Ge Weidong's "Betting" Robot: What is the background of Qianxun Intelligence?

Well-known private equity investor Ge Weidong has shifted his focus from intensive investments in AI industries like GPUs to betting on the robotics sector.

On February 24th, reporters learned from Quanshen Intelligent, a embodied intelligence company, that the company recently completed two consecutive funding rounds totaling nearly 2 billion yuan. Investors include Yunfeng Fund, a leading state-owned enterprise, Chaos Investment, Sequoia China, and others. Additionally, industry capital such as Synstellation Capital, TCL Venture Capital, Minghui Zhiyuan, as well as state-owned funds like Chongqing Industrial Investment Fund and Hangzhou Jintou, along with strategic investors like 360 Fund and Houxue Capital, participated.

With this round of financing, Quanshen Intelligent’s valuation has surpassed 10 billion yuan, officially entering the “billion-yuan club” in the embodied intelligence track.

Chaos Investment is managed by Shanghai Chaos Investment (Group) Co., Ltd., controlled by Ge Weidong. Chaos Investment invests across major global financial markets, including stocks, commodities futures, financial futures, equity investments, and bulk commodity trading.

In the equity investment field, Chaos Investment focuses on early-stage and growth-stage financial opportunities, mainly in sectors such as semiconductors, automotive intelligence, new energy, and advanced manufacturing.

This is not Ge Weidong’s first move in the primary market. Public information shows that he has invested in several prominent tech companies through his controlled Shanghai Chaos Investment Co., Ltd. and personal accounts, many of which have gone public successfully, such as Mu Xi Co., Ltd., Hite Bio, Huizhi Micro, Leap Motor, and Horizon Robotics.

What is the background of Quanshen Intelligent, which caught Ge Weidong’s attention?

Founded in January 2024, Quanshen Intelligent is a embodied intelligence company focused on building a “general brain” for robots. Since its inception, it has attracted diverse industry capital—from leading industrial manufacturers like CATL, Inovance, and TCL, to logistics, retail, and financial infrastructure investors like JD.com and China Merchants Venture, as well as tech giants Huawei and Xiaomi.

The company’s founder and CEO, Han Fengtao, is a serial entrepreneur in the robotics industry, previously co-founder and CTO of Luoshi Robotics. Co-founder Gao Yang is one of the “Berkeley Return Four” in the embodied intelligence field and an assistant professor at Tsinghua University’s Institute for Interdisciplinary Information.

In an interview with Shanghai Securities News, Han Fengtao stated that Quanshen Intelligent adopts a “software-hardware synergy” development strategy, strengthening AI algorithms while focusing on overcoming hardware performance bottlenecks. Its robot products have been piloted in new energy battery, logistics, and office scenarios.

In terms of practical applications, recently, the world’s first mass-produced humanoid embodied intelligence robot line for new energy power battery PACK production was officially put into operation at CATL’s Zhongzhou base. The humanoid robot “Xiao Mo,” developed by Quanshen Intelligent, is working on the production line.

Quanshen Intelligent states that Xiao Mo has become a core piece of equipment on the production line, enabling nearly 1,000 batteries to be produced with zero faults. Its operational speed is comparable to or even faster than skilled workers, demonstrating millisecond-level rapid adaptation to production uncertainties and superhuman flexibility.

Business-wise, Quanshen Intelligent’s models are among the industry’s top-tier. In January 2026, the open-source Spirit v1.5 model from Quanshen became the first Chinese open-source model to outperform Pi0.5 in performance. This model has strong zero-shot generalization capabilities, capable of performing complex tasks such as object wiping, hinge operation, and handling flexible objects without additional training, showing high efficiency across different scenarios.

Behind this breakthrough is Quanshen’s long-term deep cultivation of its technical path. Gao Yang, co-founder of Quanshen, said, “We adhere to the ‘data pyramid’ training philosophy. During pretraining, we did not follow the traditional ‘world model’ approach of predicting each frame—this method consumes huge computational resources and is inefficient. Instead, we pretrain on massive amounts of internet videos of humans, achieving better results with fewer parameters and significantly reducing computing costs.”

Regarding data collection, Gao added, “We have reduced data acquisition costs by 90% through self-developed equipment, enabling large-scale real-world data collection.”

After this funding round, Quanshen Intelligent will continue to increase investment in foundational embodied models and real data systems, deepening industry ecosystem co-creation.

Looking ahead, to rapidly improve robot capabilities, Han Fengtao believes that data is key. Robots need to be exposed to and collect more data.

“The core challenge of embodied intelligence is AI, or the ‘brain’ of the robot. AI is data-driven, so accumulating more high-quality real-world data is essential to enhance robots’ application ability and performance in practical scenarios. How to acquire more data for training models, accelerate model training, and enable models to quickly learn new skills in new scenarios are the next breakthroughs for embodied intelligence,” Han Fengtao said.

(Article source: Shanghai Securities News)