GenRobot Completes $100 Million Series A+ Financing in Q3

On November 2, GenRobot announced that the company completed a $100 million Series A+ financing round in the third quarter of this year. This round of financing was jointly participated by global industrial capital and top financial investors: the industrial parties include SAIC Motor, Malaysia's Eastworlds, Baidu Group, Shoucheng Holdings, and GALBOT, while the financial parties include Guanghe Ventures, SUNHELM, CITIC Jinshi Investment, China Securities Investment, CCB International, and Huatai Zijin. GenRobot was founded in May 2025, and since its establishment, the company has consistently adhered to one thing—building data infrastructure for embodied intelligence, turning human operational actions in homes, factories, and logistics into training data that models can directly use. At present, GenRobot's data flywheel, technology flywheel, and commercial flywheel have all completed closed loops. Effective data delivery in a single quarter has reached one million hours. Over the past two years, most of the excitement in embodied intelligence has been about the robots themselves—who can run, jump, and twist off a bottle cap. But when it actually comes to training models, the industry quickly discovered that the bottleneck is not the machine, but the data. GenRobot co-founder Zhu Yanming has explained their judgment: data is the nutrient essential for model growth, and it will reach scale earlier than models. This is precisely the position GenRobot has cut into: it does not build robots, but instead acts as a "data seller," and it takes an even lighter "bodyless" route: without using real robots, people wear self-developed wearable devices to record hand movements, first-person visuals, along with tactile and depth information, and then convert them into data usable by robots. Why not use real robots? The reasoning is very practical. A robot is not cheap, and data must be collected slowly one unit at a time; with a set of wearable devices, ordinary people can get started after a bit of training, and collection can be concurrent and scaled out. However, "selling data" sounds simple; the hard part is being both abundant and accurate. GenRobot's approach is to build software and hardware together. On the hardware side, it has self-developed the Gen DAS imperceptible wearable device, divided into three categories: Ego first-person view, Fingers bionic dual fingers, and Gripper bionic gripper. It is a rare data collection series in the industry that covers from "head + hands" to the whole body, and cumulative orders have already exceeded 10,000 units. Next, during CoRL, GenRobot will officially release two new products under DAS 2.0: Ego 2.0 and Dex 2.0. On the algorithm side, it has developed the Data Foundation Model, specifically built for the embodied intelligence and world model industries, covering dimensions required for model pretraining such as chain of thought (CoT), 6D pose, depth information, tactile perception, hand tracking, and full-body Mesh, while continuously expanding a standardized human skill system of "domain-scenario-task-skill." Collection has been turned into a production line. It is reported that GenRobot's Gen ADP professional data production line has covered more than 3,000 collectors and tens of thousands of real scenarios, accumulating over one million hours of data and more than 2,000 daily skills. Based on these arrangements, the effective data volume delivered by GenRobot has currently jumped from 100,000 hours per quarter to one million hours per quarter. According to existing data production capacity and delivery plans, the company expects to achieve ten million hours of delivery per quarter in 2027. Less than two years and already 8 rounds of financing. At the end of August this year, when GenRobot announced its Series A financing, it stated that in just one year since its founding, the company had completed 7 consecutive financing rounds, with cumulative financing exceeding $200 million. Among them, the seed round was led by BV Baidu Ventures, the seed+ round was led by RoboSense, the angel round was led by Shunwei Capital, and the Series A was led by Momenta. Ant Group, Didi, and Delian had all appeared on the shareholder list. The completion of the latest financing round means that in a year and a half, GenRobot has completed 8 financing rounds, with cumulative financing exceeding $300 million. In the latest round, SAIC is an automaker, Baidu is an internet giant, Galaxy General is a robot body company, Malaysia's Eastworlds brings overseas scenarios, and Shoucheng has a state-owned background. A group of players originally positioned at different points in the industrial chain are simultaneously putting money into the same data company, pointing to the same thing: no one can bypass the data foundation, and rather than repeatedly collecting data themselves, it is better to bind to a public supplier. More noteworthy is that GenRobot's core team has dual DNA in both algorithm R&D and large-scale data engineering implementation. GenRobot founder Chen Jianxing is a former senior director of algorithms at Momenta. He worked in autonomous driving for nearly a decade, led Momenta's data flywheel mass-production route, and led the delivery of more than one million intelligent driving units. The autonomous driving playbook of "data-driven, closed-loop iteration" has been brought by him into embodied intelligence. GenRobot co-founder Zhu Yanming also comes from a background in mass production of intelligent driving. However, beyond the excitement, one must also stay calm. One million hours per quarter is the company's delivery metric, and ten million hours in 2027 is still only a plan; neither is a number already in hand. Moreover, the data business is naturally prone to homogenization; collecting more does not mean collecting correctly or usefully. Whether customers will continue to pay and whether data can steadily turn into revenue still depends on subsequent orders. Original title: Seeds| Former Momenta executive does not build robots, wins $100 million by "selling data" (Statement: This article is reprinted from a third-party platform. The purpose of reposting this article is to promote information exchange and does not involve profit-making purposes. The views in this article do not represent the position of this site. Unverified information is for reference only and should not be used as a basis for any investment or trading. If there is unintentional infringement of media or personal intellectual property rights, please call or write to inform us, and this site will handle it as soon as possible.)