Unleashing Physical AI: The NVIDIA Jetson Orin Nano 2 for Advanced Drones and Robotics
NVIDIA has just unveiled the Jetson Orin Nano 2, an innovative edge robotics computer designed to elevate the capabilities of drones, robots, and vision systems. This powerful board offers a fresh entry point for developers keen on running generative AI models directly on machines, shifting away from reliance on data centers. With advancements in AI performance and energy efficiency, this launch promises to reshape how we interact with technology.
The Breakthrough in AI Performance
NVIDIA’s motivation for releasing the Jetson Orin Nano 2 stems from significant developments in the efficacy of small and medium AI models. These models now mirror the accuracy levels that only were achievable by larger frontier models just a year ago.
Deepu Talla, NVIDIA’s VP of Robotics and Edge AI, emphasized the impact of this breakthrough: “The Jetson Orin Nano 2 brings unprecedented performance and energy efficiency to millions of developers, enabling real-time reasoning right at the edge.”
This leap forward means compact edge hardware can efficiently interpret language and images, responding almost instantly. From delivery drones to vision AI systems, the capability to operate without consuming excessive power is transformative.
Jetson Orin Nano 2: Specifications and Efficiency
The Jetson Orin Nano 2 is no slouch when it comes to specifications. Here’s what makes it stand out:
- 78 trillion operations per second of AI compute power
- 8GB of memory
- An eight-core Arm CPU
This new board provides excellent AI and video-processing performance while maintaining affordability and low power consumption. Notably, it offers twice the inference performance of the previous Jetson Orin Nano Super, thanks to enhanced Tensor Cores and increased memory bandwidth—all while fitting into the same compact design.
Running in a 15-watt mode, the Jetson Orin Nano 2 operates 40% more efficiently than the Orin Nano Super without compromising performance.
Built to seamlessly integrate with NVIDIA’s open software stack, the board supports large language models and vision language models optimized for edge inference. Developers can deploy powerful models, such as Cosmos and Nemotron, as well as others like Gemma 4 and Qwen 3.
Early Adopters and Real-World Applications
Several prominent companies, including Cognex, Doosan Bobcat, and Matic, have already started exploring the Jetson Orin Nano 2. NVIDIA boasts a community of over three million developers leveraging its robotics stack, with expectations that partners will widely integrate edge AI into a range of applications—covering everything from home robots to inspection drones.
Wing, a subsidiary of Alphabet specializing in drone delivery, is particularly excited about the Jetson Orin Nano 2. Currently employing the Jetson Orin Nano Super for its fleet, Wing aims to utilize the new board to enhance real-time AI perception, ultimately speeding up and securing local deliveries.
Dinuka Abeywardena, Head of Perception at Wing, shares this sentiment: “Drone delivery depends on AI that understands the real world swiftly and reliably. Exploring Jetson Orin Nano 2 will help us create more agile and energy-efficient drones for our delivery systems.”
Matic Robots, known for its cutting-edge consumer robotics, also plans to harness the capabilities of Jetson Orin Nano 2 to enhance its home-cleaning robots. With this board, Matic can introduce conversational AI, gesture detection, and sophisticated spatial mapping, all while ensuring autonomous cleaning functionality.
Co-founder and CEO Navneet Dalal explained: “Home robots need to interact effectively, understand their environment, and navigate dynamically. Thanks to Jetson Orin Nano 2, we can integrate state-of-the-art AI models right at the edge.”
Collaborations and Future Prospects
NVIDIA is collaborating with a wide array of hardware partners to expand the ecosystem surrounding the Jetson Orin Nano 2. Key players like AAEON, ADLINK, and Advantech are developing carrier boards and hardware systems tailored for this innovative board. Additionally, companies like Antmicro and Aptiv are working on customized AI software and reference designs, facilitating faster market entry for their customers.
Talla notes, “Frontier intelligence has now reached the edge; models that previously resided in data centers can now be operated in real-time on entry-level Jetson systems. The Jetson Orin Nano 2 opens doors to applications that were previously unimaginable.”
Discover the future of physical AI by attending the Physical AI Expo in Amsterdam, London, and North America.
With the Jetson Orin Nano 2, the next chapter in AI-driven innovation is just beginning. This groundbreaking technology is set to transform industries, making autonomous solutions more accessible and efficient than ever. Are you ready to be a part of this exciting evolution? Let’s embrace the future together!
