Skip to content
aiqbench takeChips & Infra

aiqbench analysis

Minerva is an operations layer for GPU clusters, aimed at reducing downtime and alert noise in AI data centres. Its claimed "billions of data points per hour" and failure-prediction capabilities are vendor figures that independent benchmarks would need to confirm. For infrastructure teams, the practical question is whether it measurably improves GPU uptime and cuts incident-response time compared with existing observability tools.
Sources covering36Kr, Latent Space, Quantum Zeitgeist, Quantum Computing Report

Nvidia Launches Minerva, an Operations Platform for AI Factories That Can Analyze Billions of Data Points per Hour

36Kr1 min read

Nvidia has launched Minerva, an AI operations platform designed to detect failures early, improve GPU uptime and boost the output efficiency of AI factories by integrating telemetry data from infrastructure and AI workloads. The platform can analyze billions of data points per hour, identify anomalies, predict potential failures and consolidate correlated alerts into actionable incident-response items. Minerva also includes a native AI assistant that can query cluster health via natural language and provide repair recommendations. Nvidia said IREN is already using the platform, while Aolani, Bitdeer AI, IBM, Jio, Lambda, Quake AI and Tesla plan to deploy it; Dell, Lenovo and Supermicro will also incorporate Minerva into their respective AI factory operations product portfolios. (Caixin)

Read the full article at 36Kr

Related coverage