Author | Xiang Xin

At the dawn of 2025, a dark horse emerged in the embodied intelligence race, dedicated to developing a "general brain" for robots.

On January 7, Zhipingfang Technology, a pioneering firm in embodied large model research and development, announced the successful completion of a Pre-A round strategic funding totaling hundreds of millions of yuan.

This round was co-led by Fortune Capital and DH Capital, with participation from Cornerstone Capital.

Since its inception in April 2023, Zhipingfang has secured two rounds of funding.

In the angel round held in August 2023, Zhipingfang attracted investments from renowned institutions such as SEE Fund, Qingzhi Capital, and SDIC Innovation Capital.

Investors' favor towards Zhipingfang stems primarily from its formidable product competitiveness.

A robot comprises three core components: the brain, cerebellum, and body.

The brain is responsible for processing and analyzing information, performing learning, reasoning, and decision-making tasks.

A robust brain serves as the pivotal driving force for robots to transcend application limitations, expand functionalities, and diversify application fields, directly influencing their competitiveness in the intelligent era.

Zhipingfang stands as the world's first startup to propose and systematically develop AGI in the physical realm. Its domestically crafted embodied large models significantly outperform leading Silicon Valley models in terms of general spatial perception and generalization operation success rates.

Its AI2R Brain, a general embodied intelligence system, and the Alpha Bot robot have garnered technical recognition from multiple leading enterprises.

Moreover, Zhipingfang's commercial endeavors are progressing smoothly.

The company has inked contracts with numerous domestic and international top-tier automakers, high-end manufacturing enterprises, and internet giants, securing a plethora of commercial orders and realizing tens of millions in confirmed revenue in 2024. Zhipingfang is the pioneering intelligent robot company to embark on commercial promotion.

Post this funding round, Zhipingfang Technology will allocate additional resources to technology research and development, thereby accelerating its commercialization efforts.

Led by former scientists from XPeng and OPPO, Zhipingfang boasts both research and development and mass production expertise.

Zhipingfang's outstanding technical achievements are attributed to its seasoned R&D team with profound technical expertise.

Guo Yandong, the founder and CEO of Zhipingfang, has worked at Microsoft, XPeng, and OPPO, leading AI research and development for hundreds of millions of smart terminals. He possesses over 15 years of experience in AI academic research and industrial practice.

Dr. Guo Yandong graduated from Purdue University, a prestigious engineering institution in the United States. During his doctoral studies, he was mentored by Jan P. Allebach and Charles A. Bouman, both members of the US National Academy of Engineering in the field of AI.

Post-graduation, he joined Microsoft's headquarters in the United States as a researcher, spearheading the development of several cutting-edge AI technologies and products.

His early AI research significantly contributed to Microsoft AIaaS, Bing search, and other products. Notably, the Custom Vision service (image recognition service) was the world's first commercial attempt to utilize "pretrained models + scenario fine-tuning," reducing the technical threshold and cost for large-scale AI application, thereby fostering widespread AI adoption across various industries.

In 2018, Guo Yandong joined XPeng Motors, where he oversaw AI vision and perception research and development. In 2020, he joined OPPO as the Chief Scientist, focusing on intelligent perception and interaction technology.

His diverse work experiences have endowed Guo Yandong with extensive R&D experience and a forward-looking vision in machine intelligent perception and interaction.

In April 2023, Guo Yandong founded Zhipingfang Technology, one of the earliest technology companies dedicated to advancing spatial intelligence technology, particularly spatial perception technology.

For AI to transition from the digital to the physical world, it must possess three core capabilities:

The ability to perceive spatial object information

The ability to comprehend and reason about spatial relationships and layouts

The ability to make decisions and execute actions based on the aforementioned understanding

These capabilities collectively constitute spatial intelligence, the core technology where Guo Yandong and the Zhipingfang Technology team excel and continue to invest heavily in research and development.

The Zhipingfang team comprises top talents in AI, robotics, and smart terminals. Core members hail from internationally renowned enterprises such as Microsoft, XPeng, OPPO, and Momenta, as well as prestigious academic institutions like Tsinghua University, Peking University, the Chinese Academy of Sciences, the University of California, Berkeley, and Columbia University.

The team members possess both core technology R&D experience and mass production expertise. They not only have extensive experience in full-stack large model technology research and development but also rich experience in large-scale AI technology application across various smart terminal forms.

With tens of millions in confirmed revenue, does Zhipingfang's model outperform Silicon Valley enterprises?

Zhipingfang positions itself as a leader in multimodal large models and future AGI terminals, committed to the research, development, and production of general-purpose intelligent embodied terminals.

Currently, its focus is on three main areas: spatial perception basic models, end-to-end embodied large models, and embodied intelligent robots.

Zhipingfang Technology is ahead of the curve in spatial perception model research and development.

In April 2024, "AI Godmother" Li Feifei founded World Labs, a company dedicated to spatial intelligence research, drawing widespread industry attention to this domain.

However, Zhipingfang Technology had already commenced researching and developing spatial perception basic models based on years of accumulation in early 2023.

After more than a year, Zhipingfang's spatial perception basic model surpassed the SOTA models of leading Silicon Valley enterprises in multiple tasks, earning national-level certification and endorsement.

Zhipingfang's prescient anticipation of technology development trends is also evident in its proactive layout of end-to-end embodied large models.

In June 2024, the VLA end-to-end paradigm adopted by the American embodied intelligence company Physical Intelligence garnered widespread attention. This technical model is considered the "ultimate" solution for embodied intelligence, inspiring many startups to emulate.

Subsequently, Physical Intelligence's valuation soared.

In November 2024, Physical Intelligence secured a $400 million investment from investors including Amazon founder Jeff Bezos, pushing its valuation to $17 billion.

However, Zhipingfang had already established the "end-to-end" technical paradigm before PI was founded. Also in June 2024, Zhipingfang collaborated with Peking University to launch the RoboMamba end-to-end embodied large model for robot operation.

RoboMamba was not only publicly applied earlier than PI's OpenVLA but also significantly surpassed Google's RT series models in the generalization ability of unseen tasks.

This achievement led to Zhipingfang's successful selection for the prestigious AI conference NeurIPS 2024, marking the first time a Chinese embodied startup gained international recognition in the VLA end-to-end direction.

Additionally, during the same period, Zhipingfang launched the general embodied intelligence system AI2R Brain and the general intelligent robot Alpha Bot.

Built end-to-end on a multimodal large model, AI2R Brain can be deployed on Alpha Bot, enabling the robot to accurately perceive the world, achieve human-like natural interaction, and execute tasks in highly flexible, compliant, and diverse scenarios.

Zhipingfang's pursuit of embodied intelligence technology extends beyond the general brain to the body as well.

In August 2024, Zhipingfang Technology unveiled the newly upgraded robot Alpha Bot 1S.

Alpha Bot 1S boasts enhanced flexibility, a broader operating range, superior perception capabilities, and extended battery life. The robot integrates wheel design, folding and lifting structures, and a highly flexible robotic arm, featuring a 50% increase in joint count and a 200% surge in sensor count compared to its predecessor.

Equipped with the embodied large model AI2R Brain, Alpha Bot 1S can swiftly adapt to its environment and execute tasks flexibly, mimicking human capabilities.

During the 2024 World Robot Conference, Alpha Bot 1S conducted live demonstrations in work and daily life scenarios, accurately placing parts and serving coffee to attendees.

Thanks to its forward-looking strategy and exceptional product strength, Zhipingfang's commercialization endeavors have yielded initial success.

Zhipingfang's Alpha Bot series robots and embodied large model AI2R Brain have demonstrated application potential across various scenarios, including flexible industrial assembly, material handling, and public services.

Zhipingfang revealed that it has signed contracts with numerous domestic and international top-tier automakers, high-end manufacturing enterprises, and internet giants, securing a plethora of commercial orders and realizing tens of millions in confirmed revenue in 2024.

Guo Yandong believes that embodied intelligence stands on the cusp of industrialization, with embodied robots poised to fully integrate into daily life and reshape the operational mode of every industry in the next 5 to 10 years.

In terms of technology, products, and commercialization, Zhipingfang has already seized the initiative.

As a leading technology enterprise with a talented workforce, advanced technology, and substantial funds, Zhipingfang's future development will serve as an important benchmark for the entire embodied intelligence industry.