Produced by | Zhangtong Media
Its post-investment valuation has exceeded CNY 7 billion.
Recently, Shanghai Moqi Wanxiang Intelligent Technology Co., Ltd. (hereinafter referred to as "Moqi Intelligence") underwent industrial and commercial changes, adding Hangzhou Haoyue Enterprise Management Co., Ltd., an affiliated company of Alibaba, as a new shareholder, while its registered capital was increased to CNY 1.5424 million.
Founded in August 2025, Moqi Intelligence focuses on the R&D of embodied intelligent robots. Just half a year after its establishment, the company completed an angel round of financing series exceeding CNY 1 billion, with a post-investment valuation surpassing CNY 7 billion. This is currently one of the largest publicly disclosed initial funding rounds in China's embodied AI sector.
Now, less than three months after the completion of the previous funding round, an affiliated company of Alibaba has doubled down on its investment.
The Alliance of a Huawei "Genius Youth" and a Logistics Veteran
Moqi Intelligence was co-founded by Huang Qingqiu and Gao Wenli, and the backgrounds of the two founders are perfectly complementary.
CTO Huang Qingqiu comes from a technical background. He earned his bachelor's degree from the Department of Automation at Tsinghua University and his Ph.D. from the Multimedia Laboratory at the Chinese University of Hong Kong, under the supervision of Lin Dahua, Chief Scientist of SenseTime. He has long been engaged in research on robot control, computer vision, and AI (Artificial Intelligence) algorithms, and has published dozens of papers in top-tier conferences.
In 2020, he joined Huawei's Intelligent Automotive Solution Business Unit as a Huawei "Genius Youth". He later became the head of the Intelligent Driving AI Department, leading Huawei's intelligent driving algorithm breakthroughs and mass production from ADS 1.0 to 4.0. He is the first person in the industry to achieve million-scale mass production of a single-stage end-to-end system.
During his five years at Huawei, Huang Qingqiu witnessed the transition of intelligent driving from modular, rule-based systems step by step to AI-driven, end-to-end models. Having gone through several rounds of technological route changes, he has accumulated a strong sense of judgment regarding technological directions, as well as the ability to rapidly validate and iterate.
CEO Gao Wenli is the co-founder of the cross-border logistics company iMile. He built a global operation system covering 20 countries from scratch, driving the company's annual revenue to exceed CNY 4 billion. Earlier, he spent 11 years in overseas markets at Huawei's Carrier Business Group.
If Huang Qingqiu excels at "making the technology work", Gao Wenli excels at "selling the products and running through the scenarios".
The two chose to enter the embodied AI field in 2025. While outsiders might think it is somewhat late, Huang Qingqiu feels that two technological singularities are just converging in these two years.
One is the GPT moment. Large language models have solved the problems of high-level semantic understanding and task decomposition, giving robots a "cognitive brain" for the first time; the other is the end-to-end moment in intelligent driving, proving that it is feasible to use neural networks to drive a physical entity to complete real-time closed-loop control.
The combination of the two founders determined that from day one, Moqi Intelligence would not take the route of single-point demos, but instead anchor itself in the system engineering capabilities of integrated hardware and software, using commercial service scenarios as a technical training ground, and aiming for the ultimate goal of general-purpose home robots.
Putting the "Brain" Directly into the Robot Itself
In August 2026, Moqi Intelligence released its self-developed embodied model framework, MoRA.
The name MoRA itself carries an attitude. In the industry, mainstream models are mostly named according to their architecture, such as VLA, WM, and WAM. Moqi Intelligence chose to name it based on capabilities—Mo stands for Moqi, R stands for Reasoning, and A stands for Autonomy.
The working mode of most robots is that the cloud brain issues commands, and the local hands and feet execute them. Moqi Intelligence has taken a different path—putting the capabilities of the "brain" directly into the execution layer, enabling the robot to make its own decisions, memorize things, plan, and work continuously without relying on the cloud.
In terms of system design, Moqi Intelligence divides the robot into three layers. System 2 runs in the cloud, responsible for conversing with humans and decomposing instructions into tasks; System 1 is the "brain" of the robot, perceiving the environment and making actions; System 0 is the "cerebellum", converting actions into joint signals.
The core idea of Moqi Intelligence is to put more capabilities into System 1, making the edge-side brain smarter and smarter. Once System 1 becomes smarter, it brings about leaps in two key capabilities.
First, stronger task execution capability. Previously, System 2 gave System 1 detailed operations like "hold the two corners of the clothes and turn them over". Now, it can give coarse-grained tasks like "fold the clothes", allowing System 1 to decompose and complete them on its own.
Second, enhanced temporal capability. It has shifted from previous short-term, reactive actions to autonomously completing long-term or medium-term complete tasks that require error correction and backtracking capabilities.
Beyond the model, data, which is regarded as the biggest bottleneck in the embodied AI industry, is equally critical.
Moqi Intelligence has chosen an optimized non-embodied data collection route. The company's self-developed collection device, MORPHI Sense Kit, can achieve millimeter-level trajectory reconstruction in extreme scenarios such as weak textures and high reflections. The equipment goes directly into frontline commercial scenarios such as hotels and business apartments, where real operators generate data in their daily work.
At the same time, Moqi Intelligence has built an infrastructure capable of digesting large-scale data. It has currently accumulated 30,000 hours of real-scenario data and plans to complete a data reserve of 150,000 to 200,000 hours by the end of 2026.
This technology has been implemented on the MORPHI KINO wheeled service robot, focusing on daily service scenarios.
At the WRC 2026 site, Moqi Intelligence demonstrated a 15-minute long-horizon household task: organizing items across different areas, operating the refrigerator and washing/drying equipment, and folding clothes, with a series of actions completed autonomously throughout the process.
Shanghai's Embodied AI Industry Accelerates Its Momentum
In July this year, the highly anticipated Moqi Intelligence officially settled in Yangpu, Shanghai. Yangpu is cultivating embodied AI as a key future industry and has introduced special plans to clarify the development direction of the track.
In the "15th Five-Year Plan for the Development of Innovative and Modernized Industries in Yangpu District", embodied AI is included in the CNY 10 billion-level future industry track of "Future Intelligence". It proposes the development of humanoids, general exoskeletons, robotic arms, and other main bodies, and promotes industrialization breakthroughs in core components such as edge-side chips and dexterous hands.
Currently, Yangpu is accelerating the construction of an industrial development foundation covering core components, complete machines, algorithms, and application scenarios. The Wangu Embodied AI Innovation Demonstration Site focuses on real-machine data collection and scenario verification; the Fuxing Island Embodied AI Real-scene Space Experimental Training and Testing Platform opens real urban scenarios such as Music Festival CityWalk, coffee delivery, urban river patrol, and market shopping guide to robot R&D teams for practical training and testing.
Yangpu's waterfront has also gathered numerous major internet technology companies, and the effect of industrial agglomeration is accelerating. Projects such as the Meituan Shanghai Technology Center, Bilibili New Generation Industrial Park, and Douyin Group Shanghai Waterfront Center are under intensive construction and will be completed and put into use in the second half of this year, continuously shortening the spatial radius of upstream and downstream industrial synergy.
Moqi Intelligence's move to Yangpu also hides an industrial clue from the shareholder dimension.
The new shareholder, Hangzhou Haoyue, is an affiliated company of Alibaba. In recent years, Alibaba has been intensively deploying in the embodied AI track, and Shanghai is a key city for the Alibaba ecosystem's layout in the AI and robotics tracks. The DAMO Academy Shanghai team and Robbyant have successively landed. Coupled with the Alibaba Cloud Jinshan Intelligent Computing Center project under accelerated construction, a supporting layout from algorithm R&D to computing power support is being built.
Shanghai has the most complete embodied AI industrial chain and talent reserve in the country. A group of leading enterprises such as Fourier, Keenon, Agibot, and Kepler have already taken root here. For a company aiming at the ultimate goal of general-purpose home robots, settling in Shanghai also means being closer to scenario customers and closer to the most active peers in the industry.
A steady stream of embodied AI startups continues to emerge, backed by the huge growth space of the track. Shanghai is becoming an important soil for many new players to start their race.
Text | Zhouyu Editor | Yida