1X has unveiled a next-generation robotic hand for its NEO home humanoid, asserting that the component finally resolves the persistent dexterity gap in robotics. The new manipulator, designed to replicate human biomechanics, features 25 degrees of freedom against the human hand’s 27, enabling complex tasks such as pouring tea, sorting grapes by color, plugging in USB-C chargers, and executing sign language. According to Dar Sleeper, 1X’s head of product, the system represents the closest approximation to human-level manual agility achieved to date. The hand’s architecture departs from traditional direct-drive designs by situating motors within the forearm and utilizing tendon-like cables to actuate the digits. This configuration reduces weight, enhances control precision, and incorporates a compliance mechanism that allows the fingers to yield upon impact. The prototype demonstrates exceptional durability, remaining functional after repeated hammer strikes, and is fully waterproof, permitting the robot to wash its own hands. Despite their refined size, the manipulators retain sufficient torque to carry grocery bags and luggage, overcoming early iterations that were too bulky for practical domestic use. Manufacturing scalability is a central focus for 1X. The company has transitioned to a modular assembly process comparable to building blocks, allowing technicians in its Hayward, California, facility to construct the hand at high volume. Sleeper confirmed that 1X can already produce thousands of units annually, with an initial plant capable of 10,000 NEO units per year. A second facility in San Carlos is being scaled to eventually support 100,000 to 250,000 units annually. This expansion follows a $100 million funding round in 2024, which included backing from OpenAI and Samsung, and has already attracted 10,000 preorders. The engineering strategy is rooted in a deliberate design philosophy articulated by CEO Bernt Børnich. Børnich argues that the human hand is the primary interface through which intelligence probes and interprets the physical world. By matching human kinematics, 1X intends to streamline machine learning pipelines. A kinematically similar hand can directly learn from consumer video datasets, enabling the training of large-scale world models that understand domestic environments and everyday tasks. Sleeper defended this anthropomorphic approach against market skepticism, noting that consumer environments are engineered around human proportions and movement patterns, making human-like form factors the most efficient path to functional autonomy. The NEO humanoid is priced at $20,000 upfront or $500 monthly, positioning it in a growing sector of domestic automation. Competitors such as Weave Robotics and Sunday Robotics are advancing wheeled, gripper-based systems for laundry and dishwashing, while Tesla continues development of its Optimus platform. Nevertheless, 1X has identified early commercial partners and aims to begin deployments in 2026. The successful integration of a high-dexterity manipulator could accelerate the transition of humanoid robotics from laboratory prototypes to reliable household assistants, cementing the hand as the critical bottleneck for the industry’s next wave of commercialization.