Nvidia is hoping interest in artificial intelligence will translate to consumer sales of a relatively low-priced computer optimized for basic AI functionality. Last month, the company upgraded its Jetson line with a $249 “compact AI supercomputer,” the Jetson Orin Nano Super Developer Kit. At half the price of the original, the model aims to attract students, developers, hobbyists, small- and medium-sized businesses, and anyone who is AI curious. “As the AI world is moving from task-specific models into foundation models, it provides an accessible platform to transform ideas into reality,” according to Nvidia.

Available now, the Jetson Orin Nano “delivers as much as a 1.7x leap in generative AI inference performance, a 70 percent increase in performance to 67 INT8 TOPS, and a 50 percent increase in memory bandwidth to 102GB per second compared with its predecessor,” Nvidia says in a blog post that claims the enhanced performance of its latest Jetson “delivers gains for all popular generative AI models and transformer-based computer vision.”

“In a promotional video, Nvidia founder Jensen Huang broke from his usual stagecraft — posing before towering server racks that use Nvidia chips — and instead presented the new palm-sized Jetson device on a tray, as if pulling it fresh from the oven,” The Wall Street Journal writes.

The Jetson Orin Nano “consists of a Jetson Orin Nano 8 gigabyte system-on-a-module (SoM) and a reference carrier board,” reports SiliconANGLE, explaining that the package “provides a platform for prototyping hardware applications and reference architecture for edge AI applications.”

Nvidia says the SoM specs include a Nvidia Ampere architecture GPU with tensor cores and a 6-core Arm CPU, “facilitating multiple concurrent AI application pipelines and high-performance inference.” It also supports up to four cameras, “offering higher resolution and frame rates than previous versions.”

The software needn’t cost a fortune. The platform supports open-source tools including Nvidia’s JetPack SDK and popular machine learning frameworks TensorFlow, PyTorch and OpenCV. For those with a budget, it also runs Nvidia premium packages: Isaac for robotics, Metropolis for vision AI and Holoscan for sensor processing. Also, Omniverse Replicator to reduce time for synthetic data generation and the TAO Toolkit for fine-tuning pretrained AI models from the NGC catalog.

Existing Jetson Orin Nano Developer Kit owners can upgrade the JetPack SDK to get an immediat performance boost, according to Nvidia.

SilconANGLE points out that “the Orin Nano Super is capable of running almost any popular generative AI model and transformer-based computer vision model from Alibaba Group Holding Ltd., Google LLC, Hugging Face, Meta Platforms Inc., Microsoft Corp., OpenAI and more,” making it a viable go-to for many AI-powered functions, including IoT and autonomous vehicles.

Related:
Nvidia Announces $3,000 Personal AI Supercomputer Called Digits, The Verge, 1/24/25