NVIDIA Minerva: Launch of a Platform for Observing AI Factories

  • PublicPublic
  • 22 Oct, 2026
NVIDIA Minerva: Launch of a Platform for Observing AI Factories

NVIDIA has officially launched Minerva, a specialized AIOps platform engineered to maintain productivity within large-scale AI factories. The system aggregates telemetry data from both underlying infrastructure and active AI workloads, applying purpose-built artificial intelligence models to transform raw data into actionable insights. AI factories represent some of the most complex operational environments globally, where vast arrays of GPUs operate across high-speed networks to train models and serve applications. In such settings, a single degraded component can significantly slow down entire jobs, making early detection of issues critical for maintaining efficiency and output.

Traditional monitoring tools often fragment data center visibility, presenting separate views for networking, compute, site reliability, and machine learning teams. Minerva addresses this fragmentation by connecting signals across NVIDIA GPUs, networking hardware, storage systems, schedulers, and workloads. The platform constructs a dynamic knowledge graph that logically maps the physical infrastructure, including GPUs, SuperNICs, switches, and cables, alongside the jobs executing on them. This contextual mapping allows operators to trace performance bottlenecks to specific components, even in facilities containing hundreds of thousands of GPUs, ensuring that root causes of slow training jobs are identified accurately and quickly.

The platform processes billions of data points per hour using both supervised and unsupervised machine learning algorithms. It learns normal operational behavior, detects previously unseen anomalies, predicts emerging failures, and correlates related events based on the factory's specific topology. Rather than flooding operators with every individual symptom, Minerva groups event storms into focused incidents with likely root causes. The system includes a native AI assistant and connects to external agentic AI tools, providing a natural-language interface grounded in live telemetry. Operators can investigate cluster health and receive recommended remediation steps through conversation, while retaining the ability to manually implement fixes or connect to automation via infrastructure APIs.

Minerva is designed to simplify operations by aggregating related problems into single actionable items, enabling AIOps teams to shift from reactive troubleshooting to proactive maintenance. This approach helps maximize GPU uptime and improves key metrics such as yield, goodput, cluster uptime, and token production. Several major partners, including IREN, Aolani, Bitdeer AI, IBM, Jio, Lambda, Quake AI, and Tesla, plan to deploy the platform. Additionally, hardware vendors Dell Technologies, Lenovo, and Supermicro will offer Minerva as part of their AI factory operations portfolios, integrating it into broader infrastructure solutions.