Idle Power, By The Numbers

209 of the 503 mini PCs in this database carry a third-party measured idle-power reading. Nobody else aggregates that across the whole market — it exists, scattered across hundreds of individual reviews, and this page is what it says once it is all in one place. Rebuilt from the live dataset on every deploy; it will not go stale the way a written-once report does. Last built 2026-09-15.

Overview ¶

Machines with measured idle

209 of 503 (41.6%)

Median idle draw

7.3W

Range observed

1.9–46.5W

Mean idle draw

9.3W

n = 209 models with a sourced idle-power figure, out of 503 in the dataset built 2026-09-15. Idle power here is the reviewer's reported idle-minimum reading, chiefly from NotebookCheck's measured mini-PC reviews. Where the same review also reports an idle-median reading (n=209), it runs 1.7W higher on average — said here because using the more flattering of two available numbers without disclosing it is exactly the kind of thing that should sink a data page's credibility.

How idle draw is distributed ¶

77% of measured machines idle under 10W; the tail above 30W is almost entirely older, higher-TDP chips (see below).

Does a higher-TDP chip mean higher idle draw? ¶

Grouped by the CPU's rated TDP class (sourced alongside the CPU benchmark, same source) — not by idle draw itself.

Does a higher price buy lower idle draw? ¶

What idle power actually costs, per year ¶

kWh/year = idle watts × 8,760 hours ÷ 1,000. That part is arithmetic on a sourced number. Turning it into a dollar figure needs your own electricity rate, which is not something this dataset knows — so it is a small calculator, not an asserted average. Edit the rate; the figures below update.

Your rate ($/kWh, not sourced — edit it)

Lowest and highest measured idle draw ¶

Highest measured (worth knowing, not recommending):

What manufacturers don't tell you ¶

Coverage across all 503 models in the dataset, not just the 209 with a measured idle watt — this is how often each field is published or measured anywhere we could find it, for the whole catalogue.

SATA bays (6.2%) and M.2 slot count (30%) are the two worst-covered fields in the whole database — not because we stopped looking, but because almost no vendor states either on a spec sheet, and only a fraction of models have a published third-party review to pull it from. Idle power itself (41.6%) is better covered than most of this list, but still means most listings in the wild never say it at all — which is the whole reason this page and the database behind it exist.

Methodology, sample sizes and limits ¶

• We do not measure power. Every idle-watt figure is a third-party measurement, chiefly NotebookCheck's Kill A Watt-class bench, credited on the source link next to the figure it came from. IdleWatt runs its own measured idle database with a published methodology and is worth reading directly if idle draw is the only thing you care about.
• idle_w is the idle-minimum reading, not idle-median — see the overview section above for the measured gap between the two where both are available.
• 209 of 503 models (41.6%) have a sourced idle measurement. Every ranking and average on this page is computed over those 209, never extrapolated to the rest.
• Correlations are Pearson's r on the exact sample sizes stated next to each one (n=197 for TDP, n=57 for price) — not the full 209, because both need a second sourced field that not every measured machine has.
• What this page does NOT support: a causal claim about TDP or price CAUSING idle draw (these are observational correlations across whatever the market happens to ship, not a controlled comparison); any claim about idle draw for a specific model that is not in the 209-machine sample; and the dollar figures in the cost section, which depend entirely on the electricity rate you enter and are not a sourced statistic.

How to cite this page ¶

Every heading and headline number above has its own permalink (click the ¶, or the anchor is already in your address bar). A citation that will still resolve after this page updates: