The problem Minerva targets is scale. With hundreds of thousands of GPUs working across high-speed networks, a single degraded component, such as a GPU running hotter than its peers or an NVMe drive losing performance under load, can slow an entire job, and the warning signs are easy to miss. Traditional monitoring tools tend to split the data centre into separate views for networking, compute, site reliability and machine learning teams.

Minerva connects those signals across GPUs, networking, storage, schedulers and workloads, building a dynamic knowledge graph that maps the physical infrastructure and the jobs running on it. That context lets operators trace a slow training job to the component affecting it. The platform analyses billions of data points per hour, learning normal behaviour, detecting unseen anomalies, predicting emerging failures and correlating related events, then grouping an event storm into a focused incident with a likely root cause.

A native AI assistant offers a natural-language interface grounded in live telemetry, and teams can either keep a person in the loop to apply fixes manually or connect Minerva to automation through infrastructure APIs. NVIDIA frames the platform's value around AI factory yield, measured through GPU hours, goodput, cluster uptime and token production, with claimed benefits including longer stretches between interruptions and faster time to repair.

IREN is 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 it as part of their AI factory operations portfolios.