Raising 3 billion yuan within the year, why can Qianxun Intelligence attract investments from Jack Ma and Lei Jun?

Why does AI · General Brain Technology attract many celebrity investors?

Qianxun Intelligence announces an additional 1 billion yuan of financing, marking its second funding round in 2026. This round was led by Sequoia Capital China and Yunfeng Fund, with total funding this year reaching 3 billion yuan and a valuation exceeding 20 billion yuan.

On April 7th, Qianxun Intelligence announced the completion of a new 1 billion yuan funding round, jointly led by Sequoia Capital China and Yunfeng Fund, with follow-on investments from Dacheng Caizhi, Galaxy Source, Turing Fund, New Ding Capital, Gengxin Capital, and others.

Since its founding just two years ago, focusing on embodied intelligence “general brain,” Qianxun Intelligence has raised over 4 billion yuan across seven funding rounds, with a valuation surpassing 20 billion yuan.

The reason why Qianxun Intelligence can attract capital attention is closely related to the backing of its founding team. Master’s graduate Han Fengtao from Zhejiang University has extensive hardware R&D experience, Gao Yang has deep expertise in software, and Zheng Lingyin specializes in marketing.

**Jack Ma and Lei Jun jointly bet on it

Less than two months after the previous Series A funding of 2 billion yuan involving Sequoia China and Chaos Investment, on April 7th, the embodied intelligence unicorn Qianxun Intelligence announced a new 1 billion yuan funding round.

This round was led by Sequoia Capital China and Yunfeng Fund, with follow-on investments from Dacheng Caizhi, Galaxy Source, Turing Fund, New Ding Capital, Gengxin Capital, and others. Since then, Qianxun Intelligence’s valuation has exceeded 20 billion yuan, making it one of the youngest billion-yuan valuation companies in the embodied intelligence track.

Since the seed round in March 2024, the company’s seven funding rounds have featured a star-studded lineup of investors, including industry funds such as CATL’s Ba Rui Capital, Lei Jun’s Sequoia Capital China, Jack Ma’s Yunfeng Fund, JD.com’s JD Technology, Huawei’s Hubble Investment, and Chaos Investment led by Ge Weidong, as well as China Internet Investment Fund, China Merchants Venture, Prosperity7 under Amei Venture Capital, and Sequoia China. Multiple rounds of continued follow-on investments from these investors have also occurred.

Currently, Qianxun Intelligence has raised over 4 billion yuan in total. For startups, this funding is crucial for advancing R&D, accelerating production lines, and gaining industry competitiveness.

According to data from the China Academy of Information and Communications Technology, by 2025, the total financing in China’s embodied intelligence field reached 73.54B yuan, with over 740 investment events, indicating sustained industry enthusiasm. Since early 2026, the robotics sector’s primary market has been even more active.

Unlike other robotics companies that produce humanoid robots as complete machines, Qianxun Intelligence focuses on providing underlying intelligent control and decision-making capabilities for various robots, aiming to build a “general brain” for robots.

Qianxun’s “general brain” complements Xiaomi’s hardware ecosystem. As early as 2024, Lei Jun’s Sequoia Capital China led the seed round of SpiritAI, a subsidiary of Qianxun Intelligence, and continued to support subsequent angel and other funding rounds. As one of the lead investors in this 1 billion yuan round, Lei Jun’s Sequoia Capital China described Qianxun Intelligence as choosing a “difficult but correct” path, building barriers with the “data pyramid” concept.

**What makes Qianxun Intelligence attractive?

As a company valued at 20 billion yuan just two years after its founding, why does Qianxun Intelligence attract so many star investors?

For Qianxun Intelligence, solving the generalization ability of embodied products and meeting diverse operational scenarios depend on high-quality data.

Some of the data used to train foundational models come from real-world scenes such as factories and warehouses. While working, embodied intelligence products form a physical data loop, collecting data that further optimizes the “construction work” efficiency of embodied robots, improving task execution accuracy at bottlenecks.

Qianxun Intelligence fully controls the humanoid robot Moz1, officially released in June 2025. The robot, equipped with the Cash VLA model, achieved over 99% success rate in precise operations on CATL’s first large-scale production line globally.

After long-term on-site work at CATL’s factory, refining models through coding, by the end of 2025, another Qianxun robot, Mozi, officially entered the factory, performing battery insertion and removal tasks on CATL’s new energy power battery PACK production line, collecting reinforcement learning data. This marked the first deployment of Qianxun’s end-to-end model.

Building on the successful mass production of thousands of batteries at CATL, in March 2026, Qianxun Intelligence signed a cooperation agreement with JD.com (one of its investors), deploying Moz robots at JD MALL to handle coffee serving and shopping guide roles. The two sides announced plans from 2026 to 2029 to accelerate the application of embodied intelligence in retail through customized consumer products, technical cooperation, scenario deployment, and marketing collaboration.

Beyond factory scenarios, Qianxun’s self-developed wearable devices and robots continue to collect data. To date, the company has accumulated over 200k hours of multi-type real interaction data, covering internet videos, remote operation, wearable data collection, and more. By 2026, the total is expected to exceed 1 million hours.

Additionally, the company’s self-developed wearable data acquisition devices have been iterated to the 5th generation, reducing data costs to one-tenth of traditional methods, clearing obstacles for large-scale data collection.

However, after improving quality and reducing costs, investors ultimately focus on profitability. Based on the company’s recent statements, the primary goal for 2026 remains improving large model performance, with only a small revenue target of 100 million yuan.

Founder Han Fengtao stated that, under the premise of high-quality performance, expanding to more scenarios will be feasible for large-scale commercialization by late 2027 to 2028.

**The “Iron Triangle” founding team

The “Iron Triangle” behind Qianxun Intelligence has an impressive background.

Master’s graduate Han Fengtao from Zhejiang University served as co-founder and CTO of Luoshi Robotics from 2015 to 2023, leading industrial robot technology R&D and product delivery. He witnessed the domestic industrial robot localization rate rise from 3% to 50%, leading the team to mass-produce and deliver the first domestic collaborative robots, with over 20k units delivered. He also advanced Luoshi’s multi-billion yuan financing plans.

With nearly a decade of startup experience, Han Fengtao accumulated extensive R&D and delivery experience in joints, motors, and complete machines. In 2023, with the emergence of ChatGPT, Han Fengtao recognized that large models are key to robot differentiation, decisively abandoning Luoshi’s IPO opportunity, and inviting Gao Yang, a Tsinghua University graduate and assistant professor at Tsinghua’s Cross-Information Research Institute, to develop the robot’s “brain,” along with Zheng Lingyin, former Luoshi marketing director, responsible for commercialization.

In January 2024, the three jointly founded Qianxun Intelligence, a company dedicated to creating a “general brain” for robots, starting to implement the vision of integrating AI large models into the physical world through technological revolution.

At that time, the domestic robotics track was just beginning. Gao Yang, one of the “Berkeley Return Four,” along with three other Berkeley PhDs, returned to China to engage in embodied intelligence. Gao Yang, a PhD from UC Berkeley in computer science, collaborated deeply with robotics learning scholar Sergey Levine, co-founder of the American embodied intelligence company Physical Intelligence (Pi).

Despite the capital “flood,” Han Fengtao remained focused on “performance enhancement” rather than “making money.” His demand for model performance remains extremely strict.

Recently, Han Fengtao stated that before the model reaches a critical point, forced commercialization is meaningless. Current embodied models still require extensive fine-tuning for tasks. Only when zero-shot learning can achieve 70-80% success rates can intelligent robots rapidly scale at significantly reduced marginal costs.

Currently, Qianxun Intelligence faces fierce competition in the embodied track, with investors participating broadly in funding, perhaps only hoping to see a few dark horse companies emerge.

Editor | Chen Bin