Overview

Fleet Overview

Real-time metrics across your multi-cloud accelerator fleet: Nvidia, Tenstorrent, AMD, Google TPUs, AWS chips

Total GPUs / Accelerators
325

Across 55 physical hosts

Global Utilization
31.4%

Aggregated 24h average

Idling GPU Hours
5,352

Total idle time over 24h

Estimated Savings
$16.1K / day

Potential via better utilization

Fleet Utilization Trend
Aggregated 24h usage across environments
Hardware Distribution
Active GPU models in fleet
NVIDIA130 (40%)
Tenstorrent78 (24%)
AMD24 (7%)
Google93 (29%)