The humanoid robotics sector has reached another inflection point, and the latest evidence comes from the collaboration between a Chinese robotics developer and a US semiconductor giant. Galbot, a company focused on general-purpose humanoid machines, has announced that its G1 Premium robot is now being built around Nvidia’s Jetson Thor platform. This is not a peripheral upgrade or a marketing tie-in; it is a fundamental change to the computational core of the machine, and it signals how seriously both companies are taking the transition from laboratory demonstrations to commercial deployment.
The integration was made public in connection with the World Robotics Conference, held in Beijing in August 2025. It was at that event that Galbot showcased the G1 Premium running on Jetson Thor, giving attendees a live look at what the new hardware enables. According to information released by Nvidia, the Jetson Thor module delivers a substantial leap over its predecessor, the Jetson Orion. The new platform offers more than seven times the AI computing capacity of the earlier generation, and it does so while consuming significantly less power — over three times the energy efficiency, by Nvidia’s own figures. For a robot that needs to operate in real-world environments, where battery life and thermal management are constant constraints, that combination of raw compute and efficiency is not a luxury; it is a necessity.
Galbot’s positioning of the G1 Premium is clear. The company describes the robot as a general-purpose humanoid, meaning it is not being built for a single, narrow task. Instead, it is intended to operate across multiple sectors, with retail, healthcare, and logistics named as the primary target industries. The company’s stated goal with the Jetson Thor integration is to boost performance and enable a greater degree of autonomy. In practical terms, that means the robot should be able to handle more complex planning and motion tasks, and do so with a level of precision that was previously out of reach.
What is particularly notable about this announcement is the timing. The World Robotics Conference demonstration took place in August 2025, and the company has been positioning itself as an early adopter of the Jetson Thor platform. That is a meaningful distinction. Many robotics firms talk about future plans or roadmap items; Galbot has put the hardware in a product and shown it to the public. The company has also been active in securing the resources needed to scale. In June 2025, Galbot announced the completion of a new financing round totaling 1.1 billion yuan, a substantial sum that underscores investor confidence in the company’s direction. While the exact allocation of those funds has not been disclosed, it is reasonable to assume that continued hardware integration and software development are priorities.
The broader context here is also worth noting. Nvidia has made no secret of its ambitions in the robotics space. The company has positioned itself as a major supplier of tools and platforms for robot development, and its Jetson modules are designed to be embedded directly into AI robots. These modules combine Nvidia’s Blackwell GPUs, the Isaac development platform, and additional capabilities such as sensor signal processing. The result is a system that can run multiple AI workflows simultaneously, enabling real-time interactions and on-robot inference. That last point is critical: running inference directly on the robot, rather than relying on a cloud connection, reduces latency and allows the machine to react to its environment in real time. For a humanoid robot navigating a busy retail floor or a hospital corridor, that capability is essential.
Galbot is not alone in receiving early access to the latest Jetson hardware. According to available information, several prominent Chinese robotics companies were among the first to receive the newest Jetson modules, including UBTech, Unitree, EngineAI, and AgiBot, in addition to Galbot itself. International players such as Figure AI, Google DeepMind, Mentee Robotics, Meta, and Skild AI have also adopted the Jetson Thor platform. This is a broad ecosystem, and Galbot’s inclusion in that group is a signal that the company is being taken seriously by one of the most important hardware suppliers in the industry.
Product and availability details
The G1 Premium is the product at the center of this announcement, and it is worth examining what is known about it. Galbot describes the robot as a general-purpose humanoid, which means it is designed to perform a wide range of tasks rather than being specialized for a single function. This is a deliberate design choice. In sectors like retail, healthcare, and logistics, the tasks that need to be done are varied and often unpredictable. A robot that can only do one thing well is of limited use in such environments; a robot that can adapt to different situations is far more valuable.
The integration of the Jetson Thor platform is intended to make that adaptability possible. According to Nvidia, the Jetson Thor is powered by a Blackwell GPU, which enables robots to run multiple AI workflows at the same time. This is important because a humanoid robot is not just running one model; it is simultaneously processing visual data, planning movements, maintaining balance, and responding to external stimuli. Each of those tasks requires computational resources, and doing them all in real time requires a platform that can handle the load. The Jetson Thor is designed for exactly this kind of workload, and Nvidia has stated that it enables real-time on-robot inference, which is a breakthrough for high-performance physical AI applications.
Galbot has said that, by leveraging this upgrade, its robots can now perform complex planning and motion tasks with new levels of precision and efficiency. The company has also reported gains in speed, responsiveness, and real-time reasoning. These are not vague claims; they are specific performance characteristics that matter in real-world deployments. A robot that can reason about its environment in real time can navigate around obstacles, avoid collisions, and respond to unexpected events. A robot that can plan complex motions can pick up objects, open doors, and manipulate tools. The combination of these capabilities is what makes a humanoid robot useful in a commercial setting.
What has not been disclosed is equally important. Galbot has not published specific technical specifications for the G1 Premium beyond its use of the Jetson Thor platform. There is no public information about the robot’s payload capacity, its maximum walking speed, or its battery life. The company has not announced pricing, and there is no confirmed availability date for commercial deployment. The demonstration at the World Robotics Conference was just that — a demonstration — and it is not clear when the G1 Premium will move from the show floor to actual customer sites. These details may be forthcoming, but as of now, they have not been made public.
It is also worth noting that the G1 Premium is not the only humanoid robot in Galbot’s portfolio, nor is it the company’s only area of activity. Galbot has been involved in building large-scale dexterous hand datasets for humanoid robots, using Nvidia’s Isaac Sim simulation platform. This work, which was highlighted in a technical blog post by Nvidia in November 2024, suggests that the company is investing heavily in the software and data side of robotics, not just the hardware. A robot is only as good as the models it runs, and those models are only as good as the data they are trained on. By building datasets for dexterous manipulation, Galbot is addressing one of the hardest problems in humanoid robotics: getting robots to use their hands effectively.
The company’s financing situation is also relevant to any discussion of availability. In June 2025, Galbot announced the completion of a new financing round totaling 1.1 billion yuan. This is a significant amount of capital, and it gives the company the resources to continue development, expand its team, and potentially scale up manufacturing. However, the specific use of these funds has not been disclosed, and it would be speculative to assume that they are earmarked for a particular product or market. What can be said is that Galbot is financially positioned to pursue its stated goals.
What it means for buyers
For organizations considering the adoption of humanoid robots, the Galbot G1 Premium and its Jetson Thor integration represent a data point worth examining. The combination of increased compute capacity and improved energy efficiency addresses two of the most common objections to humanoid robots in commercial settings: they are not smart enough to handle complex tasks, and they do not have enough battery life to be practical. The Jetson Thor platform, at least on paper, makes progress on both fronts.
The sevenfold increase in AI computing capacity, as claimed by Nvidia, is a substantial jump. It means the robot can run larger and more sophisticated models, which in turn means it can handle more complex tasks. The threefold improvement in energy efficiency is equally important, because it directly impacts how long the robot can operate on a single charge. For a retail store that wants a robot to work a full shift, or a hospital that needs a robot to make rounds throughout the day, battery life is a critical factor. If the Jetson Thor delivers on its efficiency claims, the G1 Premium could operate for significantly longer periods between charges.
However, buyers should be cautious about drawing too many conclusions from the available information. The performance claims come from Galbot and Nvidia, and they have not been independently verified. The demonstration at the World Robotics Conference was a showcase, not a rigorous benchmark. No third-party testing results have been published, and no customer case studies are available. The robot has not been deployed in a commercial setting, at least as far as public information indicates, so there is no track record of real-world performance.
There are also unanswered questions about the total cost of ownership. The price of the G1 Premium has not been announced, and neither has the cost of the Jetson Thor platform itself. Maintenance costs, spare parts availability, and service agreements are all undisclosed. Buyers who are evaluating the G1 Premium for potential deployment will need to factor in these unknowns. It is possible that the robot will be competitively priced and that the total cost of ownership will be attractive, but that has not been demonstrated.
Another consideration is the broader ecosystem. Nvidia’s Jetson platform is supported by the Isaac development environment, which provides tools for simulation, training, and deployment. This is a mature ecosystem, and it is used by many companies in the robotics space. For buyers, this means that there is a pool of developers and integrators who are familiar with the platform, which could make it easier to find support and customize the robot for specific use cases. It also means that the platform is likely to continue evolving, as Nvidia has shown a commitment to investing in robotics.
The fact that Galbot was an early recipient of the Jetson Thor hardware is also relevant. It suggests that Nvidia sees Galbot as a strategic partner, and it gives Galbot a head start in integrating the platform into its products. Whether that head start translates into a competitive advantage in the market remains to be seen, but it is a positive signal.
For buyers in the retail, healthcare, and logistics sectors, the G1 Premium is worth monitoring. The robot is being designed for the kinds of tasks that are common in those industries, and the Jetson Thor integration addresses some of the key technical challenges that have historically limited humanoid robots. However, the lack of pricing, availability, and performance data means that any purchasing decision would be premature. The prudent approach is to track the product’s development, look for independent evaluations, and wait for more concrete information before making any commitments.
It is also worth noting that the humanoid robotics market is becoming increasingly crowded. Galbot is one of several companies, including UBTech, Unitree, EngineAI, and AgiBot, that are working with Nvidia’s latest hardware. Each of these companies is pursuing a slightly different approach, and buyers will have options. The G1 Premium is one option among many, and its ultimate success will depend on factors that have not yet been revealed, such as reliability, durability, and the quality of the software stack.
In summary, the announcement of the Galbot G1 Premium with Nvidia Jetson Thor integration is a significant development in the humanoid robotics space. The hardware upgrade promises substantial improvements in performance and efficiency, and the company’s early adoption of the platform positions it well for future development. However, many details remain undisclosed, and buyers should approach any potential purchase with caution. The robot has been demonstrated, but it has not been proven in commercial settings. Until more information is available, the G1 Premium is best viewed as a promising product in development, rather than a proven solution ready for deployment.
Sources
https://www.automationworld.com/factory/robotics/news/55325084/galbots-humanoid-robot-integrates-nvidia-jetson-thor-with-potential-for-manufacturing-use
Published by Vigla Media OÜ (Estonia).