March 11, 2026 Nvidia is planning to launch an open-source platform designed to help companies deploy artificial intelligence agents to perform workplace tasks, according to a report from WIRED citing people familiar with the company’s plans. The platform, reportedly called NemoClaw, would allow enterprise software companies to build and run AI agents regardless of whether their products use Nvidia chips.
The chipmaker has been presenting the concept to several large enterprise software vendors ahead of its annual developer conference in San Jose next week. According to the report, Nvidia has held discussions with companies including Salesforce, Cisco, Google, Adobe and CrowdStrike about potential collaboration around the platform.
It remains unclear whether those discussions will result in formal partnerships. Because the platform is expected to be open source, early collaborators could gain access in exchange for contributing to the project’s development.
The platform is expected to include security and privacy features designed for enterprise environments. AI agents, sometimes referred to as “claws,” are software systems capable of executing multi-step tasks autonomously, such as completing workflows or coordinating actions across applications.
Interest in agent-based AI tools has grown as companies explore ways to automate complex business processes. Earlier this year, a project called OpenClaw gained attention in Silicon Valley for its ability to run locally on personal computers and perform work tasks without constant human input. OpenAI later hired the project’s creator and began open-sourcing the technology.
Despite the growing interest, enterprise adoption of autonomous agents has raised security concerns. Some technology companies have restricted employee use of experimental agent software because of unpredictable behaviour or potential risks to corporate systems.
Nvidia’s reported move toward an open-source agent platform marks a shift in its broader software strategy. The company’s developer ecosystem has long centred on CUDA, its proprietary platform for programming Nvidia GPUs.
The initiative also comes as Nvidia prepares additional announcements for its developer conference. The Wall Street Journal reported last month that the company plans to introduce a new inference computing system that incorporates a chip developed by startup Groq under a multibillion-dollar licensing agreement.
