QCNEWS
Hangzhou's Heguang Quantum raised a seed round to pursue deterministic GKP error correction, a way of producing error corrected quantum states from light on demand instead of by chance.
Hangzhou-based Heguang Quantum has closed a seed round to pursue a specific hard problem in photonic quantum hardware: producing error-corrected quantum states on demand, every time, rather than when chance allows. Size, lead investors, and post-money valuation are not stated in the available material.
The wager is on a code called Gottesman-Kitaev-Preskill, or GKP, a way of encoding quantum information in continuous properties of light waves (amplitude and phase) rather than in individual particles. Photons are fragile; they get lost, they pick up noise. GKP turns that sensitivity into a controlled error-correction target by spreading each unit of quantum data across a wider wave pattern, so small disturbances become correctable shifts rather than silent failures. Most published photonic GKP work generates those wave patterns probabilistically: a usable state shows up only some fraction of attempts, and the rest are discarded. HQ's reported advance is a proprietary pluggable nonlinear module that the company says produces GKP states deterministically, the way a printer produces pages. Every run yields a usable output.
That distinction, on-demand versus probabilistic, is what the rest of the field is also working to close. A 2025 Nature paper demonstrated an integrated photonic source of GKP qubits on a chip-scale platform, giving the field a working anchor for what a deterministic photonic GKP source has to look like in practice. A separate arXiv preprint on extensible universal photonic quantum computing with nonlinearity sits on the same axis, arguing that engineered nonlinearity is what unlocks useful photonic computation. HQ is placing its bet on that axis, and using a modular hardware piece to do it.
The company's architecture emphasis is modular: separate photonic processing nodes linked by optical fiber, with reports of early-stage kilometer-scale distributed task tests. The distributed framing matters because it lets a startup scale by adding boxes rather than shrinking one chip past the point where fabrication breaks, and it gives the company a story for selling into classical compute clusters. The HQ10 quantum acceleration processor is planned as a module that plugs into GPU, CPU, and TPU systems, with sales targeted for 2027. That is a company roadmap, not an order book.
The founder, Dr. Shang Yu, trained under Academician Guo Guangcan at the University of Science and Technology of China and previously held a Marie Curie Fellowship in Europe. His prior work sits on programmable temporally encoded photonic processors and quantum simulation, both of which feed directly into the bet HQ is making. His Zhejiang Lab profile and Imperial College London research page document that background.
China quantum-photonics startup funding is moving fast enough that weekly trackers can no longer keep up by hand; The Crane China's recent startup raise roundup is one snapshot of the cohort HQ is raising inside.
The verification gap is what will move the story next. HQ's deterministic GKP generation is company-reported, mediated through an industry aggregator. Independent fidelity figures, success rates, and scalability numbers at useful levels are not yet public. Until they surface, the seed round buys a roadmap bet, not a benchmark.