Nvidia doubles down on open-source AI

Nvidia (NVDA) moved to expand its open source artificial intelligence efforts through an acquisition and a new model release.

The semiconductor company said Monday it acquired SchedMD, the main developer of the open source workload management system Slurm. Nvidia said Slurm will continue to operate as open source, vendor-neutral software focused on high-performance computing and AI.

Slurm was first released in 2002. SchedMD was founded in 2010 by Slurm developers Morris Jette and Danny Auble. Auble currently serves as the company’s chief executive officer.

Financial terms of the deal were not disclosed. Nvidia declined to comment beyond details shared in a company blog post.

Nvidia said it has worked with SchedMD for more than a decade. The company described Slurm as critical infrastructure for generative AI systems. Nvidia said it plans to continue investing in the technology and expand its use across different computing environments.

Alongside the acquisition, Nvidia released a new family of open AI models. The company said the models, called Nvidia Nemotron 3, are designed to efficiently support the development of AI agents.

The Nemotron 3 family includes several models aimed at different workloads. Nemotron 3 Nano is intended for targeted tasks. Nemotron 3 Super is designed for multi-agent AI applications. Nemotron 3 Ultra is built to handle more complex tasks.

Jensen Huang, Nvidia’s founder and chief executive officer, said open innovation plays a central role in AI development. He said the Nemotron models are intended to provide developers with transparency and efficiency when building agent-based systems at scale.

The announcement follows other recent moves by Nvidia to expand its open source and open AI offerings. Last week, the company introduced an open reasoning vision language model called Alpamayo-R1. Nvidia said the model is focused on autonomous driving research.

Nvidia also said it added new workflows and guides related to its Cosmos world models. Those models are open source under a permissive license and are intended to support the development of physical AI systems.

The recent activity reflects Nvidia’s focus on physical AI as a growth area for its graphics processing units. The company is positioning itself as a supplier of AI hardware and software for robotics and self-driving vehicle developers.