Linkerbot Develops Human-Like Robotic Hands for Real-World AI Applications

In a significant step toward physical artificial intelligence, Chinese robotics company Linkerbot is developing dexterous robotic hands that can operate with human-like precision. While large language models are considered the “brains” of AI, Linkerbot’s robotic hands aim to become its “body,” bridging the gap between digital intelligence and the physical world.

“NVIDIA CEO Jensen Huang once said the next wave of AI would be physical AI,” said the Linkerbot team. “For us, the real challenge is not just computing power, but how fast hardware can catch up.” Over recent years, the company has focused on high-DOF (Degree of Freedom) hands, integrating hardware, algorithms, and data to build a full-stack robotic system.

Moving beyond hardware, Linkerbot recently introduced a PEEK-based (Polyether Ether Ketone) reducer, making robots lighter and more energy-efficient, while exploring multiple technical designs including tendon-driven, linkage-based, and direct-drive systems. On the software side, the company has launched the Linker Creation Model, along with Linker OS and a reinforcement learning framework, creating an end-to-end system from data collection to model training.

For practical applications, Linkerbot is targeting precision manufacturing, healthcare, and home services. Its robotic hands are already being used in automotive and electronics assembly, while projects in agriculture are developing robots for breeding and high-tech farming tasks. In eldercare and rehabilitation, these hands aim to assist in daily activities such as folding clothes and serving coffee, reducing physical strain on caregivers.

“Our goal is to relieve people from dangerous, exhausting, and repetitive tasks so they can focus on creative work,” the team said. While lab demonstrations are one thing, real-world deployment presents challenges in reliability and safety, making durability and cost efficiency crucial. Linkerbot’s Linker Hand series has undergone rigorous testing, including drop tests, and has contributed to setting national robotics standards.

Globally, Linkerbot’s products are used in research labs at universities such as Stanford and Cambridge. The company has also launched Open TeleDex, a modular teleoperation system allowing developers to control robotic hands with VR devices or exoskeletons, supporting open-source experimentation and collaboration.

Linkerbot serves two main customer groups: those pushing the limits of physical dexterity for research, and those seeking stable, industrial-scale robotic systems. The company currently produces over 1,000 dexterous hands per month and aims to scale to 50,000–100,000 units annually by 2026. In December 2025, Linkerbot closed its Series A++ funding round led by Sequoia China and Creation Partners Capital.

“Delivering 100,000 robotic hands means taking over 100,000 dangerous and repetitive jobs,” the company said, highlighting its long-term commitment to using technology for the greater good. As embodied intelligence develops, Linkerbot believes industrial-scale, human-like hands may shape the future of physical AI.