Nvidia Slurm Acquisition Sparks Hardware Bias Concerns

Nvidia's acquisition of SchedMD, the developer of the open-source Slurm workload manager, has sparked concerns among AI and supercomputing experts that the chipmaker could prioritize its own hardware in future software updates. The deal, announced in December 2025, gives Nvidia control over Slurm, software that schedules computing tasks across about 60 percent of the world's supercomputers and supports AI training at major companies.

Slurm, originally developed at Lawrence Livermore National Laboratory, manages job queuing and resource allocation in high-performance computing environments. It powers operations at government facilities for weather forecasting and national security research, as well as AI model training at firms including Meta Platforms, French startup Mistral and Anthropic. OpenAI, by contrast, relies on alternative technology from Alphabet's Google.

Industry specialists, including engineers and executives familiar with mixed-vendor GPU clusters, worry that Nvidia's influence over Slurm's development could lead to optimizations favoring its chips over those from competitors like AMD and Intel. Five sources with direct experience in AI and supercomputing...