Free browser-based calculator
Home Server Power Cost Calculator
Estimate the energy use and operating cost of an always-on mini PC, NAS, or home server.
How the calculation works
The calculator uses idle power for the hours the server is not under load and load power for the active hours you enter. It then converts the weighted daily energy use from watt-hours to kilowatt-hours.
Average watts = ((idle watts × idle hours) + (load watts × load hours)) ÷ 24
Annual kWh = average watts × 24 × 365 ÷ 1,000
Annual cost = annual kWh × electricity rateFor the most useful result, measure whole-system power at the wall. CPU package power does not include memory, storage, networking, conversion losses, or attached USB devices.
What should count as load time?
Include backups, media transcoding, photo indexing, updates, scheduled jobs, and any other period when measured power is clearly above idle. If your server runs a steady workload, enter the same value for idle and load power.
Worked example from our EQi12 measurements
Our tested Beelink EQi12 showed about 12–14W in short idle observations and about 37W during sustained CPU load. Using 14W idle, 37W load, and two load hours per day produces an estimated average of 15.9W and about 139kWh per year. This is an example, not a default for every mini PC: attached drives, power settings, display output, updates, and workload schedules all change the result.
For the underlying seven-state measurements, read the EQi12 power-consumption report. Replacing its watts with your own wall-meter observations gives a more useful budget than copying a review number.
Common mini PC and NAS power, with sources
Use these rows to seed the form, then replace them with your own wall reading. Every figure is labelled by what it actually is: a manufacturer power rating is not a consumption figure, a published measurement comes from the third party named in the source column, and only the EQi12 row is our own instrumented reading. Ranges are wider than a single test because chassis, drives, workload and power settings all move the number.
| Class | Idle (W) | Load (W) | What the figure is | Source |
|---|---|---|---|---|
| Beelink EQi12 (i3-1215U), Windows | 12–14 | 37 | Measured at the wall by us, 7 states, whole system | EQi12 power-consumption report |
| Fanless Intel N100 mini PC | 6–10 | 20–25 | Published measurement | Selfhosting — silent/fanless home server guide |
| Fanless case + i5-12400T | 15–25 | 50–65 | Published measurement | Selfhosting — silent/fanless home server guide |
| Semi-passive build with SSD | 20–30 | 60–80 | Published measurement | Selfhosting — silent/fanless home server guide |
| Typical tower server | 40–80 | 100–200 | Published measurement | Selfhosting — silent/fanless home server guide |
| Low-cost N100/N150 network appliance | 13 → 10 after enabling PCIe ASPM | — | Owner report, one model | Netgate community forum mini PC thread |
| 2-bay NAS with spinning drives | 25–40 | — | Typical range quoted for this class | Home server power cost guide |
| Intel Core i3-1215U, processor ratings | — | 15 base / 55 max turbo | Manufacturer power rating, not consumption | Intel ARK product specification |
| Intel N100, processor rating | — | 6 | Manufacturer power rating, not consumption | Intel ARK, referenced in Intel community PL1/PL2 thread |
The two bottom rows exist to stop a common mistake: a processor's rated power is a thermal design figure that the firmware uses to configure power limits. It is not what the machine draws at the socket. A "6 W" N100 system routinely idles in the low teens at the wall, because the rating excludes memory, storage, networking, the power supply's conversion loss and the chipset.
The spread-between sources is a warning as much as a reference: the same class of machine can be reported anywhere from 6 W to 25 W at idle depending on ASPM settings, attached devices and firmware. Measure your own unit rather than picking the flattering number.
Measure three states before estimating cost
- Let the operating system settle, then record an idle range rather than one instant.
- Run the workload that matters—backup, transcode, compile, or database job—and record its typical draw and duration.
- Measure sleep or shutdown separately if the server will not remain online all day.
If you are measuring for a Mini PC home server, also read the EQi12 build guide to see the full host setup that sits behind those power numbers—BIOS power settings, sleep behavior, and attached peripherals all move the wall draw.
What this estimate does not include
Time-of-use tariffs, taxes, UPS conversion loss, external switches, routers, displays, and seasonal electricity pricing are outside the formula. If a UPS powers several devices, measure the complete protected group or allocate its loss conservatively. Pair this calculator with the media storage planner if you are deciding whether a always-on media server fits your monthly budget alongside its storage needs.
Frequently asked questions
How much electricity does a typical home server use per year?
A mini PC–based home server idling around 10–15W with two hours of daily load typically draws 100–160kWh per year, or roughly $15–$25 at US residential rates. A traditional tower NAS with multiple spinning drives often uses 2–3× as much.
Is a mini PC cheaper to run than a NAS?
Usually yes for light workloads. A two-bay NAS with spinning drives idles around 25–40W, while a fanless mini PC with a single USB SSD can idle below 10W. The gap narrows when the NAS drives spin down or the mini PC runs heavy transcoding.
Does putting the server to sleep save meaningful electricity?
Sleep (S3) typically draws 1–3W, so a server that sleeps 8 hours per night saves roughly 0.5–1kWh per month. For a 24/7 server the savings are small; for a part-time server the savings scale with sleep duration.
Should I use CPU package power or wall power for the estimate?
Always use wall power if possible. CPU package power excludes memory, storage, networking, USB devices, and PSU conversion losses, which together can add 20–40% to the measured draw.
How do I measure my server's actual power consumption?
Use a plug-in power meter (Kill-A-Watt, TP-Link Tapo P110, or similar) between the wall outlet and the server's power adapter. Record readings at idle, under your typical workload, and during any background jobs.
What electricity rate should I enter?
Use your effective rate including all fixed charges and taxes, not just the per-kWh energy charge. In the US the national average is about $0.16/kWh; in the EU it varies from €0.15 to €0.40+ depending on country and contract.
Can I use the CPU's rated wattage instead of measuring?
No. A processor rating such as the 15 W base and 55 W maximum turbo power published for the Intel Core i3-1215U is a power-limit figure the firmware configures against, not a wall reading. It excludes memory, storage, networking, chipset and the power supply's conversion loss. That is why a mini PC built around a 6 W-rated N100 chip can still idle in the low teens at the socket.