Tsinghua-incubated quantum AI startup Qingxingyigou raises nearly 100M yuan in two rounds within three months

Beijing — Qingxingyigou (Clear异构), a Tsinghua University-affiliated quantum artificial intelligence startup, has completed a Series A+ funding round, with investors including Jingkai Capital, Anhui High-Tech Investment, Daode Investment, Lishi Investment, Zhongzi Fund, Xuhui Capital, Senlan Group, and the founder of Jingdong Group.

Combined with its earlier Series A round, the company has raised nearly ¥100 million in total over the past three months.

From Heterogeneous Computing to Quantum-Inspired AI

Founded in 2021 by Yu Teng, a Tsinghua University computer science alumnus, Qingxingyigou initially focused on AI systems based on classical heterogeneous computing. The company has since pivoted toward the frontier of quantum-inspired next-generation AI models, with products now integrated into the AI and quantum computing industry chain.

Given that general-purpose quantum computers are still under development, the company chose not to wait for full hardware maturity. Instead, Qingxingyigou enables quantum-inspired high-dimensional intelligence to perform large-scale inference on existing GPU hardware. By restructuring models from parameters to architecture, the company addresses industry pain points and significantly optimizes token usage efficiency.

In the era of general artificial intelligence (AGI), where demand for computing power is soaring, this approach offers substantial room for growth. The market has long awaited models with higher knowledge density and energy efficiency for real-world AI deployment.

Early Focus on Heterogeneous Computing

Qingxingyigou was established in 2021. The team initially developed parallel computing system software to address AI system construction and performance issues in heterogeneous processor environments. It soon launched a prototype "program factory" product and collaborated with upstream and downstream enterprises. In 2022, the company completed three consecutive funding rounds within six months.

For quantum AI models to enter real computing environments, algorithms, chip adaptation, and system operation are all essential. This is why, when the team shifted its R&D focus to quantum AI in 2025, it chose Shanghai as its base—a city that brings together quantum computing hardware, algorithms, AI computing power, and application scenarios, offering faster access to equipment and market demand.

RiverONE: A Quantum-Inspired Vision-Language Model

This year, Qingxingyigou officially launched RiverONE, a quantum-inspired vision-language model. According to the company, the model uses simulated quantum computing to generate parameters during the construction phase. Once trained, the model runs inference on classical GPUs without requiring real-time quantum computer involvement. This allows the team to develop, deploy, and iterate products using existing hardware.

According to third-party test reports, in the specific task of quantum calibration chart understanding, RiverONE achieved at least 95% of the performance of Nvidia's Ising Calibration 1 model, using only 1.9 billion parameters—less than one-tenth of the comparison model. This provides a lighter-weight solution for specialized scenarios.

Industry Integration and Partnerships

Since 2023, Qingxingyigou has been an official AI system supplier to global computing giant AMD. In August of this year, Liberation Daily reported that the company had made phased progress in its cooperation with upstream quantum computing and AI computing infrastructure providers.

• Biren Technology disclosed that the two parties conducted adaptation tests on its GPU products for mainstream quantum computing frameworks including Cirq, PennyLane, and Qiskit. Within the test scope, 48-hour stability tests showed no abnormal interruptions affecting conclusions.
• Muxi Technology's Xiyun C-series GPUs, in collaboration with Qingxingyigou, completed adaptation of the Ising quantum calibration multimodal model on the vLLM inference framework and passed the international calibration benchmark QCalEval.
• Taichu Electronics, in partnership with Loongson's pure domestic computing base, achieved Day-0 adaptation of Qingxingyigou's full suite of quantum-inspired models and operators.
• Qianhe Yibang, a chip company incubated by NetEase for gaming scenarios, signed a cooperation agreement with Qingxingyigou to adopt its game acceleration and adaptation support services.

Two Funding Rounds in Three Months: Expanding the Quantum AI Horizon

Can quantum-inspired AI grow into a sizable company?

The answer is becoming clearer. As AI technology becomes more widespread and deployed globally, computing power remains in short supply. Models with higher knowledge density and energy efficiency will have greater market opportunities—and this is precisely where quantum computing can contribute.

Multiverse Computing in Spain offers a relevant case. Starting from quantum software, the company uses tensor network methods for AI model compression, launching CompactifAI to reduce storage and computing resource requirements for large models. Its technical papers document model compression experiments and performance trade-offs across different tasks.

Multiverse has since expanded its commercial path: offering compressed models and APIs, exploring offline operation on devices like smartphones, and driving adaptation with hardware vendors and enterprise clients. In 2025, the company announced a €189 million Series B round; in 2026, it announced plans to raise up to €500 million in a Series C round. Japan's Marubeni Corporation has signed a memorandum with Multiverse to study the promotion of its compression technology in the Japanese market.

Computational methods derived from quantum science can indeed find entry points in the existing AI market. Customers care about whether models can run on their own devices, reduce costs, and maintain required performance.

While Multiverse primarily compresses existing models for broad enterprise deployment, Qingxingyigou focuses on using simulated quantum computing and quantum-inspired algorithms to comprehensively reconstruct models—from parameters to architecture. The company has advanced this future-oriented technology exploration into stages including model release, private deployment delivery, and cloud service internal testing.

Looking Ahead

Ultimately, the market will focus on product delivery: Can quantum-inspired large models like RiverONE maintain performance on more real-world data? Can adaptation experience be distilled into replicable software capabilities? Can cloud-edge-end representative cases generate sustained orders?

Looking further ahead, as the quantum computing industry chain gradually takes shape, Qingxingyigou aims to open up downstream applications for China's quantum computing industry, carving out new paths to serve the explosive growth market of general AI.

Xuhui Capital commented: "The quantum AI赛道 has enormous industrial potential. Dr. Yu Teng's team at Qingxingyigou has chosen to enable quantum-inspired high-dimensional intelligence for large-scale inference on existing GPU computing power. By restructuring models from parameters to architecture, they have achieved breakthroughs and industrial implementation, significantly optimizing token usage efficiency and solving industry pain points."

"Xuhui Capital continues to invest in the frontier of quantum-AI交叉. We will leverage our state-owned innovation platform advantages to connect local computing power, scenarios, and industry partners in Xuhui, providing ongoing support for portfolio companies. We look forward to Qingxingyigou seizing opportunities to open downstream application tracks for China's quantum computing industry and build a benchmark for Chinese quantum-inspired AI in the general AI incremental market."

Anhui High-Tech Investment added: "Qingxingyigou is deeply engaged in the frontier of quantum-AI融合 innovation, focusing on core quantum AI technology self-development and engineering implementation. It possesses rare original technical barriers and a differentiated development path. Quantum AI is a core future赛道 for breaking through traditional intelligent computing bottlenecks and cultivating new productive forces. Anhui High-Tech Investment adheres to the philosophy of investing early and in hard tech. We are optimistic about Qingxingyigou's technical accumulation and growth potential, and will leverage our state-owned platform advantages to continuously empower the company's technology iteration and scenario implementation, helping Anhui seize the high ground of quantum intelligent industry development."

The window for industrialization has opened. The imagination space for China's AI + quantum computing is expanding.