Nvidia plans NemoClaw launch, an open-source platform for AI agents
Nvidia is planning to launch NemoClaw, an open-source platform for AI agents, next week.
The product is reportedly already being pitched to the usual big names (e.g., Salesforce, Cisco, Google, Adobe, and CrowdStrike) with the promise of helping them deploy AI agents to perform tasks that support their employees, Wired reported on Monday. It would be a welcome relief for companies eager to finally see some ROI from AI investments, especially given Gartner estimates that more than four in ten agentic AI projects will be dead in the water by 2027.
Anonymous sources cited by Wired say even non-Nvidia customers will be able to get in on the action with access to the new platform open to all companies, whether they’re running on Nvidia chips or not.
Nvidia claws its way deeper into the AI agent ecosystem
Stirrings of NemoClaw come at an interesting time for the chipmaker and the industry at large as conversations increasingly turn to “claws,” i.e., a new type of AI assistant that runs locally and performs tasks autonomously.
AI power users are likely already familiar with the OpenClaw story, the open source autonomous AI agent that quickly became GitHub’s most-starred project — and then almost even more quickly got acquired by OpenAI, with owner Peter Steinberger signing on to join Sam Altman’s organization.
Along the way, OpenClaw (and its claw contemporaries) have come under fire for security risks, with one researcher hijacking the agent in fewer than two hours.
With Wired reporting Nvidia’s plans to offer an additional layer of security and privacy tools, it seems NemoClaw may be their attempt to address tech companies’ reservations about claws’ security shortcomings.
The new open-source agent platform may also be Nvidia’s response to the fast-changing tides of AI development.
As AI labs begin to bring chip-building in-house, Nvidia is likely looking for ways to maintain its central role in AI software and agent development — a notable strategy shift apparent in the chipmaker’s recent releases of Nemotron and Cosmos, foundational models designed to support the full AI agent lifecycle.
Going open source is a telling move, too. It could be a signal that Nvidia aims to focus on ecosystem development and growth instead of short-term licensing revenue.
Expected unveiling next week in San Jose
So far, Nvidia and suspected partners have kept mum about any formalized agreements. But since NemoClaw is an open-source platform, it would make sense that any early partners signing on would also be active contributors to the project.
Noise on NemoClaw comes ahead of Nvidia’s annual GTC developer conference, taking place March 16-19 in San Jose, California, when the industry can expect a full dive into the platform’s capabilities and release timeline.
There’s also reporting in The Wall Street Journal that Nvidia will announce a new chip system for inference computing featuring a Groq-designed chip, the startup with which Nvidia signed a licensing deal in 2025.
The post Nvidia plans NemoClaw launch, an open-source platform for AI agents appeared first on The New Stack.
