Jetson Orin Nano 2: NVIDIA's entry robotics board promises double inference
· TexaHub · Product & Platform Reviews, Robotics, Semiconductors
NVIDIA's Jetson Orin Nano 2 claims twice the inference of the Orin Nano Super at lower power, but pricing is undisclosed and shipments wait until 2027.
What NVIDIA announced
NVIDIA announced the Jetson Orin Nano 2 on 25 August, positioning it as an entry-level edge AI computer for robots, drones and vision systems [1]. The module offers 78 trillion operations per second (TOPS) of AI compute, 8GB of memory and an eight-core Arm CPU, in the same compact form factor as its predecessor [1].
NVIDIA claims double the inference performance of the Jetson Orin Nano Super and 40% lower power consumption at equivalent performance in 15-watt mode [1]. The Robot Report reported that the gains come from improved Tensor Cores and higher memory bandwidth, and that the board uses the same GPU architecture as NVIDIA's data centre products [2]. For comparison, NVIDIA rated the Orin Nano Super, launched in December 2024 as a $249 developer kit, at 67 INT8 TOPS with 102GB/s of memory bandwidth [3].
Early evaluators named by NVIDIA include Wing, Matic Robots, Cognex and Doosan Bobcat [1]. "Wing is exploring Jetson Orin Nano 2 to give us a path to more responsive, energy-efficient drones," said Dinuka Abeywardena, Wing's head of perception [1]. Matic is integrating it into home cleaning robots with conversational AI and gesture detection, according to The Robot Report [2].
Our assessment
Three gaps limit how far this can be judged today. First, NVIDIA has not disclosed a price [1], and price was the headline of the Orin Nano Super launch [3]; for hobbyist, education and low-volume robotics buyers, cost per TOPS matters as much as peak throughput. Second, the module and developer kit are not due until the first half of 2027 [1], so teams designing products now must plan around a part they cannot yet test. Third, the doubling claim is NVIDIA's own and comes without published benchmark methodology in the announcement material we reviewed.
The software story is stronger. NVIDIA says the board runs language and vision-language models optimised for memory-efficient edge inference, including its own open Cosmos and Nemotron models as well as Gemma 4 and Qwen 3 [1]. The Robot Report noted that more than 3 million developers build on NVIDIA's robotics stack [2], and continuity with that stack lowers switching costs for existing Orin users.
Commercially, edge devices remain a small part of NVIDIA's business. In its second-quarter fiscal 2027 results, released a day after the announcement, Edge Computing revenue was $7.2 billion, up 27% year over year, against total revenue of $96.2 billion [4]. The Orin Nano 2 is best read as an ecosystem play: keeping the next wave of low-cost robots on NVIDIA's platform, while the higher-end Jetson Thor underpins the Isaac GR00T reference humanoid design NVIDIA unveiled alongside its results [4].
Sources
[1] NVIDIA Announces Jetson Orin Nano 2 Robotics Computer to Redefine Entry-Level Edge AI — NVIDIA, 25 Aug 2026
[2] Jetson Orin Nano 2 doubles inference performance for robotics at the edge, says NVIDIA — The Robot Report, 25 Aug 2026
[3] NVIDIA Unveils Its Most Affordable Generative AI Supercomputer — NVIDIA, 17 Dec 2024
[4] NVIDIA Announces Financial Results for Second Quarter Fiscal 2027 — NVIDIA, 26 Aug 2026