for the early fault-tolerant era remains undefined
- Superconducting: error correction threshold crossed; independent all-microwave route established
- Photonic: on-chip MBQC engineering (4-photon 16-bit GHZ state + Grover 0.987) + boson sampling scale breakthrough (Jiuzhang 4.0: 3,050 photons + 8,176 optical modes)
- Quantum networks: 420-km quantum memory entanglement + Xinghan-2 14.5-km quantum repeater
- Trapped ions: Origin Wukong cloud platform — 52 million visits from 192 countries, 1 million tasks
- Quantum computing says goodbye to "the more you correct, the worse it gets" - Toutiao
- What's remarkable about the "Zuchongzhi 3.2" superconducting processor - Toutiao
- "Zuchongzhi 3.0" selected among China's top ten science news of 2025
- Major breakthrough in quantum error correction
- When computers outpace time itself
- From "quantum" to "mass production"
Both routes have room to scale. Microwave control requires fewer cables, is easier to array, and needs no complex DC bias circuitry — a key bottleneck for engineering million-qubit superconducting processors.
4. Several key judgments
First: general-purpose quantum computers within 5-10 years. CAS academician Guo Guangcan's judgment: "within 5 to 10 years, general-purpose quantum computers will emerge, with device sizes far smaller than current supercomputers," and "there is reason to believe this could come sooner."
Second: physical qubit count ≠ usable compute. Physical qubits are unpolished bricks; logical qubits are the standard building material. Surface codes may need ~100,000 physical qubits per logical qubit at equivalent correction distance. The community (including Caltech's Preskill group's recent "mitten codes") is working to compress this ratio, but all remain at the theoretical validation stage.
Third: fault-tolerant quantum computing requires running four tracks in parallel — compute, cooling, calibration, and ecosystem. Even with error correction solved, engineering challenges remain:
All of this says: now is not the time to pick a single horse. Superconducting, photonic, trapped-ion, and neutral-atom platforms each have strengths — the real race is just beginning.
5. Coordinates for China's quantum industry
Placing Zuchongzhi 3.2 alongside Origami (Benyuan) PSE-CZ, Silicon-Zhen MBQC, 420-km quantum memory entanglement, and Jiuzhang 4.0's manipulation of 3,050 photonic states sketches China's quantum industry coordinates as of August 2026:
"Full-stack domestic + multiple parallel routes + engineering deployment" is the true posture of China's quantum technology today.
---
Sources