Quake AI with over 22,000 GPUs, pushing to adopt NVIDIA Minerva

Quake AI is moving to adopt Nvidia's Minerva AI data-center operations platform, according to an announcement from RUM Group ($RUM), which disclosed that Quake AI already runs more than 22,000 Nvidia GPUs across its AI training and inference infrastructure business. Minerva aggregates and analyzes telemetry from GPUs, networking, storage, job schedulers and AI workloads, using a dynamic knowledge graph to link equipment status with AI jobs and correlate alerts for fault root-cause tracing. The move matters because it ties a large third-party GPU fleet to Nvidia's management stack, potentially improving utilization and reliability across tens of thousands of accelerators. Watch for confirmation of the deployment timeline and scope, and for any disclosure of how Minerva affects uptime or utilization metrics at Quake AI.