# Is Silicozen Chip China's Best Bet on Photonic Quantum Computing?

Hefei-based Silicozen Chip Technology has closed a financing round of over 100 million yuan, co-led by Weihao Venture Capital and publicly listed Gomtec (002655), with follow-on participation from Huace Film & TV (300133), Suzhou Venture Capital, Suzhou Financial Holding, and Chuangling Capital. The capital will go toward two targets: R&D of 10,000-bit dedicated optical computing clusters, and development of general-purpose quantum computing chip sets. The company positions itself as China's only enterprise offering a full-stack, general-purpose photonic quantum computing system — and explicitly benchmarks itself against [PsiQuantum](https://quantumintel.tech/companies/psiquantum) and [Xanadu](https://quantumintel.tech/companies/xanadu) as its long-term peers.

The company's most concrete near-term credential is its QRNG-10 quantum random number generator chip — the first millimeter-scale QRNG chip in China, and the first to clear evaluation by the State Cryptography Administration's Commercial Cryptography Testing Center. Orders for approximately 200,000 units have already been placed, with the chip deployed in SIM cards, State Grid electric meter encryption modules, and home cameras. This revenue stream, modest by global venture standards, gives Silicozen a live commercial loop that most photonic quantum startups lack entirely.

---

## What Silicozen Actually Built — and What It Claims

Founded in Hefei in 2020, Silicozen emerged from the Key Laboratory of Quantum Information of the Chinese Academy of Sciences, the group led by Academician Guo Guangcan at the University of Science and Technology of China (USTC). Its three co-founders carry serious institutional weight: Chen Wei (Changjiang Scholar and Executive Director of USTC's Department of Optics and Optical Engineering), Ren Xifeng (National Outstanding Youth Fund recipient and Deputy Director of the Key Laboratory of Quantum Information), and Ding Yuyang (USTC PhD, quantum chip R&D focus). The team claims more than 20 years of collective experience in integrated optical quantum systems.

In November 2025, Silicozen publicly released a general-purpose programmable quantum computer built on silicon photonic integrated chips, accessible via cloud platform. The company states this is the only optical quantum computing system in China capable of running general quantum algorithms — a claim that, if it holds up to scrutiny, defines a meaningful category lead in the domestic market.

On the research side, Silicozen and USTC's Key Laboratory jointly demonstrated the stable generation of a 4-photon 16-qubit GHZ state and a single-photon 4-qubit cluster state on chip. According to the source, this represents the largest-scale [entanglement](https://quantumintel.tech/glossary/entanglement) demonstrated on optical quantum chips to date. The team also ran Grover's algorithm on the same platform, reporting an average recognition probability of 98.7%. These are specific, citable figures — but they have not yet been independently peer-reviewed based on the available source material, and readers should weight them accordingly.

In dedicated optical computing (a distinct product line from general-purpose quantum), the company reports completing internal verification of 5,000-bit-level equipment, with commercial launch targeted for the end of 2026.

---

## The Qubit-Fidelity Tension Silicozen Is Navigating

General Manager Ding Yuyang addressed the qubit count vs. [gate fidelity](https://quantumintel.tech/glossary/gate-fidelity) trade-off directly in the 36Kr interview. His framing is technically sound: at a single-qubit error rate of 0.1% (equivalent to 99.9% fidelity), combining 10 qubits yields a circuit-level correct rate of 0.999 to the power of 10 — approximately 99%. Scale to hundreds of physical qubits without improving fidelity, and the system becomes computationally useless before it becomes useful. This is precisely the dynamic that has made the [NISQ](https://quantumintel.tech/glossary/nisq) era so commercially frustrating across all hardware modalities.

Photonic systems have a structural advantage here that is worth stating plainly: photons don't require cryogenic cooling, operate at room temperature, and can be routed and integrated using semiconductor fab processes. The disadvantage is equally structural — photon-photon interactions are weak, making two-qubit gates hard to implement deterministically, and photon loss is a persistent source of error that doesn't have a clean analog in superconducting or trapped-ion systems. Silicozen's silicon photonic integration approach is one of several strategies the field is pursuing to address this, but the company has not disclosed its two-qubit gate fidelities or photon loss rates in the source material — gaps that enterprise buyers and technical due diligence teams will need to probe.

---

## Commercialization Runway and the Investor Logic

The financing structure reveals something about investor confidence — and its limits. Gomtec and Weihao leading the round, with film company Huace Film & TV and regional government-linked vehicles (Suzhou Venture Capital, Suzhou Financial Holding) following, suggests this is a round anchored by strategic and policy-aligned capital rather than deep-tech global funds. That's not inherently a weakness for a Chinese quantum hardware company operating in a domestic ecosystem shaped heavily by state procurement and national standards — but it does affect how Silicozen's international competitive positioning should be read.

The three-horizon commercialization strategy Ding Yuyang outlined is coherent: QRNG chips fund operations now; 10,000-bit dedicated optical computing clusters provide medium-term revenue; million-bit general-purpose optical quantum computers are the long-term target. The QRNG business already has verified repurchase cycles and the State Cryptography Administration's stamp of approval — an important unlock for government and critical infrastructure procurement in China. Ding predicts annual QRNG shipments could reach millions of units once national anti-quantum cryptography standards are formally introduced.

The 10,000-bit dedicated optical computing cluster — distinct from the general-purpose quantum computer — is better understood as a classical-adjacent photonic accelerator. This segment competes with companies like Lightelligence and LightMatter internationally, though the source does not make those comparisons. It may have a faster path to revenue than full fault-tolerant quantum computing, but conflating the two product lines, as press coverage often does, obscures the very different technical maturity levels involved.

---

## Where This Sits in the Global Photonic Race

Internationally, [PsiQuantum](https://quantumintel.tech/companies/psiquantum) and [Xanadu](https://quantumintel.tech/companies/xanadu) are the most cited photonic quantum computing companies building toward fault-tolerant systems. PsiQuantum's approach relies on photonic qubits with silicon photonics manufacturing at GlobalFoundries scale; Xanadu's Borealis system demonstrated photonic quantum advantage claims in 2022. Neither has disclosed a commercial general-purpose system running arbitrary user algorithms at scale as of this writing.

Silicozen's claim to be China's sole full-stack general-purpose photonic quantum computing enterprise — if accurate — makes it a credible domestic counterpart to those efforts, albeit at an earlier technology readiness level. The 4-photon, 16-qubit GHZ state result is meaningful as a laboratory benchmark, but the gap between a 16-physical-qubit entangled state and the millions of physical [photonic qubits](https://quantumintel.tech/glossary/photonic-qubit) required for [fault-tolerant quantum computing](https://quantumintel.tech/glossary/fault-tolerant-quantum-computing) is not incremental — it is architectural. The company's own roadmap acknowledges this with its "long-term" million-qubit framing.

For enterprise buyers evaluating quantum cloud access today, the cloud-accessible silicon photonic system released in November 2025 is the relevant near-term product. Its gate fidelities, circuit depths, and supported algorithm classes have not been disclosed in the source material available to us.

---

## Key Takeaways

- **Silicozen Chip closed a financing round of over 100 million yuan**, co-led by Weihao Venture Capital and Gomtec (002655), with multiple follow-on investors including Suzhou Venture Capital and Suzhou Financial Holding.
- **Funds are targeted** at 10,000-bit dedicated optical computing cluster R&D and general-purpose quantum computing chip sets.
- **The QRNG-10 chip** has received orders for approximately 200,000 units, passed State Cryptography Administration evaluation, and is shipping into SIM cards and critical infrastructure encryption applications.
- **The largest entanglement milestone disclosed**: a 4-photon 16-qubit GHZ state and single-photon 4-qubit cluster state on chip, with 98.7% average recognition probability on a Grover search demonstration.
- **The company positions itself** as China's only full-stack general-purpose photonic quantum computing enterprise, benchmarking against PsiQuantum and Xanadu for long-term trajectory.
- **Key technical gaps** not disclosed in available source material include two-qubit gate fidelities, photon loss rates, and coherence-equivalent metrics for the cloud-accessible system.
- **5,000-bit dedicated optical computing equipment** has completed internal verification; commercial launch is targeted for end of 2026.

---

## Frequently Asked Questions

**What is Silicozen Chip and who founded it?**
Silicozen Chip Technology (Hefei Silicozen Chip Technology Co., Ltd.) was founded in 2020 in Hefei, China. It was incubated by USTC's Key Laboratory of Quantum Information, led by Academician Guo Guangcan. The three co-founders are Chen Wei (Changjiang Scholar, USTC), Ren Xifeng (National Outstanding Youth Fund recipient), and Ding Yuyang (USTC PhD).

**How much did Silicozen Chip raise and from whom?**
The company raised over 100 million yuan in its latest round, co-led by Weihao Venture Capital and Gomtec (002655). Follow-on investors include Huace Film & TV (300133), Suzhou Venture Capital, Suzhou Financial Holding, and Chuangling Capital, per 36Kr's reporting.

**What makes Silicozen's photonic quantum computer significant in China?**
According to the company, its silicon photonic integrated chip-based quantum computer — released in November 2025 and accessible via cloud — is the only optical quantum computing system in China capable of running general quantum algorithms. This claim has not been independently verified by a third party in the available source material.

**What is the QRNG-10 chip and why does it matter commercially?**
The QRNG-10 is Silicozen's quantum random number generator chip — described as the first millimeter-scale QRNG chip in China and the first to pass State Cryptography Administration evaluation. It has approximately 200,000 units on order and is integrated into SIM cards, electric meter encryption chips, and home cameras. It represents the company's primary near-term revenue driver.

**How does Silicozen compare to PsiQuantum and Xanadu?**
Silicozen explicitly names PsiQuantum and Xanadu as its long-term global benchmarks. All three pursue photonic approaches to fault-tolerant quantum computing, but at significantly different scales and funding levels. PsiQuantum and Xanadu have raised substantially more capital internationally and have published more peer-reviewed technical results, though none has demonstrated a fault-tolerant logical qubit at scale as of mid-2026.

**What are the main technical risks in Silicozen's approach?**
Photonic quantum computing faces known structural challenges: weak photon-photon interactions make deterministic two-qubit gates difficult, and photon loss is a persistent error source. The source material does not disclose Silicozen's two-qubit gate fidelities or photon loss rates — metrics that enterprise buyers and technical investors will need before drawing comparisons to superconducting or trapped-ion competitors.