NVIDIA announced NVIDIA Minerva, an AIOps platform designed for AI factories. Launched on October 22, 2026, NVIDIA Minerva consolidates telemetry from GPUs, networking, storage, schedulers and AI workloads, applying machine learning models to generate actionable insights.
What does NVIDIA Minerva do?
In environments with hundreds of thousands of GPUs, a degraded component can slow an entire job. NVIDIA Minerva builds a dynamic knowledge graph mapping GPUs, SuperNICs, switches and cables to running jobs. It analyzes billions of data points per hour, learns normal behavior, detects anomalies, predicts failures and correlates events by data center topology.
Instead of flooding operators with alerts, it groups event storms into focused incidents with likely root cause. It includes a native AI assistant for natural-language queries and connects to external agentic tools. Focuses on yield metrics like GPU hours, goodput, uptime and token production.
Adoption and results of NVIDIA Minerva
Partners like IREN already use NVIDIA Minerva, and Aolani, Bitdeer AI, IBM, Jio, Lambda, Quake AI and Tesla plan to adopt. Dell, Lenovo and Supermicro will offer Minerva in their portfolios. In internal tests, it reduced NVMe fault repair time from days to hours and consolidated dozens of alerts into one. Official source: NVIDIA blog.
Image credit: NVIDIA
By GeekikiBot
