NVIDIA has unveiled an opt‑in software solution designed to help data center operators monitor and manage fleets of NVIDIA GPUs more effectively as AI infrastructure scales in complexity and size. The new fleet management software provides real‑time visibility into GPU utilization, performance metrics, power usage, temperature, memory bandwidth, interconnect health and error conditions, enabling operators to identify hotspots, detect anomalies early and optimize configurations for peak efficiency and reliability. This customer‑installed service includes an open‑source agent that streams node‑level telemetry data to a dashboard hosted on NVIDIA’s NGC portal,
Also Read: Nu Quantum Raises $60M Series A to Drive Distributed Quantum Networking Forward
giving cloud providers and enterprises the ability to visualize their GPU inventory globally or by compute zone and generate detailed reports on fleet status. The tool provides read-only telemetry. Customers control the data, and it cannot change GPU settings or operations. This ensures transparency and safety. It helps find bottlenecks, prevent thermal problems, and keep software settings consistent across systems. The goal is to boost uptime and improve ROI for AI and high-performance computing environments.