TOKENPOST
1 min read
Add as a preferred source on Google

Nvidia Introduces OpenShell Architecture for AI Agents With Limits

Nvidia also said DSX MaxLPS can cut electricity use per AI token by up to 30%, while DSX targets near-zero facility cooling water use at a 2-gigawatt scale.

By Simon Yoon
Published
A networking chip sits beneath a clear protective enclosure. / TokenPost.ai
A networking chip sits beneath a clear protective enclosure. / TokenPost.ai

Nvidia introduced an OpenShell reference architecture on Oct. 21 for deploying autonomous AI agents within defined limits.

The reference architecture supports building, controlling, observing and running secure agents across enterprise infrastructure and edge systems.

Nvidia also said DSX MaxLPS can reduce electricity use per AI token by up to 30%. At a 2-gigawatt deployment scale, DSX could bring water use for facility cooling close to zero.

OpenShell is intended to let AI agents perform useful work while staying within clearly defined boundaries.

Simon Yoon

Reporter

Simon Yoon reports on blockchain technology for TokenPost. Send corrections or tips to info@tokenpost.com.

Loading…