Linkerbot, a company specializing in dexterous hands for embodied intelligence, has completed a Series A+ round of funding, raising a nine-figure RMB sum. The round drew participation from investors including Zhejiang Provincial Innovation Investment Group, Deqing Industrial Investment, Leju Robotics, CDH Investments, Ginnva, Dongfang Precision, Aux, Huafu Investment, and Hongyi Fund, among others.
This marks Linkerbot’s fourth financing event in eight months. Its investor base now spans financial institutions and industrial backers such as Ant Group, HongShan, Wankai New Materials, and Dioo Microcircuits.
As embodied intelligence matures, Linkerbot has begun accelerating commercialization. Zhou Yong, founder and CTO of Linkerbot, told 36Kr that the company now operates across three primary scenarios: scientific research, humanoid robots, and industrial applications. Monthly orders have reportedly exceeded 1,000 units, and overseas clients include Samsung and Siemens.
Since its founding, Linkerbot has kept one long-term goal: to become the first company to produce one million dexterous robotic hands.
Today, Linkerbot’s portfolio includes three actuation technologies, built on tendon-driven, direct-drive, and linkage-driven mechanisms. Its products span 11–42 degrees of freedom (DoF).
The Linker Hand series of high-DoF robotic hands is said to be the first of its kind to enter mass production, surpassing 1,000 units.
Among its flagship products, the L20 features 21 DoF and comes in both general-purpose and industrial versions. It can operate various human tools and is priced at RMB 49,900 (USD 6,986). The top-end L30, with 22 DoF, supports precision assembly, flexible grasping, and surgical assistance tasks, selling for RMB 99,900 (USD 13,986).
Linkerbot’s lower-DoF products aim to reduce the entry barrier for customers adopting dexterous hand technology.
For industrial scenarios, the company offers the five-finger, six-DoF L6 at RMB 19,900 (USD 2,786). The L6 uses a drive design that combines a brushless motor and a ball screw to reduce mechanical friction, increasing drive efficiency to about 90%. Ball screw lifetime tests show it surpasses one million cycles.
At the lower end, the O6, priced at RMB 6,666 (USD 933.2), is a five-finger dexterous hand with six active DoF and is currently the lightest dexterous hand in the world at 370 grams. Linkerbot also offers a subsidized O6 lite version at RMB 3,999 (USD 559.9).
In-house development has become central to Linkerbot’s ability to maintain strong value-for-money performance. Beginning with components as fundamental as gear reducers, the company develops everything internally. Linkerbot has established a supply chain that covers materials, motors, and electric cylinders, which gives it a cost advantage over outsourcing or custom manufacturing.
Even so, scaling production and fulfilling orders remain global challenges for a young category like dexterous hands.
Zhou told 36Kr that typical complications include stabilizing newly built production lines, maturing manufacturing processes, and managing concentrated order surges that lead to long queues. Linkerbot is currently understood to be the only company that has mass produced high-DoF dexterous hands at a scale of 1,000 units. Based on publicly disclosed data, its shipments account for more than 80% of the global market for high-DoF dexterous hands.
To keep capacity aligned with demand, Linkerbot built its own production lines and retained control of its manufacturing operations. The company now operates four subsidiary factories and works closely with suppliers to ensure delivery.
Monthly production capacity has surpassed 4,000 units, and Linkerbot aims to deliver between 50,000–100,000 dexterous hands in the next year.
Hardware alone, however, is insufficient for long-term competitiveness. Zhou noted that once large technology companies devote significant resources to the category, they will likely be able to outcompete smaller firms on cost. This means Linkerbot must expand across the software stack and ecosystem layer.
For data collection, the company has launched Open TeleDex, a modular robot teleoperation system that is said to support any external device, any robotic arm, and any dexterous hand. The system is designed to reduce integration costs for customers building data collection setups.
Linkerbot has also developed LinkerSkillNet, a multimodal data collection system aimed at building what it describes as the world’s largest database of human hand skills.
Zhou explained the reasoning. “The value of the human hand lies in its unmatched generalization, for it can manipulate thousands of objects, perform thousands of operations, and use thousands of tools,” he said. “We want to build a massive ‘skill store’ where users can draw from a wide range of hand skills. From a business model perspective, when tech giants eventually enter the market, if we already have data for thousands of mature skills accumulated, they will be more inclined to partner with us rather than simply replicate the hardware.”
For training, Linkerbot uses real-world reinforcement learning. The company incorporates end-effector force sensing, giving robotic fingers the ability to perceive contact forces. It has also expanded dexterous operation training to non-force-controlled robotic arms and increased automation across the training pipeline.
As a result, Linkerbot reports a near 100% success rate in demonstration tasks such as screwdriving and USB port insertion.
To strengthen industry coordination, Linkerbot has partnered with CK Capital to launch an initiative that supports innovation across the value chain, including component suppliers, model algorithm developers, and vertical application solution providers, offering both capital and industrial resources.
The company has also introduced a program based on the L20 to deploy robots equipped with Linkerbot hands on production lines, enabling robots to learn continuously from real-world tasks and improve autonomously.
On the team front, Linkerbot’s leadership includes Zhou, who joined Lenovo’s research institute in 2007 and brings 15 years of combined internet and robotics experience.
The head of algorithms and model training, Cao Gang, is a researcher who has led major national artificial intelligence projects. He is also part of Beijing’s municipal talent program and oversees a cross-institution team of 100 researchers focused on large model development. His team has incubated and invested in companies including Z.ai (formerly Zhipu AI), Galbot, ModelBest, DeepLang, LibLibAI, and Xiaoyu.
KrASIA Connection features translated and adapted content that was originally published by 36Kr. This article was written by Fu Chong for 36Kr.