AI and robotics continue to push the boundaries of technology, and a recent collaboration between NVIDIA and 1X Technologies is a prime example. At 1X Technologies’ San Francisco headquarters, the company’s humanoid robot, NEO Gamma, presented NVIDIA CEO Jensen Huang with a custom black leather jacket.
As part of this collaboration, 1X Technologies is providing NVIDIA with access to a special tool that grants real-world data insights from offices and homes. Additionally, the firm has developed software enabling NVIDIA’s powerful GPUs to run predictions at a frequency of five times per second, whether embedded within the robot’s head or connected externally.
NEO Gamma, backed by OpenAI, is 1X’s latest humanoid robot designed to assist with household tasks, marking a significant step toward integrating AI-powered robotics into everyday life.
Following the GTC 2025 keynote on March 18, during which Jensen Huang showcased the NEO Gamma robot’s ability to load dishes, the collaboration took a fashionable turn. 1X presented Huang with a custom leather jacket, designed by California-based ERL. Featuring a studded NVIDIA logo on the chest and an embroidered version on the back, the jacket blended Huang’s signature look with a modern tech aesthetic.
Developing AI for humanoid robots like NEO Gamma comes with challenges in safety, dexterity, and visual processing. To accelerate progress, 1X has created a dataset API that allows NVIDIA to access real-world data, aiding in AI model training.
Additionally, an inference software development kit (SDK) enables continuous 5Hz model predictions using NVIDIA GPUs, providing enhanced real-time responses for robotic tasks. A crucial step in AI development involves verifying new learning models, ensuring they synchronize across data collection, training, and runtime inference.
Working alongside NVIDIA’s GEAR team, 1X utilized the GR00T N1 model to teach NEO Gamma complex tasks, such as autonomously grasping a cup, transferring it between hands, and loading it into a dishwasher. These tests were conducted in a 1X employee’s home, where researchers exchanged insights on control frequencies, imitation learning, and action spaces. Such advancements suggest a near future where humanoid robots can seamlessly handle everyday chores, making life easier for users.
NEO Gamma boasts several upgrades that enhance its safety and usability. Soft covers provide safer interactions, while an advanced visual manipulation model allows it to handle diverse objects in unfamiliar environments. Using neural networks, the robot predicts actions based on raw sensor data, improving adaptability.
The introduction of a Companion feature set further enhances human-robot interactions, incorporating a custom language model for natural conversations and responsive body language. A conversational voice interface powered by 1X’s proprietary large language model (LLM) enables more intuitive communication, bridging the gap between AI and practical applications.
With significant hardware improvements, NEO Gamma is now quieter, more reliable, and better suited for real-world environments. The latest iteration is reportedly 10 times more reliable and operates at a noise level 10 dB lower, comparable to the sound of a refrigerator.
