NVIDIA Unveils Minerva AIOps Platform for AI Factory Infrastructure Observability
NVIDIA unveiled Minerva on October 22, 2026, an AIOps observability platform that connects infrastructure and workload telemetry across hyperscale AI factories. The platform targets a direct threat to enterprise AI profitability: silent hardware degradation and cascading job failures that erode compute yield during multi-week training runs and continuous inference serving.
Minerva synthesizes telemetry from physical hardware - including NVIDIA GPUs, SuperNICs, switches, cables and storage - together with workload schedulers and distributed training jobs, building a dynamic knowledge graph that logically maps the factory and everything running on it. In high-density clusters, a minor issue such as a GPU running hotter than its peers or an NVMe drive losing performance under load can stall an entire synchronous training job without triggering a hard failure.
The platform analyzes billions of data points per hour using supervised and unsupervised machine learning to learn normal behavior, detect previously unseen anomalies, predict emerging failures and correlate related events based on factory topology. Rather than forwarding every symptom to operators, it collapses alert storms into a focused incident with a likely root cause.
A native AI assistant, which also connects to external agentic AI tools, gives site reliability engineers a natural-language interface grounded in live telemetry: operators can query cluster health conversationally and receive recommended remediation, then apply fixes manually or connect Minerva to automation through infrastructure APIs.
IREN is among the NVIDIA partners already using Minerva, and Aolani, Bitdeer AI, IBM, Jio, Lambda, Quake AI and Tesla plan to deploy it; Dell Technologies, Lenovo and Supermicro will offer Minerva as part of their AI factory operations portfolios. NVIDIA positions the platform around AI factory yield - GPU uptime, goodput and token production - shifting incident management from reactive troubleshooting to predictive maintenance.
For infrastructure directors and data center operators, the buyer cautions are practical: observability identifies and correlates root causes but does not itself fix failing hardware, so spare capacity or automated node substitution is still required. NVIDIA has not published standalone pricing or licensing terms for Minerva, which is expected to integrate into broader NVIDIA AI Enterprise and DGX SuperPOD agreements, and full knowledge-graph coverage requires deep telemetry integration across supported networking and storage fabrics, a constraint for heterogeneous data centers.