Nvidia has unveiled NemoClaw, a software stack that integrates with OpenClaw, the autonomous AI agent platform that became one of the fastest-growing open-source projects in recent history. The announcement came at Nvidia’s annual GTC conference in San Jose.
NemoClaw is not a replacement for OpenClaw. It functions as an enterprise-ready distribution of the platform, installing in a single command and adding security, privacy, and governance controls that most organizations require before deploying autonomous agents on production systems.
The stack has two main components. The first is Nvidia’s Nemotron family of open models, which can run locally on dedicated hardware instead of routing data through external cloud services. The second is OpenShell, a new open-source security runtime that runs each agent inside an isolated sandbox. Administrators define policies in YAML files, specifying which files, network connections, and cloud services an agent can access. Everything outside those limits is blocked.
Nvidia CEO Jensen Huang framed the announcement in broad terms: “Claude Code and OpenClaw have sparked the agent inflection point — extending AI beyond generation and reasoning into action.”
NemoClaw can run on Nvidia GeForce RTX PCs, RTX PRO workstations, and the company’s DGX Spark and DGX Station systems. A built-in privacy router directs sensitive queries to local models while routing tasks that require more capability to cloud providers, without exposing private data to those external endpoints.
Enterprise software companies including Box, Cisco, Atlassian, Salesforce, SAP, and Adobe are listed as integration partners. Nvidia developed NemoClaw in collaboration with OpenClaw founder Peter Steinberger.
Sources: VentureBeat, The New Stack
