NVIDIA says it has launched Minerva, an operations platform for monitoring AI factories. It combines data from GPUs, networks, storage, schedulers and AI workloads, then uses AI models to flag unusual behavior and connect related problems to likely causes.
Operators can ask Minerva’s built-in assistant about cluster health and get suggested fixes. Teams can apply those fixes themselves or connect the platform to automation. IREN is already using Minerva, while eight other organizations plan to deploy it, according to NVIDIA.
Why it matters
For teams running large AI clusters, Minerva could make it easier to trace slow jobs to infrastructure problems, though NVIDIA’s announcement does not provide measured results or pricing.
