China’s Jiuzhang 4.0: A Cubic-Level Quantum Computing Leap

Chinese researchers have unveiled “Jiuzhang 4.0” (Nine Chapters Four), a photonic quantum computing prototype that achieves cubic-level computational power by manipulating and detecting up to 3,050 photons — a more than tenfold increase over its predecessor. The breakthrough, published in the journal Nature on May 13, 2026, was led by Pan Jianwei and Lu Chaoyang of the University of Science and Technology of China (USTC), in collaboration with seven other institutions including Tsinghua University and the Shanghai Artificial Intelligence Laboratory.

The system performs a Gaussian boson sampling task in just 25 microseconds — a feat that would take the world’s fastest supercomputer, El Capitan, more than 10^42 years to complete, yielding a quantum advantage ratio of 10^54. As CGTN reported, this represents a new world record for optical quantum information technology.

Breaking the Photon Loss Bottleneck

The most significant challenge in photonic quantum computing has long been photon loss — as optical networks grow larger, photons are absorbed or scattered at every stage of propagation, coupling, and detection, causing effective multi-photon interference events to decline exponentially with system size.

Jiuzhang 4.0 overcomes this through two key innovations. First, the team developed a high-efficiency optical parametric oscillator (OPO) light source achieving approximately 92% single-source efficiency and 51% total system efficiency. Second, and more fundamentally, the researchers created a “spatiotemporally hybrid-coded interferometer” that enables cubic-level expansion of connectivity without proportionally increasing hardware components.

According to the USTC official announcement, the system integrates 1,024 high-efficiency squeezed-state optical fields into an 8,176-mode spatiotemporally hybrid-coded circuit, operating within a Hilbert space of 10^2461 dimensions.

Wang Kunkun, a professor at Anhui University, explained the innovation using an analogy: “Space is like roads, time is like schedules. The same set of hardware can be reused in different time windows, with fiber delay rings achieving ‘time caching,’ allowing photons to interfere in both spatial and temporal dimensions. Without significantly increasing devices, a higher-dimensional, larger-scale quantum network can be constructed.”

A Decade of Quantum Progress

The Jiuzhang series, named after the ancient Chinese mathematical classic The Nine Chapters on the Mathematical Art, has progressed rapidly since its inception:

• 2020: The original Jiuzhang achieved 76-photon quantum advantage (ratio: 10^5)
• 2021: Jiuzhang 2 reached 113 photons (ratio: 10^10)
• 2023: Jiuzhang 3 pushed to 255 photons (ratio: 10^16)
• 2026: Jiuzhang 4 reaches 3,050 photons (ratio: 10^54)

China is now the only country to have achieved “quantum computational advantage” on both the photonic route (Jiuzhang series) and the superconducting route (Zuchongzhi series), as noted by Xiao Lei, a professor at Southeast University, in comments to Xinhua News.

Architecture Innovation Over Brute Force

What distinguishes Jiuzhang 4.0 is not merely the increase in photon count but a fundamental rethinking of photonic quantum computing architecture. The spatiotemporal hybrid coding scheme allows computational power to grow cubically with hardware additions rather than linearly, circumventing the traditional trade-off where adding more optical components increases photon loss.

Xue Peng, a professor at Southeast University, told Xinhua that the “cubic-level expansion of connectivity” represents a “disruptive innovation” in optical quantum computing architecture. Rather than simply adding more hardware — which introduces more loss — the system reuses existing components across different time windows.

Implications and the Road Ahead

While Jiuzhang 4.0 remains a specialized quantum simulator designed for Gaussian boson sampling rather than a general-purpose programmable quantum computer, its achievements have significant practical implications. Gaussian boson sampling has applications in image recognition, graph theory computations, and generating bosonic error-correcting codes. The demonstration also provides a foundation for building large-scale quantum entangled cluster states and fault-tolerant optical quantum computing hardware.

Lu Chaoyang noted that the results offer new possibilities for constructing “trillion-qubit-mode three-dimensional cluster states” and future fault-tolerant optical quantum computing hardware, as reported by Guangming Daily.

Looking ahead, experts predict a heterogeneous quantum computing ecosystem. Xiao Lei of Southeast University described a future where “superconducting and ion trap routes will explore toward universal quantum computing, while photonic quantum computing will be irreplaceable in quantum internet, distributed computing, and specialized sampling tasks.”

The Nature paper acknowledges that universal quantum computing will require millions of qubits with error correction, but Jiuzhang 4.0’s demonstration of scalable photonic architecture with dramatically reduced photon loss provides a viable path forward. As China continues to invest heavily in quantum technology as a strategic national priority, the Jiuzhang series represents a steady march from laboratory demonstrations toward practical, real-world quantum systems.