NVIDIA Unveils Minerva for AI Factory Observability
NVIDIA has introduced Minerva, a new AIOps platform engineered to bring end-to-end observability, telemetry integration, and proactive maintenance to massive AI factories.
Radwan AbdulhadiEditor and technology writer
Introduction to NVIDIA Minerva
NVIDIA has officially launched [NVIDIA Unveils Minerva to Bring End-to-End Observability for AI Factories](https://blogs.nvidia.com/blog/minerva-platform-infrastructure/), a new AIOps platform built to help teams discover problems earlier and keep large-scale AI operations productive. As modern infrastructure scales to support heavy workloads, AI factories have grown into some of the most complex computing environments in the world, featuring vast numbers of GPUs operating across high-speed networks.
At such immense scale, even a single degraded component—such as a GPU running hotter than its peers or an NVMe drive losing performance under load—can significantly slow down an entire job. Warning signs can easily escape traditional monitoring tools, which often divide the data center into fragmented views where networking, compute, site reliability, and machine learning teams each observe only a fraction of the operation.
Connecting Telemetry and Mapping Infrastructure
The new [NVIDIA Minerva](https://www.nvidia.com/networking/minerva) platform addresses these challenges by consolidating telemetry from both underlying infrastructure and active AI workloads. By applying purpose-built AI models, the system turns raw data into timely and actionable insights focused squarely on AI factory yield—a crucial measure of useful work produced through GPU hours, goodput, cluster uptime, and token production.
Minerva connects signals across NVIDIA GPUs, networking equipment, storage systems, schedulers, and workloads. It builds a dynamic knowledge graph that logically maps out physical infrastructure elements like GPUs, SuperNICs, switches, and cables alongside the jobs running on top of them. This holistic context enables operators to accurately trace a slow training job back to the specific component affecting it, even within clusters spanning hundreds of thousands of GPUs.
Advanced Analytics and Agentic AI Capabilities
Analyzing billions of data points every hour, the platform utilizes both supervised and unsupervised machine learning to learn normal operational behavior, detect previously unseen anomalies, predict emerging failures, and correlate related events based on the factory's exact topology. Instead of overwhelming operators with an event storm of individual symptoms, Minerva groups related alerts into a focused incident report complete with a likely root cause.
Additionally, the platform comes equipped with a native AI assistant and supports connections to external agentic AI tools. This integration provides a natural-language interface grounded in live telemetry, enabling operators to investigate cluster health and receive recommended remediations through conversation. Teams can maintain manual control by keeping a person in the loop, or they can choose to connect Minerva to automation workflows via infrastructure application programming interfaces and external systems.
Key Performance Metrics and Industry Deployment
According to the details shared on the [NVIDIA Blog](https://blogs.nvidia.com/blog/minerva-platform-infrastructure/), the platform delivers tangible value across several key metrics that AI factories rely on to maintain operational efficiency. These include a longer time between interruptions as trouble is spotted early, a faster time to repair through precise problem pinpointing, and greater insight by filtering out data noise to focus only on critical alerts.
IREN stands out as one of the early partners utilizing the technology. Denis Skrinnikoff, chief technology officer of IREN, noted that as their AI factories scale, maximizing GPU productivity is essential for customer service. He explained that the unified view provided across the network, GPUs, and running jobs allows their team to shift from reactive troubleshooting toward proactive maintenance.
Beyond IREN, a diverse roster of organizations including Aolani, Bitdeer AI, IBM, Jio, Lambda, Quake AI, and Tesla plan to deploy the platform. Furthermore, major infrastructure providers such as Dell Technologies, [Lenovo](https://news.lenovo.com/europes-ai-moment-gtc-berlin-26), and Supermicro intend to offer Minerva as an integrated part of their respective AI factory operations portfolios.
Sources
- NVIDIA BlogNVIDIA Unveils Minerva to Bring End-to-End Observability for AI Factories
Continue chronologically
Related entity coverage
Related stories
Humanoid Robots and AI Emerge as Helpers, Not Enemies
Rapid AI advances bring humanoid robots like Boston Dynamics' Atlas out of the lab and onto factory floors, serving as helpers rather than enemies.
Nvidia RTX Spark Laptops Launch With Devices From Major Brands
Nvidia RTX Spark laptops are officially on sale from brands like Microsoft, Asus, HP, Lenovo, and Dell, with prices starting at $2,599.
European Cities Deploy NVIDIA AI Agents for Smart Operations
Cities across the EMEA region are deploying NVIDIA-powered physical AI agents and digital twins to optimize municipal operations and public services.
NVIDIA Launches DSX Ready Program for AI Infrastructure
NVIDIA introduces DSX Ready to help AI factory builders qualify power and cooling solutions, beginning with battery storage and cooling units.
