US Senator Warren Questions Nvidia Acquisition of Slurm Software

Senator Elizabeth Warren has formally requested information from the U.S. Departments of Energy and Defense regarding the national security implications of Nvidia’s acquisition of SchedMD. The deal, which closed in December, grants Nvidia control over Slurm, a workload management software that powers approximately 60% of the world’s supercomputers [1].

Slurm is a critical component of the U.S. government’s computing infrastructure, used for tasks ranging from ballistic missile simulations to nuclear weapons development [1]. Because the software also serves as the backbone for major AI companies like Meta and Anthropic, critics fear that Nvidia’s ownership could allow the company to prioritize its own hardware over that of competitors like AMD and Intel [2].

In her letter to Energy Secretary Chris Wright and Defense Secretary Pete Hegseth, Warren argued that the acquisition could turn previously free, open-source software into a proprietary tool that reduces competition [1]. By controlling both the AI chips and the software layer that manages them, Nvidia is in a position to make its hardware easier to deploy while potentially boxing out rival firms [1].

Industry analysts note that Nvidia’s control over the Slurm roadmap gives it significant "soft power" to shape how efficiently competing chips perform in shared environments [2]. While Nvidia has stated that it will continue to provide Slurm as open-source, vendor-neutral software, some experts remain skeptical [2]. They point to Nvidia’s 2022 acquisition of Bright Computing as a precedent, where the software was increasingly integrated into Nvidia’s own hardware stacks rather than maintaining a neutral, multi-vendor role [2].

For users of mixed-vendor clusters, the primary concern is whether Nvidia will subtly favor its own ecosystem through code prioritization or faster integration timelines for its forthcoming hardware [2]. Dr. Danish Faruqui, CEO of Fab Economics, noted that the risk of under-optimizing support for rival chips is a "feasible outcome" now that Nvidia manages the official code review process [2].

The departments have yet to detail whether they have conducted a formal assessment of the risks posed by the transition of this critical infrastructure to a single, dominant vendor. The ultimate test of Nvidia’s commitment to neutrality may emerge in how quickly it integrates support for next-generation competitor chips compared to its own proprietary networking technologies [2].