The brains for 200+ embodied robots have raised an angel round.

A dark horse emerges in the embodied brain sector.

PE Daily has exclusively learned that Lingjing Zhiyuan, incubated by the Shanghai Artificial Intelligence Research Institute, has completed two consecutive financing rounds — Angel and Angel+ — totaling over 100 million RMB. The funding was led by Matrix Partners China, with strategic follow-on investment from State-owned Assets of Minhang District, Shanghai.

Prior to this, the company operated in "stealth mode," and both rounds were oversubscribed. Sources reveal that without any public announcement of its financing plans, Lingjing Zhiyuan attracted dozens of top-tier financial investors, industrial CVCs, and industry capital parties for intensive due diligence within a financing window of less than two weeks. The company ultimately closed the round early and exceeded its fundraising target.

Founded in July 2025, Lingjing Zhiyuan positions itself as a domestic full-stack computing infrastructure provider native to embodied intelligence, rather than a mere upper-layer algorithm solution vendor. It focuses on the underlying infrastructure of embodied intelligence, offering a domestically developed, fully controllable end-to-end hardware-software integrated full-stack solution that is universally compatible with all robot forms including humanoid, wheeled, and quadruped models.

In other words, when addressing the challenges of the embodied brain, Lingjing Zhiyuan does not solely concentrate on "soft" algorithms and models, but organically unifies three core components: models, chips, and systems.

Notably, the lead investor Matrix Partners China doubled down on its investment after leading the Angel round, directly securing the vast majority of shares in both the Angel and Angel+ rounds.

The old shareholder's decision to increase its stake significantly demonstrates strong confidence in the project. Matrix Partners China has laid out investments across the entire value chain, covering large models, embodied intelligence, GPUs, and computing infrastructure. Its portfolio companies including Muxi, Unitree Robotics, Galaxy Universal, Minimax, and Wuwu Xinqiong have now entered a harvest phase. This continuous heavy investment in Lingjing Zhiyuan carries certain indicative significance for the market.

Currently, the differentiation trend in the embodied intelligence industry has become very clear: financing windows for well-known robot body manufacturers such as Unitree, Agibot, and Galaxy Universal have long been closed, and dozens of embodied enterprises have collectively entered the "exam-taking period" of share reform and IPO capitalization. In the "shovels-selling" segments of upstream components, computing infrastructure, and underlying systems, capital enthusiasm continues to rise, and securing an unassailable industrial position has become another widely recognized strategic logic.

At present, the core bottleneck restricting embodied intelligence from moving from demos to real-world deployment lies in the embodied brain. From Vision-Language-Action (VLA) models to world models, new technological narratives and startup teams are emerging in endlessly. Lingjing Zhiyuan focuses on developing solutions for robot brains, and from the very beginning, it has taken a differentiated, more "asset-heavy" path.

"The Android of the Robotics Era"

Established in July 2025, Lingjing Zhiyuan positions itself as a provider of embodied intelligence computing infrastructure and a native operating system, dedicated to building a domestically developed full-stack solution that integrates chips, the MS-OS native system, and a brain-inspired collaborative model for the cerebrum and cerebellum, aiming to be the "Android of the robotics era."

Both founders of Lingjing Zhiyuan are serial entrepreneurs with over 20 years of industrial experience, and the core team members come from leading enterprises in the chip, communications, and industrial measurement and control sectors such as Huawei, ZTE, and Kunlunxin.

Sun Bo, Founder & CEO, holds a bachelor's and master's degree in Artificial Intelligence from Shanghai Jiao Tong University and a Ph.D. from Zhejiang University. He previously co-founded JITI Measurement & Control, a publicly listed industrial measurement and control company, and served as its General Manager, leading the company to achieve an annual revenue of over 1 billion RMB and advance toward an upcoming Hong Kong Stock Exchange IPO. Co-founder Xu Haijiang holds a master's degree from the Institute of Electronics, Chinese Academy of Sciences. He is the founder of BlueRidge Technology and a former core communications expert at Ericsson China R&D.

Song Haitao, President of the Shanghai Artificial Intelligence Research Institute and Co-Leader of the National Information Technology Standardization Committee's Humanoid Robot (Embodied Intelligence) Working Group, and Yan Weixin, Chief Scientist of the Research Institute, are all members of Lingjing Zhiyuan's Technical Committee, providing top-tier scientific research expertise, cutting-edge technology guidance, and comprehensive connections to industrial resources for the company.

Lingjing Zhiyuan's differentiation lies in the fact that it does not only focus on "soft" algorithms and models, but organically integrates the three key links: models, chips, and systems.

This strategic direction stems from the company's insight into a widespread industry pain point: in the robotics sector, the "cerebrum" for cognitive decision-making and the "cerebellum" for motion control are usually developed separately. The traditional general-purpose computing architecture based on CPU+GPU cannot achieve a millisecond-level real-time closed-loop design for "perception-decision-execution," which leads to problems such as delayed perception preprocessing, lagging motion control responses, and disjointed scheduling between hardware and software. These issues cause robots to generally "be able to think but fail to execute properly," which is also the core bottleneck restricting large-scale mass production.

Lingjing Zhiyuan has embarked on a differentiated path, breaking away from the mainstream approach that only optimizes upper-layer models. Instead, it reconstructs the entire robot computing system from the ground up. The company has pioneered a super-heterogeneous architecture, following the logic of "physical integration, logical separation," and created a groundbreaking integrated collaborative system for the cerebrum and cerebellum. This system integrates the cognitive reasoning cerebral cortex and the real-time motion control cerebellum on a single physical hardware platform, while retaining the flexibility of layered iteration. It incorporates non-inferential instinctive response capabilities into the system closed loop, significantly reducing the over-reliance on large general-purpose models, and fundamentally solving problems such as robot motion jitter, response delay, and unstable long-term operation.

Accordingly, Lingjing Zhiyuan has adopted a capital-intensive full-stack in-house R&D strategy, covering three major segments: a self-developed domestically produced embodied-specific operating system MS-OS, which is analogous to CUDA for NVIDIA and is compatible with all types of robot hardware; the "Dvořák" heterogeneous hardware computing solution, which supports both mature overseas ecosystems and fully localized full-stack development; and a brain-inspired collaborative model system, which reduces reliance on the computing power of a single large model and enhances real-time motion control and adaptive capabilities in complex environments.

Sources reveal that Lingjing Zhiyuan will officially launch the design of its dedicated embodied native chip in mid-2026, complete the chip tape-out in 2027, and form a fully closed-loop domestic competitive barrier covering "chips-systems-models." Currently, the FPGA prototype has completed full-process verification.

Securing Over 200 Embodied Intelligence Orders, Including Partnerships with Leju, StarMap, and Parsini

Backed by the Shanghai Artificial Intelligence Research Institute, Lingjing Zhiyuan has easier access to policy support, scientific research resources, industry partners, and state-owned capital at its early stage compared to ordinary startups.

Within just one year of establishment, Lingjing Zhiyuan has demonstrated highly efficient implementation capabilities. The company discloses that, currently in the stage of continuous iterative verification, it has connected with over 300 embodied system manufacturers, covering more than 70% of the industry. Over 200 customers have completed mass deployment and placed small-batch orders, with some leading customers placing orders for more than 1,000 units each.

At present, Lingjing Zhiyuan's client list includes many leading players, covering mainstream humanoid and quadruped robot manufacturers such as Leju, StarMap, Parsini, Zhifang, and Fourier Intelligence.

In the past few years, industry financing resources have been highly concentrated in robot body manufacturers. Now, terminal manufacturers have generally entered the capitalization "exam-taking period." According to incomplete statistics, there are more than 20 domestic embodied body enterprises with a valuation exceeding 10 billion RMB; Unitree Robotics is about to list on the Sci-Tech Innovation Board; and multiple companies such as Deep Robotics, Galaxy Universal, StarMap, and Zhifang are all in the process of going public.

As the industry's capitalization process significantly outpaces the speed of commercial implementation, the differentiation in investment and financing is intensifying: investors have become more conservative in investing in terminal robot bodies, and capital continues to flow into the "shovels-selling" segments such as components, computing layers, and infrastructure. The embodied brain has become one of the hottest strategic directions in the market.

At the recent embodied intelligence special session of the PE Daily Summit, multiple guests stated that developing the robot brain is becoming a mainstream strategy. Correspondingly, the number of players focusing on the embodied brain is increasing, and "world models" have replaced VLA as the most popular technological direction and investment theme recently.

In this context, Lingjing Zhiyuan, which conducts full-stack in-house R&D of both hardware and software, is following a technological and business path that goes against conventional market consensus.

Of course, there are also objective challenges facing Lingjing Zhiyuan and other embodied brain/infrastructure enterprises. Although 2026 is regarded as the "first year of mass production," with industry forecasts putting the total annual robot shipment at 50,000 to 100,000 units, the overall market size is still limited, and the short-term market ceiling for embodied brains/infrastructure is not very high. It will still take several years for embodied intelligence to achieve large-scale commercial deployment. Full-stack in-house R&D of chips and operating systems requires long-term capital investment, so the company needs continuous financing to support R&D, mass production ramp-up, and the construction of industrial standards and ecosystems.

On the other hand, in the long run, once the industry enters the phase of large-scale mass production, Lingjing Zhiyuan's accumulated frontline engineering experience, self-developed chips, hardware-software adaptation data, and customer cooperation foundation will all be transformed into competitive barriers that are difficult for competitors to overcome in the short term.

In conclusion, at a time when various technologies in embodied intelligence have not yet converged, the differentiated development of Lingjing Zhiyuan represents a new paradigm and noteworthy case worthy of attention.

This article is from the WeChat Official Account "PE Daily", authored by Cao Weiyu, and published with authorization from 36Kr.