Reference

14. GPU & Compute Dashboards

Bill Rondell, Nick Adams & Dade Murphy — reference — 1 min read — 48 words

Source: https://github.com/nuclide-research/AI-LLM-Infrastructure-OSINT/blob/main/shodan/queries/14-gpu-compute

Section verified: April 2026

Telemetry and management interfaces for the underlying compute. These surfaces disclose GPU inventory, utilization, workload metadata, and, in the case of cluster schedulers, the ability to submit arbitrary jobs that will run on someone else’s hardware.

Shodan Query	Notes
`"NVIDIA DCGM" port:9400 "/metrics"`
`"nvidia-smi" port:8080`
`"RunPod" port:8888`
`"Vast.ai" port:8080`
`"GPUStack" port:80`