NVIDIA Launches AI Factory Operations Platform Minerva, Analyzing Billions of Data Points Per Hour

Odaily reports: NVIDIA has launched Minerva, an AI operations platform designed to integrate infrastructure and AI workload telemetry data to detect failures in advance, improve GPU uptime, and boost AI factory output efficiency. The platform can analyze billions of data points per hour, identify anomalies, predict potential failures, and consolidate related alerts into actionable fault-handling items.

Minerva also features a native AI assistant that enables natural language queries about cluster health and provides remediation suggestions. NVIDIA stated that IREN is already using the platform, while Aolani, Bitdeer AI, IBM, Jio, Lambda, Quake AI, and Tesla plan to deploy it; Dell, Lenovo, and Supermicro will also incorporate Minerva into their respective AI factory operations product portfolios.