NVIDIA NemoClaw: Secure, Always-On AI Agents With One Command
Announced at GTC 2026, NemoClaw is NVIDIA's open-source stack that adds privacy and security controls to OpenClaw. It bundles Nemotron models and the new OpenShell runtime into a single-command install for autonomous AI agents.
Key Takeaways
- 01
NemoClaw adds enterprise-grade privacy and security guardrails to OpenClaw — what Jensen called 'the operating system for personal AI.'
- 02
Uses a privacy router that splits inference between local Nemotron models and cloud frontier models, keeping sensitive data on your hardware.
- 03
Microsoft Security already reports a 160x improvement in finding AI-based attacks using Nemotron and OpenShell.
- 04
Deploys with one command on GeForce RTX, RTX PRO workstations, DGX Station (up to 1T parameter models), and DGX Spark.
Key Architecture
- Privacy router — routes inference between local open models (Nemotron running on your hardware) and cloud frontier models, with policy-based security guardrails
- OpenShell runtime — isolated sandbox with data privacy and security controls for autonomous AI agents ("claws")
- NVIDIA Agent Toolkit — the software layer that secures OpenClaw agent operations
- One-command install:
curl -fsSL https://nvidia.com/nemoclaw.sh | bash
Where It Runs
NemoClaw deploys on any dedicated compute: GeForce RTX PCs/laptops, RTX PRO workstations, DGX Station (748GB coherent memory, 20 petaflops, runs models up to 1 trillion parameters), and DGX Spark (supports clustering up to four systems into a compact "desktop data center").
Why It Matters
Jensen made the enterprise security case explicit during the keynote: agentic systems inside corporate networks can access sensitive information, execute code, and communicate externally. He paused and told the audience to think about the implications. NemoClaw is designed to fill exactly that security gap.
Microsoft Security is already using Nemotron and OpenShell for adversarial learning, reporting a 160x improvement in finding and mitigating AI-based attacks.
Jensen's bottom line: just as every company needed an HTTP strategy, a Linux strategy, and a Kubernetes strategy, every company now needs an OpenClaw strategy.
