Mini PC power consumption for a 24/7 home server: how to measure and reduce it
Updated 2026-09-25
Idle power is the number that matters for a home server, because it sits idle most of the day. It is also the number most often guessed. We publish a model's idle figure only with its source, configuration and method; until then the model page says "Not measured yet". The published readings so far are collected on the low-power list.
What changes idle power
- BIOS power settings: deep CPU package C-states and PCIe ASPM enabled make the biggest difference. Some ex-office units ship with performance profiles.
- Drives and cards: each NVMe SSD, 2.5" drive and especially a 10GBASE-T card adds idle draw.
- The operating system: Linux with
powertop --auto-tuneapplied usually idles lower than a default install. - The power adapter: a higher-wattage brick is not less efficient at idle by default, but a failing third-party adapter can be.
- Displays and USB devices: measure headless if the server will run headless.
How to measure it properly
- Use a plug-in wall meter between the adapter and the socket (this measures what you pay for).
- Boot into the real OS and workload (e.g. Proxmox with your VMs idle) and wait 10 minutes to settle.
- Note the lowest stable reading over a few minutes, plus CPU, RAM, drives, OS, BIOS settings and mains voltage.
Measured one? Send us the reading with those details and we will add it to the model page with credit, labelled as a community measurement.
From watts to cost
Yearly energy = watts × 24 × 365 ÷ 1000 kWh. So every 1 W of continuous draw is 8.76 kWh per year. Multiply by your electricity price per kWh: for example, a 10 W difference at 0.30 per kWh is about 26 per year in your currency.
Official sources to check
Many of these models were ENERGY STAR certified, and the ENERGY STAR product finder lists an idle-state power figure measured under a standard test for specific certified configurations. It is a useful comparison point, but your configuration and OS will differ.