Nvidia acquires US AI software provider SchedMD

Nvidia has acquired SchedMD, a US-based AI software provider known for its open-source workload manager Slurm.

The deal comes as Nvidia increases its focus on open-source technology to strengthen its position in the AI sector amid rising competition.

SchedMD’s Slurm software is used to schedule and manage large-scale computing jobs, and is widely adopted by researchers and companies working with high-performance computing and AI.

Nvidia said it plans to keep distributing SchedMD’s software as open source.

Financial details of the acquisition were not disclosed.

SchedMD was founded in 2010 in Livermore, California, and employs about 40 people.

🔗 Source: Nvidia

🧠 Food for thought

Implications, context, and why it matters.

Slurm’s role in high-performance computing makes this an infrastructure move

More than half of the top 10 supercomputers in the TOP500 list 1 (an independent ranking of the world’s most powerful supercomputers) use Slurm. As of November 2024, 5 of the top 10 systems run it 2.
The scheduler handles 100,000+ nodes or GPUs, 17 million+ jobs per day, and 120 million+ per week 1. That scale fits exascale environments (systems operating at a scale of a billion-billion operations per second). Owning the company lets Nvidia shape how well its hardware gets used.
Rivals such as AMD are building GPU stacks with Slurm integration 3. AMD’s MI300X (a data center GPU) docs lay out Slurm-based multi-node training configs 3, which makes the scheduler central to GPU cluster management.

Cloud providers can speed up managed Slurm to stand apart from Nvidia

In October 2024, Crusoe became the first cloud to virtualize AMD’s MI300X GPUs with Linux Kernel-based Virtual Machine (KVM) 4. Providers can offer hosted Slurm that stays vendor neutral across multiple GPU architectures.
Nebius built Soperator, an open-source Kubernetes operator (an automation controller) that runs and manages Slurm clusters as Kubernetes resources 5. Others could fork or extend this to fix autoscaling gaps, then ship distinct services.
Nvidia says it will keep distributing Slurm as open source 6, so providers can add features like stronger security, compliance tooling, or easier deployment automation for customers worried about vendor lock-in with an Nvidia-run scheduler.