Nvidia announced NemoClaw, a security and privacy layer for openclaw ai, at its GTC conference in San Jose, CEO Jensen Huang said. Huang described OpenClaw as the platform that made personal agents possible and urged companies to adopt an agentic system strategy.
Huang praised OpenClaw, previously known as Clawdbot and Moltbot, and said the project gave the industry what it needed at the right time. He compared OpenClaw to foundational technologies and called it a turning point for personal agents and the industry.
Nvidia said NemoClaw is an open source stack that adds privacy and security controls to OpenClaw. The company said the stack includes a network guardrail and a privacy router to keep claws from executing unsafely inside organizations.
In its announcement, Nvidia said NemoClaw can let users run always on, self evolving agents anywhere with a single command. The company promoted the tool with a build a claw event where attendees can create custom agents.
Related Announcements And Background
The company also highlighted other platform moves, saying a new inference system will incorporate technology from Groq. Nvidia noted it reached a deal with Groq, described as an AI chip startup, and said the inference system draws on that technology.
Huang projected significant demand for Nvidia hardware and said he expected strong interest in the Blackwell and Rubin AI chips through 2027. The projection was presented alongside the NemoClaw announcement as part of Nvidia’s broader AI strategy.
The OpenClaw project will continue as open source after its creator Peter Steinberger joined OpenAI, Nvidia said. Steinberger added in a statement released by Nvidia that OpenClaw brings people closer to AI and that Nvidia and others are building claws and guardrails for secure assistants.
