Four minutes of machine time versus 2.6 billion years on a conventional supercomputer makes the old computing hierarchy look fragile. That is the point of Jiuzhang, a photonic quantum computer built by Chinese scientists at the University of Science and Technology of China, under Pan Jianwei and Lu Chaoyang. It did not run a general-purpose app. It attacked a narrow problem that classical machines hate, and it did so with a speed gap that is hard to wave away.
The experiment was published in Science and set a world record by detecting 76 photons at once. The hardware is built around light, not silicon logic gates, using lasers, mirrors and photon detectors to steer particles through an optical circuit. This choice shows where the field is heading: toward machines that do not try to imitate ordinary computers, but use different physics to work through certain problems in a radically different way.
What changed
Jiuzhang was used for Gaussian boson sampling, a task that asks the machine to sample from an extremely tangled probability distribution produced by many photons moving through a random optical network. A normal computer can only check possibilities in sequence. Jiuzhang leans on superposition, the quantum property that lets a qubit occupy multiple states at once, so the system can explore a much wider space of outcomes in parallel.
The machine is not doing every kind of computing faster. One class of calculation, the kind that explodes in complexity as the number of particles grows, becomes a terrible job for classical hardware and a plausible one for a quantum system. The four-minute result is therefore less about raw speed and more about a split in method. One side counts through options; the other is built to live inside them.
Jiuzhang is also photonic, which sets it apart from the better-known superconducting and trapped-ion approaches. It uses photons rather than electrical currents or suspended ions, and in the reported setup it operated at room temperature. This is not a solved engineering problem, but it hints at a route that may be less awkward to scale in some use cases than cryogenic systems that need extreme cooling.
Why it matters
The obvious headline is the speed gap. The less obvious point is that the gap was achieved on a task almost perfectly chosen to expose classical weakness. Gaussian boson sampling is not a toy benchmark. It is a stress test for probabilistic computation, and it shows how quickly a problem becomes unmanageable once the number of interacting possibilities gets large.
For South African businesses, the immediate lesson is not that laptops are about to disappear. The next wave of computing advantage will be uneven, specialised, and commercially strategic. In finance, this could reshape risk modelling, portfolio optimisation, and fraud detection. In logistics, it could improve route planning, fleet scheduling, and supply chain decisions across ports, mining, and freight. In security, the stakes are sharper. Current public key systems, including RSA and ECC, sit in the path of a future quantum threat, while post-quantum cryptography becomes a procurement issue rather than an academic debate.
Medicine and climate modelling are the other obvious fronts. Drug discovery depends on simulating molecular behaviour, and weather or climate work depends on processing vast, messy systems where more computing power only gets you so far. Quantum hardware will not instantly solve those fields, but it offers a different ceiling.
What happens next
The mistake is to read Jiuzhang as a replacement for standard computing. It is not a consumer device, and it is not heading into offices or homes any time soon. Quantum machines remain difficult to stabilise, difficult to scale, and prone to error. The promise is real, but so are the engineering limits.
The result proves the race is no longer theoretical. Chinese researchers have shown that a carefully built photonic system can produce a result classical machines cannot match in anything like the same timeframe. That will push more money, more talent, and more industrial strategy into quantum research, especially where governments and firms care about cryptography, advanced manufacturing, and high-value simulation.
The commercial consequence is simple. Organizations that wait for quantum computing to feel mature will arrive too late. The useful work starts earlier, with security planning, algorithm research, and domain-specific pilots in sectors where computation is already a competitive weapon.
