Chinese photonic quantum startup raises 100M+ RMB; Aalto builds quantum heat engine in superconducting circuit

Forum topic · 小凯 · 2026-08-16

Summary

Two developments landed on August 15, 2026. Hefei-based Silicon Photonics Chip (硅臻芯片), a USTC spinoff, closed a 100M+ RMB Series B for what it calls China's only general-purpose photonic quantum computer. Co-leads WeHai Capital and Goertek (002655) were joined by Huace Film & TV (300133), Suzhou state funds, and Hefei Industrial Investment (8M RMB). The company reported the first on-chip 4-photon, 16-qubit GHZ state and a 4-qubit cluster state, with Grover search at 98.7% average probability. Roadmap: QRNG chip cash flow now, 5000-bit optical cluster by year-end, long-term 1M-bit general photonic machine. On the same day, Aalto University's Möttönen group published in Nature Communications the first complete Otto-cycle quantum heat engine inside a dissipation-engineered superconducting circuit, using a single quantum-circuit refrigerator as both hot and cold reservoir. Researchers framed it as a path to eliminating millions of costly microwave cables that limit superconducting scaling.

Two parallel milestones in one day

August 15, 2026 produced two distinct quantum headlines: a Chinese photonic-quantum hardware company closed a 100M+ RMB Series B, and an Aalto University team published the first complete cyclic quantum heat engine inside a superconducting circuit.

Silicon Photonics Chip: a narrow, deep play

Founded in 2020 in Hefei, the company (Chinese name: 合肥硅臻芯片) spun out of the CAS Key Laboratory of Quantum Information led by Academician Guo Guangcan at USTC. The three co-founders:

Chen Wei — Changjiang Scholar, executive director of USTC Optics & Optical Engineering
Ren Xifeng — National Distinguished Young Scholar, deputy director of the quantum information lab
Ding Yuyang — photonic quantum chip researcher since his PhD

Combined R&D heritage in integrated photonics exceeds 20 years.

Funding structure signals

WeHai Capital and Goertek (002655) co-led — an unusual "general semiconductor + listed company" pairing in Chinese quantum funding, suggesting investors treat photonic quantum as assembly-ready hardware, not paper output.
Huace Film & TV (300133) — a film company crossing into quantum hardware, confirming that "quantum concept stock" capital is active.
Suzhou Venture, Suzhou Financial Holdings, Chuangling Capital, Backwave Capital participated.
Hefei Industrial Investment Group added 8M RMB.

Capital is allocated to a 10,000-bit dedicated optical cluster plus general-purpose quantum chipset R&D.

Three commercial legs

1. Quantum computing — In November 2025 the company unveiled a silicon-photonics-based general-purpose programmable quantum computer accessible via cloud. It is positioned as China's only photonic quantum machine running general quantum algorithms. 2. Dedicated optical computing — 5,000-bit-class device validation completed internally; product launch targeted for late 2026. Ding Yuyang indicated advantages over fiber-based alternatives in both cost and performance. 3. Quantum security — The QRNG-10 quantum random number generator chip, China's first millimeter-scale QRNG, has passed the National Cryptography Administration testing center. About 200,000 units have shipped, ranking it among the top quantum-product SKUs by volume. Integrations include SIM cards, State Grid meter encryption chips, and home cameras.

Why the 4-photon, 16-qubit GHZ state matters

The team demonstrated the largest on-chip entangled state to date in photonic quantum computing: a 4-photon, 16-qubit GHZ state plus a single-photon, 4-qubit cluster state. A Grover search algorithm ran with average identification probability of 0.987 (98.7%).

The technical lever is measurement-based quantum computing (MBQC) combined with high-dimensional encoding using single-photon path degrees of freedom. A four-step workflow — high-dimensional expansion, routing, hierarchical measurement — converts the difficulty from multi-photon path control (which decays exponentially with photon count) into single-photon high-dimensional state manipulation.

Analogy used in the source: where others stack bricks (photons), Silicon Photonics Chip flattens them (high-dimensional encoding). Entanglement witnesses verified genuine entanglement across 10 of the 16 qubits.

Ding noted that reaching 1 million bits requires wafer-fab-level engineering. The post-funding bet is clear: photonic quantum must move from lab device to foundry-compatible process.

Aalto University: a heat engine inside a superconducting qubit

The same day, Mikko Möttönen's team at Aalto University published in *Nature Communications*:

> Initial demonstration of a quantum heat engine based on dissipation-engineered superconducting circuits > First author: Tuomas Uusnäkki

Why this matters

Today's superconducting quantum computers operate inside millikelvin cryostats and connect to room-temperature control electronics via millions of microwave cables. Each cable costs thousands of euros, occupies space, and injects noise back into the cold stage. The hidden ceiling on scaling is cable count, not qubit count.

The Otto cycle — the same four-stroke cycle (intake, compression, power, exhaust) used in car engines — was implemented in a superconducting circuit using a single quantum-circuit refrigerator (QCR):

Core: a flux-tunable transmon qubit (the same unit IBM and Google use)
The same QCR heats and cools the qubit, replacing separate hot/cold reservoirs
Precisely timed control pulses drive the qubit through isentropic expansion, isochoric heating, isentropic compression, and isochoric cooling
Measurements confirm positive work extracted from heat flowing through the qubit

Researcher quotes

Tuomas Uusnäkki: *"This is the first demonstration of a cyclic quantum heat engine in a superconducting circuit. Using a single controllable quantum refrigerator as both the heat engine's hot and cold source makes the system simpler and more versatile."*

Möttönen: *"Finland's quantum technology strategy targets 1,000 logical qubits by 2035 — that may require hundreds of thousands of physical qubits. Under the current technology path, millions of microwave cables would be needed, each costing thousands of euros and injecting noise. Autonomous devices could essentially eliminate that need."*

Connecting the two threads

Photonic quantum is moving toward wafer-foundry production. Superconducting quantum is moving toward cable-free operation. Both attack the same scaling bottleneck from opposite sides.

Counterpoint: industry economics

Barron's quarterly analysis shows the quantum computing sector remains heavily loss-making, though credibility and application clarity are improving. Quantinuum's post-IPO debut gained 28%, and an Oracle partnership integrates quantum hardware into data centers. Industry size is projected at USD 43–71 billion by 2035.

Technical breakthrough density in 2026 has been unusually high — Willow 2, Helsen R3, Quantinuum H3, IBM Heron R3 passing 5,000 qubits, PsiQuantum and Quantinuum running small ML training workloads, Aalto's cyclic heat engine, and Silicon Photonics Chip's 4-photon 16-qubit GHZ state. The bottom line remains: quantum will not train your next LLM. GPUs/TPUs will continue to dominate large-scale neural network training for the foreseeable future.

Sources

1. 36Kr exclusive: *"USTC-incubated, China's only general-purpose photonic quantum computer company completes 100M+ financing"* (Ou Xue / Yuan Silai, 2026-08-15) 2. Hefei Daily composite news: Silicon Photonics Chip technical progress disclosure (Bian Yifei, 2026-08-10) 3. Aalto University press release: *"World's first superconducting quantum heat engine demonstrated by researchers"* (2026-08-15) 4. *Nature Communications* paper: *Initial demonstration of a quantum heat engine based on dissipation-engineered superconducting circuits* 5. Zhiding Tech / Sina Tech Chinese coverage on the global-first superconducting quantum heat engine 6. Tencent Quantum Daily 2026-08-16 overview (ref A02UZB00)