NVIDIA has launched the Jetson Orin Nano 2, a new entry-level robotics computing platform designed for edge AI applications such as robotics and drones.

Article author and source: Sina Tech

NVIDIA is extending the capabilities of artificial intelligence (AI) further from data centers to real-world devices such as robots.

On Tuesday, the 25th, Eastern Time, NVIDIA announced the launch of the Jetson Orin Nano 2, its next-generation entry-level robotics computing platform, targeting edge AI applications such as robotics and drones that are sensitive to cost, power consumption, and real-time computing capabilities.

NVIDIA states that, while maintaining the same compact form factor, the Jetson Orin Nano 2 delivers twice the inference performance of its predecessor, the Jetson Orin Nano Super, and achieves the same performance at 40% lower power consumption in a 15-watt power mode.

From a hardware perspective, the Jetson Orin Nano 2 features an 8-core processor and 8GB of memory, delivering an AI computing power of 78 trillion operations per second. NVIDIA states that the new platform enhances AI inference capabilities through upgrades such as improved tensor cores and increased memory bandwidth, while maintaining compatibility with NVIDIA’s software ecosystem designed for robotics and edge AI. It can run memory-optimized large language models and vision-language models, including Cosmos, Nemotron, Gemma 4, and Qwen 3.

The significance of this new product launch goes beyond simply introducing a more powerful small AI computer. As AI models continue to shrink and inference efficiency improves, tasks previously reliant on cloud processing—such as visual understanding, language interaction, and real-time decision-making—are increasingly migrating to end devices like robots and drones. NVIDIA aims to further lower the barrier to entry for physical AI in entry-level devices with the Jetson Orin Nano 2.

Small AI models are accelerating deployment, driving increased demand for edge intelligence in robotics.

The launch of the Jetson Orin Nano 2 coincides with the artificial intelligence industry's shift from the cloud to the edge.

Unlike data center AI, robots and drones face stricter constraints on their computing platforms: limited device size typically prevents the use of high-power processors, while requiring continuous visual perception, environmental understanding, and decision-making during movement. Therefore, the amount of AI computing power delivered per unit of power consumption is often more important than merely pursuing peak computational performance.

NVIDIA states that the Jetson Orin Nano 2 delivers the same performance as its predecessor while consuming 40% less power in 15W mode. For battery-powered robots and drones, this means manufacturers can reduce energy consumption under the same performance requirements, or redirect the saved power toward extended battery life, improved thermal management, or enhanced sensor configurations.

Meanwhile, the capabilities of small AI models are rapidly improving, reshaping the boundaries of edge AI applications.

Tara, head of NVIDIA’s robotics and edge AI business, said that today, the accuracy of some small and medium-sized frontier models has reached the level of last year’s large frontier models, enabling real-time intelligence on edge devices.

In the past, edge AI primarily handled tasks such as object detection and image classification; however, as model capabilities have improved, endpoint devices now possess more advanced abilities in language understanding, visual reasoning, and multimodal interaction.

This also enables the robot to evolve from “being able to perceive the environment” to “being able to understand the environment and take action.”

Moving from visual recognition to multimodal understanding, real-time decision-making becomes critical.

NVIDIA is positioning the Jetson platform as a critical computing foundation for robots to achieve "physical intelligence."

Currently, the information robots need to process is no longer limited to images captured by cameras, but also includes natural language commands, voice, spatial data, and inputs from various sensors. For robots, rapidly fusing this information locally and responding with low latency is a critical condition for achieving autonomous operation.

The Jetson Orin Nano 2 is therefore targeted at multiple typical scenarios, including home robots, vision AI systems, delivery drones, and inspection drones.

For example, Wing, Google’s drone delivery company, has previously integrated the Jetson Orin Nano Super and NVIDIA’s software stack into its delivery drones and plans to evaluate the Jetson Orin Nano 2. NVIDIA states that the improved performance and energy efficiency are expected to help Wing further enhance the real-time AI perception and reasoning capabilities of its drones.

Home robotics company Matic plans to leverage the Jetson Orin Nano 2 to enhance its robots' natural language interaction, gesture recognition, spatial mapping, and semantic understanding, enabling them to perform tasks such as cleaning with greater autonomy.

In addition, NVIDIA demonstrated a case of the Jetson Orin Nano 2 running the Reachy Mini robot. NVIDIA stated that this platform can simultaneously handle language models, speech models, and real-time vision AI tasks, showcasing the capability of an entry-level edge computing platform to run multiple AI models.

For robot manufacturers, the importance of this capability lies in the fact that robots are no longer limited to performing predefined actions based on fixed programs, but can instead understand their environment through visual information and language instructions, and dynamically adjust their actions accordingly.

This is also why NVIDIA has consistently emphasized "physical AI" in recent years: artificial intelligence is gradually moving from software services on screens into the real world, ultimately driving physical devices such as robots, cars, and drones.

Developers and partners expand; NVIDIA continues to strengthen its robotics ecosystem.

In addition to chip performance, the software ecosystem has always been a key competitive advantage of the NVIDIA Jetson platform.

NVIDIA states that over 3 million developers are currently building on its robotics technology stack. With the launch of Jetson Orin Nano 2, NVIDIA is further expanding the software, hardware, and developer ecosystem around robotic AI. Cognex and Matic

On the software side, the Jetson Orin Nano 2 continues to support NVIDIA’s robotics development platform and can run large language and vision-language models optimized for memory efficiency on edge devices, including NVIDIA’s Cosmos, Nemotron, as well as models like Gemma 4 and Qwen 3.

This means developers can directly deploy language understanding, visual understanding, and reasoning capabilities onto robotic endpoints using relatively limited hardware resources, without fully relying on cloud servers.

The hardware ecosystem is also expanding in tandem.

NVIDIA disclosed that partners including AAEON, Advantech, Aetina, Antmicro, Aptiv, Connect Tech, and Seeed Studio are developing carrier boards, hardware systems, AI software, and reference designs around the Jetson Orin Nano 2 to help robotics manufacturers shorten their product development and time-to-market cycles.

NVIDIA also stated that the first companies to adopt or explore the Jetson Orin Nano 2 include Cognex, Doosan Bobcat, Matic, and Wing plans to evaluate this platform.

From a longer-term perspective, NVIDIA is not merely trying to build a standalone robot chip business, but rather a comprehensive ecosystem encompassing AI models, software tools, computing platforms, and robotic endpoints.

This model shares similarities with NVIDIA’s competitive advantage in data center AI: creating synergies between hardware performance and software ecosystems, then further enhancing platform appeal through a large developer community.

The new product is expected to launch in the first half of next year, with commercial contributions yet to be realized.

However, please note that the Jetson Orin Nano 2 is still in the release phase and has not yet become available on the market.

NVIDIA stated that the Jetson Orin Nano 2 module and development kit are expected to be available in the first half of 2027. Therefore, the direct contribution of this new product to NVIDIA's near-term financial performance remains to be seen.

From a product perspective, NVIDIA is gradually refining its robotics and physical AI product ecosystem: on one hand, by delivering higher-performance robotics computing platforms to meet the demands of complex applications; on the other hand, by extending AI inference to cost- and power-sensitive end devices through entry-level platforms like the Jetson Orin Nano 2.

If small language models and vision-language models continue to rapidly improve in capability while inference costs keep falling, the demand for on-device AI computing power in devices such as robots and drones may further increase.

For NVIDIA, this means the market it is competing in is gradually expanding from traditional data center AI computing to the AI "brains" behind real-world devices such as robots and drones. The Jetson Orin Nano 2 is a significant step toward further penetrating this broader end-market.