The world of artificial intelligence is abuzz with the news of 1X's ambitious venture into the realm of humanoid robotics. With the establishment of the 1X World Model Lab, the company is setting its sights on a future where fully autonomous humanoids roam the earth. This move is not just a technological leap but a strategic one, as 1X aims to revolutionize the field by focusing on the core of AI's potential: data. The lab's mission is to accelerate the development of embodied intelligence, a concept that goes beyond traditional AI boundaries. By embracing a holistic approach, 1X is determined to break free from the constraints of narrow tasks and static environments, paving the way for truly general humanoid robots.
The key to this endeavor lies in the 1X World Model Lab's approach to pretraining. Unlike conventional methods, they argue that robotics is not a fine-tuning problem but a fundamental challenge. By pretraining on a diverse range of data, including web-scale media, egocentric human videos, simulation data, and remote-operated robot data, the lab aims to build world models that can generalize and adapt. This comprehensive data strategy is what sets 1X apart and positions them as a leader in the field.
One of the driving forces behind this initiative is Sam Sinha, a renowned expert in scaling large multimodal generative foundation models. His previous work at Luma AI has pushed the boundaries of image and video generation, and his expertise is now being harnessed to create world models tailored for physical intelligence and humanoid robotics. Sinha's vision aligns perfectly with 1X's goals, emphasizing the importance of data curation and scaling in the pursuit of embodied AI.
The 1X World Model Lab's focus on large-scale pretraining and the creation of frontier world models is a bold move. By connecting the entire learning loop, from pretraining to real-world deployment, they aim to accelerate the development of foundation models for humanoid robotics. This comprehensive approach addresses the challenges of data collection, annotation, and scaling, ensuring that the lab's efforts are not just theoretical but have tangible, real-world applications.
In the words of Bernt Børnich, CEO & Founder of 1X, "World models will free humanoid robots from the trap of narrow tasks in static environments by unlocking generalization, adaptability, and self-learning. But to reach full autonomy fastest, you must own the entire stack — down to pre-training your own video foundation models. That’s why we built the 1X World Model Lab."
As 1X embarks on this journey, it invites the world to join them. The lab is actively seeking exceptional talent in data, infrastructure, and learning to contribute to this groundbreaking endeavor. With a focus on innovation and a commitment to pushing the boundaries of AI, 1X is poised to shape the future of embodied intelligence, one step closer to the realization of fully autonomous humanoids.
