BlueRun Ventures Angel Portfolio Company Kaimat Completes Hundreds of Millions of RMB in Angel+ Round, Building AI4S-Driven Materials Intelligence Infrastructure | BlueRun Family

On March 24, Kaiwuji, an innovator in the AI for Science (AI4S) space, announced the completion of a Series Angel+ funding round worth hundreds of millions of RMB. The round was led by Monolith, with Luminous Ventures and JAFCO Asia participating, and BlueRun Ventures and other existing shareholders significantly increasing their stakes. The proceeds will be used primarily to deepen capabilities in materials foundation models, advance multiple proprietary materials pipelines toward commercialization, and continue building out a world-class team of interdisciplinary talent.

BlueRun Ventures led Kaiwuji's angel round. From BlueRun's perspective, AI is becoming the core force driving a paradigm shift in scientific research. The firm has long tracked and invested in AI-driven scientific discovery (AI4S), from biopharmaceuticals to new materials, consistently backing top scientific teams in deeply integrating cutting-edge algorithms with experimentation and engineering to accelerate the path from fundamental innovation to industrialization. Kaiwuji represents a critical piece of this portfolio, and BlueRun expects the company to build reusable materials intelligence infrastructure for the industry.

As a full-stack materials innovation platform building a closed-loop "model — experiment — mass production" system, Kaiwuji focuses on the complete IP lifecycle for materials, from large-scale model training and inference to experimental validation and ton-scale manufacturing, aiming to enable efficient conversion of materials from discovery to production.

Kaiwuji was founded by Dr. Ziheng Lu, an internationally leading scientist in materials science driven by AI4S. Dr. Lu previously served as Principal Researcher at Microsoft Research and head of the materials team at its AI for Science Center, with extensive experience in large-scale deep learning and its applications in materials design, plus over a decade of hands-on work in energy materials, laboratory research, and industrial deployment. Co-founder Dr. Mengyang Yang was previously Senior Research Manager at Microsoft, with a cross-disciplinary background in optics, electronics, and materials. He led teams contributing to multiple frontier Microsoft storage and materials R&D projects including HSD, Genesis, and Silica, driving efforts from proof-of-concept and prototype construction through to system-level deployment. Industrial CTO Dr. Yu Ren brings more than 20 years of industrialization experience, having spent years at materials leaders such as BASF responsible for strategic product selection, engineering of complex technical systems, and mass production scale-up.

Additionally, the company has assembled a top-tier interdisciplinary team spanning model pre-training, Agentic AI, laboratory R&D, and industrial translation, with core members drawn from leading global institutions including Microsoft Research, Google DeepMind, BASF, University of Cambridge, Imperial College London, Columbia University, Tsinghua University, and Beijing Zhongguancun Academy — forming end-to-end capabilities from foundational research and model development to engineering scale-up, and giving Kaiwuji an "end-to-beginning" industrial perspective.

Kaiwuji adheres to a weak-prior-constrained scaling approach, having built synthetic data and high-throughput physical experimentation environments with underlying high-concurrency mechanisms. Through large-scale pre-training, its models learn more generalizable representations of materials and chemical space, enhancing their ability to search, generate, and screen unknown materials systems — thereby breaking through the limitations of human chemical priors on the innovation space and more efficiently identifying high-value, breakthrough source materials IP. Based on this capability, the company has constructed a dual-engine architecture of "Prophet," a predictive engine for high-precision, broad-spectrum property prediction and screening, and "Creator," a generative engine for cross-element materials generation and inverse design. At the same time, the company has established million-scale high-concurrency data infrastructure, and has begun building out laboratories, automated high-throughput platforms, and kilogram-scale validation platforms, accelerating the formation of a foundational capability system where models, experiments, and engineering advance in concert.

The founding team has accumulated deep experience in materials foundation models and industrial translation. Related research has validated the effectiveness of Scaling Law in materials foundation models, published in Nature in 2025. In prior research and industrial practice, the team has also advanced multiple classes of high-value materials from model design through experimental validation to device-level deployment: significantly compressing R&D cycles in solid-state electrolytes; conducting the first systematic exploration of thermal conductivity distributions across over 640,000 inorganic crystal structures in thermal management materials, with some key thermal conductivity materials validated by third parties; and in recyclable PCB substrate materials, achieving material synthesis and processing into device-level products meeting real operating conditions. These accomplishments form an important foundation for Kaiwuji's current technical approach and capability system.

Kaiwuji insists on an **"end-to-beginning"**推进方式，from project initiation simultaneously considering manufacturability, stability, and scale-up constraints, pushing materials from laboratory iteration through kilogram-scale validation, pilot scale-up, and real operating condition application. The company both advances proprietary materials pipelines and engages in collaborative R&D with industrial partners, exploring new business models in AI-enabled materials. Currently, it has focused exploration on high-value directions including new energy batteries, cold storage, embodied intelligence thermal management materials, and superconducting materials, while continuously expanding into additional materials scenarios with industrial potential. Going forward, Kaiwuji will further deepen the model-experiment closed loop, driving incubation and industrial delivery of multiple materials pipelines.

Dr. Ziheng Lu, founder of Kaiwuji, stated: "AI is reshaping modern society at a staggering pace. What Kaiwuji hopes to build is not merely faster R&D tools, but a reusable, scalable materials intelligence infrastructure — moving materials IP generation away from dependence on individual experience and low-frequency serendipity, toward predictability and scalability. We look forward to advancing along this path, from models to mass production, to actually make materials that deliver real value."

BlueRun angel portfolio company Puyu Technology completes Series funding worth hundreds of millions of RMB, accelerating industrialization of nanoimprint lithography equipment | BlueRun Family

BlueRun seed portfolio company Lingchu Intelligence completes RMB 2 billion funding, tackling "dexterous manipulation" challenges | BlueRun Family

BlueRun Ventures angel portfolio company Zhijian Power completes RMB 2 billion funding, accelerating embodied intelligence toward reality | BlueRun Family