With four rounds of financing secured in just six months, the professor from Westlake University has shot to fame.

It is learned from PEdaily AI that Westlake Robotics, a general embodied intelligence brain company, recently completed its Series A financing. The interval from the closing of the previous round of financing is less than one month, which is also the fourth round of financing completed by Westlake Robotics in just 6 months, with a total scale reaching 500 million yuan.

Looking at the investor lineup, it covers many well-known institutions such as SAIF Partners, Primitive Venture, Henan Investment Group Huirong Fund, and Haiyuan Fund, building a high-quality capital matrix that brings together state-owned capital, industrial capital and top-tier venture capital firms.

The frequent and intensive capital injection fully confirms the market's high recognition of Westlake Robotics' technical route, product implementation capability and long-term development potential. After the funds from this round are in place, Westlake Robotics will focus on investing in the R&D of the unified full-body large model for humanoid robots, and build a talent cultivation base for embodied intelligence.

The GPT moment for embodied intelligence is coming at an accelerated pace.

Top-tier Entrepreneurial Team, 4 Rounds of Financing in Half a Year

Consolidate the Underlying Foundation of Embodied Intelligence

Westlake Robotics' business started in 2024, derived from the industrial transformation and implementation of the first independent high-quality achievement in the field of artificial intelligence and robotics of Westlake University.

Its founder Wang Donglin is a leading figure in the field of embodied intelligence and robot learning in China, serving as the chief scientist of the National Science and Technology Innovation 2030 Major Project, tenured professor of Westlake University, deputy head of the Department of Artificial Intelligence, and the winner of the 2026 AAAI Best Paper Award. The co-founder Zhang Yue is a tenured professor of Westlake University, vice dean of the School of Engineering, and an internationally renowned expert in the field of natural language processing, leading the exploration of cutting-edge large model technologies.

Looking at the core R&D members of the team, they come from leading technology companies such as Alibaba, ByteDance, Tencent, Huawei, as well as top universities at home and abroad including Oxford, Cambridge, Carnegie Mellon University, UC Berkeley, Technical University of Munich, Tsinghua University and Zhejiang University. They have published more than 300 papers in top conferences such as ICML, NeurIPS, CVPR and RSS, and a number of self-developed cutting-edge scientific research achievements have been applied to underlying algorithm R&D and engineering implementation.

A founding team with profound academic foundation and industrial perspective thus emerges.

This is closely related to Westlake Robotics' strategy of attaching great importance to talents. "The competition of embodied intelligence ultimately lies in the density of talents," said Wang Donglin, the founder. Headquartered in Hangzhou, Westlake Robotics fully relies on Hangzhou's high-level overseas talent introduction and cultivation system to establish a complete mechanism from global recruitment to local settlement.

After several years of development, Westlake Robotics has become a rare general brain company in the industry. Adhering to the all-stack self-developed route of "general brain + full-body humanoid cerebellum + self-developed humanoid ontology", Westlake Robotics is one of the very few embodied intelligence enterprises in China that can fully connect the three core sectors of "embodied AGI cognitive reasoning, whole-body motion control, and integrated humanoid hardware".

Westlake o1 interacts with on-site visitors at WAIC 2026

In 2026, the investment logic of the embodied intelligence track is undergoing a fundamental shift, and the focus of industry competition has begun to shift from the comparison of hardware joint parameters to the general basic model that determines the scene generalization capability of robots — which is exactly the in-depth track where Westlake Robotics has formed its first-mover technical advantage.

As a result, batches of investors are placing intensive bets on the company.

In the past six months, Westlake Robotics has successively secured 4 rounds of financing, with state-owned capital, industrial capital and top-tier venture capital gathering together. The investment logic behind this is not difficult to understand: for state-owned capital, Westlake Robotics covers the full industrial chain layout of humanoid complete machines, general brain and cerebellum, which highly aligns with China's strategy of independent and controllable upgrading of high-end manufacturing and artificial intelligence; market-oriented institutions are optimistic about the company's clear implementation path of integrating software and hardware, whose multi-scenario commercialization model can be quickly scaled up and replicated, with clear growth space.

Closed-loop Commercial Implementation

Accelerate the Arrival of the GPT Moment for Embodied Intelligence

The global embodied intelligence industry has reached a critical watershed.

The problem above the watershed is that the humanoid robot industry is generally trapped in the industry dilemma of disjointed software and hardware, solidified action scripts, and weak environmental generalization capabilities, and most products can only complete preset fixed actions. In this regard, Westlake Robotics chooses to break through the conventional hardware parameter involution in the industry.

As the investor of this round stated: "Westlake Robotics' general dual pre-training architecture for brain and cerebellum, GAE general motion large model, and Westlake o1 humanoid complete machine form a complete technical closed loop, which connects the full chain of 'basic large model - physical humanoid robot - real industrial scenario'. It is one of the few scientific research teams in China that has the capability of mass delivery of general humanoid robots, with outstanding long-term development potential."

First, look at the full-body motion control level. At a time when most humanoid robots' motion control still highly relies on pre-programming and single-task customized training, with limited action flexibility and scenario expansion space, Westlake Robotics' original self-developed general motion model - GAE Avatar System reconstructs the robot motion logic, with three core barriers of millisecond low-latency teleoperation, programming-free action generalization, and cross-ontology compatibility.

GAE realizes cross-ontology remote teleoperation across 1400 kilometers

That is to say, without pre-entering action scripts, the robot can reproduce any real-time action of human beings in real time. With only one set of motion capture equipment, the operator can remotely control multiple robot clusters synchronously, realizing the teleoperation group control mode of one person controlling multiple machines. At the same time, GAE realizes the decoupling of software and hardware. In addition to being compatible with the self-developed humanoid complete machine Westlake o1, it can also output a standardized motion control base to the outside world, which is compatible with the vast majority of humanoid robot ontologies on the market, with extremely high large-scale commercial value.

At the level of collaborative reasoning between brain and cerebellum, in addition to the motion control base, Westlake Robotics has created the industry's unique general dual pre-training architecture for brain and cerebellum, solving the industry pain point of the separation of cognitive reasoning and whole-body motion: the general brain integrates the Vision-Language-Action (VLA) model and the World Model (WM), responsible for environmental perception, task decomposition, complex logical deduction, and global path planning; the GAE general cerebellum undertakes millisecond-level limb output and dynamic posture calibration.

The two models flow data in both directions and form a closed loop, which guarantees the robot's environmental generalization performance and task response efficiency from the underlying architecture, stably drives long-cycle, multi-step complex tasks, and embarks on an original self-developed route with extremely high barriers.

In this way, the complete technical link connecting perception, decision-making and execution is opened up, and a technical closed loop can be implemented for commercialization. After the two sets of underlying core models are mature, Westlake Robotics launched the full-stack self-developed humanoid complete machine Westlake o1 in 2026, completing the complete technical closed loop of "underlying algorithm - pre-trained large model - humanoid hardware complete machine".

President Shi Yigong of Westlake University shakes hands with Westlake o1

At present, Westlake Robotics has reached in-depth cooperation with enterprises and universities such as Alibaba, Intel, JD.com and the Hong Kong University of Science and Technology, continuously accumulating high-quality full-body motion datasets for humanoid robots. It is reported that the company has won orders of nearly 100 million yuan, covering scenarios such as scientific research and teaching, data collection, and power inspection. In the future, it will also output complete sets of solutions for working conditions that are difficult for humans to approach, such as high-risk industrial inspection, post-disaster search and rescue, and remote precision assembly, to extend human operation capabilities.

At the same time, Westlake Robotics has also signed a strategic cooperation agreement with the People's Government of Longyou County, to jointly build a full-domain real-scene training base for humanoid robots, promote Westlake o1 to enter the real production and living environment, and collect the scarce high-quality full-body motion data of humanoid robots in the industry relying on real-scene training. Later, the company will run through the closed-loop link of "data collection - model training - real machine scenario verification", build an open and collaborative embodied intelligence innovation ecosystem, and continuously promote the large-scale and standardized commercial implementation of humanoid robots.

Westlake Robotics signs an agreement with the People's Government of Longyou County to build a full-domain real-scene training base

With the completion of the latest round of financing, Westlake Robotics' "technology + talent" two-wheel drive strategy will be further implemented in depth.

"This round of financing will increase investment in the research of the unified full-body large model for humanoid robots, promote the iterative upgrading of the self-developed brain, and continuously optimize the GAE Avatar System to empower the intelligent capability leap of humanoid robots," said Wang Donglin, the founder.

It is understood that Westlake Robotics has built an independent talent training system, focusing on tackling core underlying technologies. This system is deeply coupled with Hangzhou's talent policy, forming a normalized introduction channel covering overseas doctors, postdoctors to industrial engineering talents. Internally, it has formed a complete R&D team covering algorithms, systems and hardware.

Original technology takes root downward, top talents continue to gather, and industrial capital accelerates entry. The GPT moment of embodied intelligence for Chinese humanoid robots is coming at an accelerated pace.

This article is from the WeChat official account "PEdaily AI", author: Feng Yuchen, published with authorization from 36Kr.