RUM Group Inc. plans to add NVIDIA Minerva to the systems supporting Quake AI’s computing services, saying the software may help staff identify infrastructure problems sooner.
Quake AI, RUM Group’s AI infrastructure unit, provides bare-metal GPU capacity for AI training and inference. The company says its AI data centers also run CPU systems and include storage and networking equipment. Founder and CEO Chris Pavlovski said customers value direct access because it gives them control over their computing resources.
Minerva gathers operating information from GPUs, networking and storage, then correlates it with job scheduler records and AI workload data, RUM Group said. This lets operations staff assess equipment signals alongside information about the computing tasks underway. The platform is designed to detect unusual behavior, help identify likely causes and flag potential failures. RUM Group said the system could help staff move toward preventive maintenance and faster incident resolution; Pavlovski said it will help teams “identify issues earlier.”
Pavlovski said Quake AI’s GPU fleet has surpassed 22,000, and NVIDIA Vera Rubin deployments are part of the company’s plans for 2027. He said bare-metal service lets customers decide how their compute is used. RUM Group presents Minerva as a way to help its staff support the reliability, uptime and performance customers expect.
Quake AI appeared on NVIDIA’s Oct. 22 list of organizations preparing to deploy Minerva. Aolani, Bitdeer AI, IBM, Jio, Lambda and Tesla were also listed as prospective users, while Dell Technologies, Lenovo and Supermicro planned to offer the software through their AI factory portfolios. EconoTimes’ earlier coverage of Minerva noted that IREN was already using the platform. That coverage also cited NVIDIA Mission Control release notes describing Minerva 1.0.0 as pre-release beta software with limited access and no general release date listed.
