# Akash GPU allocation ratio versus actual utilization

Allocated GPUs divided by total GPUs describe a capacity snapshot. They do not measure kernel utilization, hours occupied over the day or the share of revenue already settled.

- Canonical: https://iotahome.site/en/learn/akash-gpu-allocation-ratio
- Publisher: IOTA Watch
- Topics: Akash
- Published: 2026-10-26
- Updated: 2026-10-26

## At a glance

Allocated GPUs divided by total GPUs describe a capacity snapshot. They do not measure kernel utilization, hours occupied over the day or the share of revenue already settled.

## What to check

Match the akash1 provider address, source time and both capacity fields. Do not calculate a ratio with a missing or zero denominator. A provider may span several servers, so its aggregate ratio cannot be assigned to each physical device. Consult actual workload telemetry and lease records for activity over time.

## Example

Example: 3 allocated GPUs out of 8 yield a 37.5% capacity ratio at that instant. This does not mean every GPU ran for 37.5% of the day, and it cannot establish electricity cost or net income.

## Continue

[Open project monitor](https://iotahome.site/en/projects/akash)

## Sources and implementation

- [akash.network](https://akash.network/docs/providers/getting-started/)
- [console.akash.network](https://console.akash.network/providers)