Galaxy General Robotics rakes in $1.1 billion in new funding, led by Ningde Times

On June 23, humanoid robotics star company Galaxy General announced the completion of a 1.1 billion RMBfinancingIt is led by Ningde Times Company and Pryuquan Capital, a subsidiary of Ningde Times, and followed by China Development Bank's CDB Sci-Tech, Beijing Robotics Industry Fund and Jiyuan Capital.This is the largest funding round that has been disclosed in the Body Intelligence track this year.

As early as February this year, there is news on the market that Ningde Times plans to invest in a human-shaped robot body manufacturing company, has contacted Galaxy General and other companies. A person with knowledge of the matter confirmed to the Nandu reporter at that time. In addition, in November 2024, Park Rui Capital, founded by Ningde Times co-founder and vice chairman Li Ping, also invested in the body-shaped intelligence company Qianxu Intelligence.

Galaxy General said, as an industrial strategic lead investor, Ningde era will continue to deepen the industrial synergy, for the galaxy general body with intelligent big model in the industrial field of technology landing and scale application to provide key support.

Up to now, Galaxy General has accumulated more than 2.4 billion yuan in financing, and the financing scale is at the head of the domestic humanoid robot startups. Previously in June and November 2024, Galaxy General successively announced angel round and strategic round financing, respectively, into 700 million yuan and 500 million yuan. Among them, the former set a record for the largest domestic angel round financing in 2024.

Galaxy General did not disclose the company's valuation after the new round of financing, but the company's founder and CTO Wang He revealed in late April at an event, the first half of this year, Galaxy General has become a valuation of 1 billion U.S. dollars "unicorn". According to the reporter's statistics, with the body of intelligent robotics camp jumped over 1 billion U.S. dollars in valuation of the startups include at least Yu Shu Technology, Zhiyuan robotics, Galaxy General and Fourier Intelligence.

A head with body intelligence company practitioners previously told reporters frankly, due to the large-scale commercialization of temporary difficulties, only to melt enough money startups can survive.

From the background of the startup team, Galaxy General adopts the "scientists + industry veterans" model common in the field of body intelligence. Wang He taught at Peking University; co-founder Yao Tengzhou studied under the famous robotics expert Professor Wang Tianmiao, who worked in the robotics research and development center of industrial robotics multinational ABB Group. The reporter learned that in the division of company affairs, Wang He is responsible for software algorithms, while Yao Tengzhou undertakes the work of hardware manufacturing. According to four informed investors, Galaxy General's CEO is Yin Fangming, whose more familiar identity is the first investor in Yu Shu Technology.

In mid-June 2024, Galaxy General released Galbot G1, the first generation of wheeled humanoid robot, which adopts a "two-arm + one-leg + wheeled chassis" body structure instead of a fully humanoid two-legged form to ensure stable performance and a cheaper price, and to take an incremental rather than a one-step approach to development.

Since this year, Galaxy General has repeatedly announced its progress in the research and development of embodied intelligence models and scene landing. on January 9, Galaxy General released the end-to-end embodied grasping base big model GraspVLA. the training of GraspVLA contains pre-training and post-training two parts. The company introduced that pre-training is based on tens of billions of synthetic data. After pre-training, the model can be generalized to never-before-seen, ever-changing real scenes and objects. Post-training for special needs, through small sample learning and fine-tuning, so that the model quickly understand the new task and "learn by example", the acquired skills automatically migrate to different SKU (product style).

As a "synthetic data faction" in the embodied intelligence players, Wang He believes that if you completely bet on the real machine data, you have to first mass-produce humanoid robots to millions of units, and then employ millions or even ten million people to do the data collection before you can hope to reach the generalization, "This will make the development of humanoid robots or VLA fall into a long-term situation that can not be put into practice. long-term situation of not being able to land and not being able to be generalized."

Wang He will synthesize the data analogous to the compulsory education of mankind, at this stage, with the help of tens of billions of scale data, so that the robot thoroughly learn to grasp and release the skills; a small amount of real data after the training is like the vocational education of the on-the-job training, and the level of remote operation work required is the level of the current industry can support.

Based on this with the body to grasp the basis of the large model, Galaxy General and the United States Mission cooperation, the humanoid robot landing to the intelligent pharmacy scene. Wang He revealed at the beginning of June on the Beijing Smart Source Conference, Galaxy General robot has been in Beijing 7 unmanned pharmacy "on duty", this year will be promoted to Beijing, Shanghai, Shenzhen, 100 pharmacies, can identify and capture 5,000 types of drugs SKUs. robots according to the user's order requirements to take down the drugs, delivery riders can enter the password to The robot will be rolled out to 100 pharmacies in Beijing, Shanghai and Shenzhen within this year.

Galaxy General Robotics working in an unmanned pharmacy. Photo: Galaxy General

Wang He said that the operation of a 24-hour store requires "three shifts", it is estimated that the three-year salary of three employees need more than 700,000 yuan. He believes that as long as the price of humanoid robots cheaper than more than 700,000, you can open the market. A Galaxy General staff told reporters that the configuration of the secondary development interface wheeled dual-arm robot Galbot G1 is currently priced at 700,000 yuan.

At the same time, Galaxy General is also promoting the application of cooperation in the field of automotive industry. on June 17, Galaxy General announced that it had established a joint venture company "Boyin Hechuang" with Bosch Capital, a subsidiary of the Bosch Group, a giant in the global automotive industry. The company introduced, the joint venture will focus on complex assembly, intelligent quality inspection and other high-precision manufacturing scenarios, research and development of dexterous robots, to promote the scale of the body artificial intelligence in the industrial field landing.

In addition, on February 11, new energy vehicle supply chain company Tianqi announced the establishment of a joint venture with Galaxy General, with each side holding half of the shares. Eyes on the sky, the joint venture company wuxi Tianqi Galaxy Robotics Limited has been registered on May 26th. Prior to May 6, Tianqi shares chairman and general manager Huang Bin in the results of the meeting to answer the Nandu reporter's question, said the company's cooperation with the Galaxy General focus on the scale of the application of intelligent robots in the industrial field, around the automotive manufacturing scenarios, has been realized in a number of OEMs, battery plants customized application scenarios to validate.

Galaxy General Robotics conducts a scene-on-scene proof-of-concept at the Extreme Krypton factory. Photo: Galaxy General

In addition to establishing joint ventures with the automobile manufacturing industry chain, Galaxy General has also reached cooperation with Mercedes-Benz, Extreme Krypton, Korea Hyundai and other OEMs in robot scene landing. Wang He cited that at the Extreme Krypton plant, Galaxy General's robots are able to flexibly handle boxes randomly unloaded on the ground and complete handling tasks with skills trained from visual observation and synthetic data. "Such work capacity is expected to replace long hours of (manual) labor."