Nvidia will introduce the open-source project NemoClaw, enabling the development of custom AI agents without using its own chips.

Nvidia plans to launch an open project called “NemoClaw” – a platform for building AI agents

The new initiative, named NemoClaw, is intended to become a development tool for autonomous AI agents. According to Wired (citing its own sources), the platform will be open and available not only to customers using Nvidia hardware.

What developers can expect
* Open source code – any programmer will be able to access the core of the project.
* Security and privacy – the suite includes data protection tools and agent behavior controls.
* Partnerships before the conference – next week Nvidia will hold its annual developer meet‑up. In anticipation of the event, the company reached out to Salesforce, Cisco, Google, Adobe, and CrowdStrike with a proposal for collaboration on the NemoClaw project. The current status of negotiations is unknown.

Why it matters
Demand for AI agents is growing: everyday users already employ solutions like OpenClaw, which runs agents directly on personal computers to perform routine tasks. Although leading companies (OpenAI, Anthropic) have improved the reliability of their models, chatbots still require constant oversight due to potential failures.

For Nvidia, NemoClaw could become:
* A new channel for attracting corporate clients;
* A tool for enhancing AI agent security;
* A step toward promoting open AI models in data‑center infrastructure while maintaining the company’s leading position.

Related developments
The company has already announced the creation of its own AI accelerators for inference (running trained models). These accelerators will be based on Nvidia’s acquired Groq technology, underscoring Nvidia’s strategic focus on high‑performance AI processing.