At CES 2026, AGIBOT makes a decisive move onto the global stage, marking its official U.S. market debut while presenting one of the most complete and operationally mature humanoid robot portfolios at the show. At a time when much of the humanoid robotics sector is still centered on prototypes and tightly controlled pilots, AGIBOT arrives with a different profile: more than 5,000 robots already manufactured and deployed in real-world environments, spanning industrial, commercial, and public-facing use cases.
AGIBOT XR2 dancing at CES Pepcom Digital Experience
Founded in 2023, AGIBOT was built around a clear vision: humanoid robots must evolve beyond single-purpose automation and become general-purpose embodied systems capable of learning and adapting across environments. This philosophy is reflected in the company’s motto, “One Robotic Body, Three Intelligences.” The AGIBOT architecture integrates locomotion, interaction, and task intelligence into a unified platform. Rather than optimizing for isolated demonstrations, AGIBOT’s approach prioritizes system-level consistency, enabling robots to move naturally, communicate with people, and perform useful work under real-world constraints.
At CES, AGIBOT is showcasing its full lineup of embodied robots. The A2 Series represents the company’s full-sized humanoids, designed for guided tours, reception, showroom presentations, and public engagement, where autonomous navigation and natural interaction are critical. The X2 Series, roughly half the height of an adult human, targets education, research, and entertainment, with expressive movement and humanlike walking that make it well-suited for learning environments and demonstrations. We saw the X2 in action at Pepcom ahead of CES, where it performed a brief dance routine to showcase its balance, fluidity, and expressive motion.
AGIBOT A2
The G2 Series addresses industrial use cases with humanoid robots engineered for factory floors and production lines. In a briefing ahead of CES, AGIBOT emphasized that these robots are designed to operate beyond the rigid assumptions of traditional industrial automation. Instead of relying exclusively on fixed programming and predefined object positions, G2 robots use reinforcement learning combined with vision-based perception, allowing them to recognize and manipulate objects even when layouts change. This adaptability is aimed at real production environments, where variability is the norm rather than the exception.
AGIBOT G2
The portfolio is further expanded by the D1 Series quadruped robots, intended for inspection, patrol, and operations in complex or hazardous environments, including security monitoring and infrastructure checks. Complementing the humanoid platforms is OmniHand, AGIBOT’s dexterous robotic hand system. Equipped with on-hand sensing and vision, OmniHand enables fine-grained force control, allowing robots to distinguish between fragile objects and rigid components, and to adjust grip strength accordingly, a capability that is critical for both industrial and service-oriented tasks.
AGIBOT D1 Pro
A key differentiator highlighted during the interview is AGIBOT’s platform-level scalability. All robots in the lineup share a common foundation, meaning skills learned by one robot can be transferred across others. This design reduces training time when deploying fleets across multiple locations and supports customization without fragmenting the underlying system. AGIBOT views this as essential for scaling deployments across logistics, manufacturing, hospitality, tourism, and public-service scenarios, where consistency and rapid rollout matter as much as raw capability.
Alongside its hardware portfolio, AGIBOT is also launching Genie Sim 3.0 at CES 2026, marking the public debut of its third-generation simulation and evaluation platform for embodied intelligence. Built on NVIDIA Isaac Sim, Genie Sim 3.0 is designed to connect real-world robot operations with high-fidelity simulation, providing a unified workflow spanning scene generation, data collection, physics-based simulation, and automated evaluation.
AGIBOT positions Genie Sim 3.0 as a response to one of the major bottlenecks in humanoid robotics: the cost and complexity of training robots exclusively in physical environments. The platform introduces an open-source simulation pipeline supported by more than 10,000 hours of synthetic data, covering 200+ tasks across 100,000+ simulation scenarios. Crucially, the system supports 1:1 digital twins of real industrial environments, enabling robots to be trained and evaluated virtually before deployment, thereby shortening validation cycles and reducing reliance on physical hardware.
“Bringing our full robotics portfolio to CES marks a defining moment for AGIBOT,” said Dr. Yao Maoqing, Partner at AGIBOT, Senior Vice President, and President of the Embodied Business Unit. “It demonstrates our transition from advanced R&D to real-world deployment at scale, while laying the groundwork for an open ecosystem around embodied intelligence.”
AGIBOT will host live, coordinated demonstrations throughout CES 2026 at the Las Vegas Convention Center, North Hall, Booth 10715, offering attendees a rare opportunity to see humanoid robots operating beyond controlled lab conditions and into practical, scalable deployment scenarios.
Filed in . Read more about AI (Artificial Intelligence), CES, CES 2026, Humanoid Robot, NVIDIA and Robotics.