NVIDIA has acquired SchedMD, the company best known as the primary commercial steward behind Slurm, the open-source workload manager widely used to schedule jobs across high-performance computing (HPC) clusters.
The deal, announced December 15, 2025, positions Slurm even closer to the center of modern AI infrastructure—right as more organizations build GPU-heavy clusters for training, inference, and large-scale data processing.
Why NVIDIA wants Slurm
Slurm sits in the unglamorous but essential layer between users and compute: it queues jobs, schedules workloads, and allocates resources across large clusters. As HPC and AI systems scale, that scheduling layer becomes a performance feature in its own right—determining how efficiently expensive hardware gets used.
NVIDIA is pitching the acquisition as a way to strengthen the open-source ecosystem around Slurm while improving how accelerated compute is managed in real deployments.
NVIDIA says Slurm will remain open source and vendor-neutral
In its announcement, NVIDIA said it will continue to develop and distribute Slurm as open-source software and keep it vendor-neutral, supporting the broader HPC and AI community across diverse hardware and software environments.
That commitment is likely to be closely watched. Slurm is deeply embedded across universities, national labs, and hyperscale operators—and those environments tend to care as much about neutrality and portability as they do performance.
Slurm’s footprint in supercomputing and AI
NVIDIA and SchedMD also pointed to Slurm’s prevalence across top-tier supercomputing systems, saying it’s used in more than half of the top-ranked machines in the TOP500 list.
Beyond traditional HPC, Slurm is now a familiar tool in AI clusters as well. It’s used to orchestrate training runs, manage shared GPU pools, and handle the practical realities of multi-tenant infrastructure—especially as model sizes and job complexity grow.
What changes for customers
NVIDIA said it will continue offering support, training, and development around Slurm for SchedMD’s customer base, and that the acquisition should help expand Slurm’s reach to new systems—while still supporting heterogeneous clusters.
SchedMD CEO Danny Auble framed the acquisition as validation of Slurm’s importance in demanding environments, and said Slurm will remain open source as NVIDIA invests in its ongoing development.
The bigger picture
This move isn’t just about software—it’s about control of the plumbing that makes large clusters usable.
As AI spending shifts from single-node experimentation to fleet-scale deployment, companies are increasingly competing on infrastructure maturity: scheduling, telemetry, orchestration, and reliability at scale. By bringing SchedMD in-house, NVIDIA is making a long-term bet that job scheduling is a strategic layer of AI compute—not a commodity.