Solving the pain points of experimental tools with Physical AI, Shuye Technology has secured its Pre-A round of financing.
Physical AI for Science tool revolution explorer Shuye Technology has recently completed a Pre-A round of financing of tens of millions of yuan. This round is led by Topo Life Sciences Investment, followed by Origin Capital. The financing will be used for core technology R&D, marketing expansion, team building and replenishment of working capital.
Physical AI Must "Grow" in the Laboratory
What is Physical AI? At the beginning of 2026, Jensen Huang, founder of NVIDIA, declared at CES: "The ChatGPT moment for Physical AI has arrived." In his view, AI models capable of understanding and interacting with the real world are enabling autonomous machines to perceive, understand and perform complex operations in the real physical world. Gartner has listed Physical AI as one of the top 10 strategic technology trends in 2026. According to market research institutions, the global Physical AI market is expected to grow from 1.5 billion US dollars in 2026 to 15.24 billion US dollars in 2032.
Different from AI Agents that remain in the digital world, Physical AI emphasizes a fundamental logic: intelligence must grow through interaction with the real world. The capabilities of autonomous driving are developed on roads, and the capabilities of robots are developed through actions — while the capabilities of Physical AI for life sciences must be developed in real experimental operations.
Most traditional life science analysis tools are designed "people-centered": interaction relies on closed GUIs and physical buttons, hardware structures are adapted to human operation habits, each device stores results independently with inconsistent formats, and the devices themselves lack edge-side intelligence for perception, reasoning, decision-making and self-correction. Wet and dry experimental data are naturally separated, making it difficult to form a complete closed loop of "design-execution-detection-learning".
The core judgment of Shuye Technology is that the scientific research hardware base left over from the 20th century cannot support the full potential of AI autonomous experiments. Tools in the AI era must be upgraded from "execution tools" to "embodied intelligence carriers". This is exactly what Shuye Technology is doing — redefining and redesigning from the hardware mechanical layer, model and multi-agent layer, control logic layer, software and interface layer to the business and industrial value layer, making each device a carrier of embodied intelligence, and turning each experimental action into data that can be understood and optimized by AI.
VLA Operon: Upgrading Devices from "Execution Terminals" to "Intelligent Agents"
Founded in 2018 and headquartered in Hangzhou, Shuye Technology has a product studio in Weimar, Germany. Its founder Li Su holds a doctorate in engineering implementation and engineering robotics from the Technical University of Munich, Germany, and used to be a full-time lecturer in human-computer interaction at the Bauhaus University Weimar. Co-founder Wang Yixuan used to be the assistant to the chairman of a large industrial group, and co-founder Xie Tianchu holds dual master's degrees from Boston University. More than 50% of the company's employees are R&D personnel, and it has accumulated more than 85 intellectual property rights.
Its core barrier lies in the construction of the VLA Physical Multi-Agent Cluster (VLA Operon) — introducing the vision-language-action model into laboratory scenarios, so that devices are no longer isolated execution terminals, but an intelligent agent cluster with perception, decision-making and collaboration capabilities, realizing decentralized scheduling through local intranet communication. In Shuye's architecture, each experimental device is assigned a different role: some are responsible for global scheduling, some are responsible for visually identifying the position of the plate rack and the barcode of the sample tube, and some are responsible for specific experimental steps such as sample adding, purification, and thermal cycling. These agents form AI agent clusters on demand, and adopt the "Publish-Arrange-Commit" consensus protocol to realize decentralized collaborative scheduling.
Shuye's agent architecture also builds a key positive feedback loop — the data flywheel. Every pipetting action, every experimental step, and every deviation correction will be automatically recorded and structured for output, directly feeding back to the AI iteration algorithm. The device cluster continuously accumulates real-scene data during the experiment execution process, making the system smarter the more it is used.
In terms of core product matrix, the world's first AI voice interactive @Pette series intelligent pipette weighs only 99 grams, and has won three international industrial design awards: Red Dot Award: Best of the Best, Good Design Award (Japan), and London Design Gold Award. The @Robot series experimental robot has an original "robot superimposed on robot" solution, supporting modular deployment and X/Y/Z axis 3D deformation, covering more than ten common scenarios such as PCR, ELISA, and pre-sequencing processing. The self-developed "Shu Xin No.1" general AI engine is equipped with edge computing AI chips, which can provide plug-and-play intelligent upgrading for traditional instruments.
The Intersection of Policy "Tailwind" and Industrial "Spark" Usher in a Window Period for Lightweight Self-Driven Laboratories
Just one week before the closing of this round of financing, Beijing released the "Implementation Plan for Accelerating Artificial Intelligence Empowered Scientific Research in Beijing (2026-2028)", which is China's first medium and long-term AI for Science (AI4S) plan covering the complete system of autonomous laboratories, scientific research agents, and large scientific models. The Implementation Plan clearly proposes to build autonomous laboratories with the capabilities of autonomous perception, autonomous decision-making and autonomous execution, construct a "new closed-loop scientific research model of two-way feedback between computing and experiments for dry and wet labs", and deploy 18 specific tasks around five directions: autonomous laboratory system, high-value scenario application, scientific model system, scientific data foundation, and innovation ecosystem.
The Implementation Plan particularly emphasizes the development of scientific research agents and tool sets, tackling core technologies for multi-agent collaboration in scientific research, and building "AI scientists". Beijing has launched the Uni-Lab-OS operating system to support the evolution of traditional laboratories to intelligent and autonomous operations. A relevant person in charge of the Beijing Municipal Science and Technology Commission stated that scientific research agents can, like scientific researchers, call different scientific tools and experimental instruments to independently complete the whole scientific research process of "hypothesis formulation - scheme planning - computational simulation - experimental verification - innovation discovery".
From a market perspective, AI4S (AI for Science) is listed by NVIDIA as one of the three key AI directions alongside large language models and embodied intelligence, and the long-term market size is expected to reach hundreds of billions of US dollars. The global AI4S market is expected to grow from 4.538 billion US dollars in 2025 to 26.23 billion US dollars in 2032. The global AI in life science market is expected to grow from 21.58 billion US dollars in 2026 to 69.34 billion US dollars in 2031. China's life science instrument market has exceeded 100 billion yuan.
At present, China's AI-enabled scientific research field still faces pain points: traditional automation solutions have high deployment costs and long cycles, creating extremely high thresholds for small and medium-sized scientific research institutions and innovative pharmaceutical enterprises; they lack flexibility and are difficult to adapt to variable experimental scenarios; most laboratories still have "stuck points, broken points, blind points and busy points" such as large manual operation errors, low efficiency at key nodes, and difficulty in forming data closed loops. Shuye Technology adopts a differentiated path of "small, flexible, intelligent and cost-effective" — instead of pursuing large-scale full deployment, it focuses on precise efforts at single-point links to achieve flexible deployment and plug-and-play.
In terms of commercialization, the company has cooperated with nearly 100 universities and scientific research institutes such as the Chinese Academy of Sciences, Westlake University, Zhejiang University, Fudan University, and Peking University Third Hospital, and developed ecological applications with leading enterprises including Dian Diagnostics, Ortho Clinical Diagnostics, Biosan, and Kangrun. It has established 14 business line agency distribution channels and launched an overseas sales network. With the intersection of policy "tailwind" and industrial "spark", Shuye Technology's lightweight self-driven laboratory solution is embracing a broad market space.