According to 1M AI News monitoring, Nvidia announced in December 2025 that it would acquire SchedMD, a company that is a primary developer of the open-source workload scheduling software Slurm. Slurm is widely used to manage the scheduling of computing tasks in supercomputers and AI data centers. It is said that about 60% of the world’s supercomputers are using it, including AI training clusters from companies such as Anthropic, Meta, and Mistral, as well as supercomputers used by governments in multiple countries for weather forecasting and nuclear weapons development. Slurm is especially good at managing Nvidia chips, but it is also widely used on non-Nvidia hardware.
This acquisition has recently begun to raise concerns among AI experts and supercomputer users. They worry that Nvidia may subtly steer toward its own chips (such as CUDA and InfiniBand) through software updates, thereby harming the hardware performance of competing hardware from companies such as AMD and Intel. Some people cite Nvidia’s prior acquisition of Bright Computing, arguing that similar integration could lead to performance penalties for other chips. Intersect360 Research CEO Addison Snell said the concern is that Nvidia may transform this general-purpose open-source tool into a product that is “better suited to, or exclusive to, its own hardware,” affecting fair competition.
Nvidia responded that Slurm is still open-source software. The company said it will continue to provide enhancements and support for all users, emphasized its commitment to “open, vendor-neutral” development, and said that customers will generally benefit. The company plans to maintain training and technical support for SchedMD’s hundreds of customers, while denying that past acquisitions would damage multi-hardware compatibility. This event is viewed as an important test of whether Nvidia in the AI and high-performance computing space truly remains open.