NVIDIA is preparing to launch NemoClaw, an open-source platform for deploying AI agents across enterprise workforces, according to people familiar with the company's plans who spoke to WIRED. The full reveal is expected at GTC 2026, NVIDIA's annual developer conference running March 15–19 in San Jose, where CEO Jensen Huang delivers the keynote on March 16.

What NemoClaw Does

NemoClaw will allow companies to dispatch AI agents — automated tools that can plan, reason, and execute multi-step tasks independently — on behalf of their employees. Unlike NVIDIA's existing CUDA platform, which ties developers to NVIDIA's GPU ecosystem, NemoClaw is hardware-agnostic: companies can run it regardless of whether their infrastructure uses NVIDIA chips.

The platform will ship with built-in security and privacy tools, a direct response to concerns about AI agents running unsupervised in enterprise environments.

Courting Enterprise Partners

NVIDIA has reportedly held preliminary conversations with Salesforce, Cisco, Google, Adobe, and CrowdStrike about early partnerships. None of those companies has confirmed a deal. Since NemoClaw is expected to be open source, early partners would likely receive access in exchange for contributing to the project rather than paying license fees.

The Bigger Picture

The move signals NVIDIA's ambition to extend its dominance from hardware into the software orchestration layer of enterprise AI. Jensen Huang recently described OpenClaw — the consumer-facing AI agent that OpenAI acquired earlier this year — as "the most important software release probably ever." NemoClaw appears to be NVIDIA's answer for the enterprise market: a structured, secured version of the same concept built for corporate deployment.

Whether official partnerships are announced at GTC remains to be seen.