Insider Brief
- China’s Jiuzhang 4.0 photonic quantum computer was among 20 projects honored with the 2026 World Internet Conference Leading Science and Technology Awards.
- In past experiments, the USTC prototype detected up to 3,050 photons in a single run, compared with 255 for Jiuzhang 3.0, while performing Gaussian boson sampling.
- Researchers claim a substantial speed advantage over classical computers, but boson sampling benchmarks have faced challenges and the task has no known commercial use.
- Image: A view of the programmable photonic quantum computing prototype “Jiuzhang 4.0.” /CMG
China’s Jiuzhang 4.0 photonic quantum computer has won one of the World Internet Conference’s top science prizes.
The prototype, built by researchers at the University of Science and Technology of China , or USTC, was among 20 projects honored with the 2026 WIC Leading Science and Technology Awards at a ceremony Monday in Wuzhen, in eastern China’s Zhejiang province, according to state broadcaster CGTN .
Organizers received more than 430 submissions from 43 countries and regions, CGTN reports. The winners fell into three categories: two in basic research, 10 in key technologies and eight in engineering and development. Other winners included optical transmission technology for space-based quantum information and an AI agent platform for intelligent vehicles.
A Record for Light-Based Computing
Jiuzhang uses particles of light, called photons, and computes by steering them through a network of optical parts and counting where they arrive. Other types of quantum computers hold information in superconducting circuits or trapped atoms.
The team described the machine in the journal Nature in May. It fed 1,024 beams of squeezed light — or, light prepared so that its natural flicker is reduced in one measurable property — into a circuit with 8,176 channels, called modes. Detectors recorded up to 3,050 photons in a single run. Jiuzhang 3.0 handled 255.
The machine performs a task called Gaussian boson sampling. In that task, light is sent through the network and records the pattern of photons that comes out. As the number of photons grows, predicting those patterns becomes far harder for ordinary computers.
USTC professor Lu Chaoyang said the device produces its most complex sample in about 25 microseconds, “shorter than the blink of an eye,” according to the official Xinhua News Agency . He said the world’s most powerful supercomputer would need more than 10 to the 42nd power years to calculate the same result. The team’s earlier preprint names that machine as El Capitan, at Lawrence Livermore National Laboratory in California. The team puts the machine’s speed at more than 10 to the 54th power times that of the supercomputer.
A Contested Benchmark
The Jiuzhang series began in 2020 with a machine that detected 76 photons. Jiuzhang 2.0 reached 113, The Quantum Insider reported in 2023 . The program comes from physicist Pan Jianwei’s group at USTC.
Speed claims for boson sampling have drawn challenges. A 2024 study in Nature Physics by researchers at the University of Chicago and Argonne National Laboratory used photon loss, a common flaw in optical experiments, to simulate earlier boson sampling results on conventional computers. The researchers said their classical method matched the ideal results better than the experiments did.
The Jiuzhang 4.0 team aimed its new work at that method. The team estimates that the method would take 1 followed by 42 zeros years to reproduce the result on the El Capitan, a world-leading classical supercomputer. The study, published in the preprint arXiv , also reports that the machine’s outputs could be told apart from several classical imitations.
Boson sampling has no known commercial use, but, rather serves as a test of whether a quantum device can outpace classical machines.
Lu said the team’s light sources and circuit design open a path to large three-dimensional “cluster states,” webs of linked light that researchers see as the raw material for optical quantum computers that can correct their own errors.