---
format: "aidr-story-markdown/v1"
id: "a8072c454a9382cef3d3b0ff964b1d26feb78133bd23145cb00d5811d674ae97"
canonical_url: "https://aidr.today/a8072c45?lang=en"
title: "Nvidia Jetson Orin Nano 2 Doubles Inference Performance to Bring LLMs to Entry Level Robotics"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-08-25T16:48:16.000Z"
category: "Chips"
topics: ["nvidia","chips","robotics","inference"]
source_urls: ["https://huggingnews.com/ai/nvidia-jetson-orin-nano-2-doubles-inference-performance-to-bring-llms-to-a414a252","https://x.com/nvidianewsroom/status/2092271330196394453","https://x.com/wallstengine/status/2092267127583769087","https://x.com/StockSavvyShay/status/2092270299391951182","https://x.com/Techmeme/status/2092274744251027555"]
summary: "The new edge AI computer delivers 78 TOPS of AI compute with 8GB of memory and an 8-core Arm CPU for robots, delivery drones, and vision systems. It provides twice the inference performance of the Orin Nano Super in the same form factor and uses 40% less power to maintain the same performance in 15W mode. The hardware allows for local execution of vision-language models and LLMs, including Gemma 4 and Nemotron, without relying on cloud connectivity. Cognex, Doosan Bobcat, and Matic are among the first adopters, while Alphabet's Wing will evaluate the module for real-time reasoning in delivery drones. Nvidia will release the module and developer kit in the first half of 2027."
---

# Nvidia Jetson Orin Nano 2 Doubles Inference Performance to Bring LLMs to Entry Level Robotics

> [Open the canonical story](<https://aidr.today/a8072c45?lang=en>)

**Published:** 2026-08-25T16:48:16.000Z
**Category:** Chips
**Topics:** nvidia, chips, robotics, inference

## Summary

The new edge AI computer delivers 78 TOPS of AI compute with 8GB of memory and an 8\-core Arm CPU for robots, delivery drones, and vision systems\. It provides twice the inference performance of the Orin Nano Super in the same form factor and uses 40% less power to maintain the same performance in 15W mode\. The hardware allows for local execution of vision\-language models and LLMs, including Gemma 4 and Nemotron, without relying on cloud connectivity\. Cognex, Doosan Bobcat, and Matic are among the first adopters, while Alphabet's Wing will evaluate the module for real\-time reasoning in delivery drones\. Nvidia will release the module and developer kit in the first half of 2027\.

## Sources

- [Story source](<https://huggingnews.com/ai/nvidia-jetson-orin-nano-2-doubles-inference-performance-to-bring-llms-to-a414a252>)
- [Story source](<https://x.com/nvidianewsroom/status/2092271330196394453>)
- [Supporting source](<https://x.com/wallstengine/status/2092267127583769087>)
- [Supporting source](<https://x.com/StockSavvyShay/status/2092270299391951182>)
- [Supporting source](<https://x.com/Techmeme/status/2092274744251027555>)