Quick Summary (TL;DR): The cheapest 24/7 NASAliExpress price or home server is not necessarily the model with the lowest processor TDP. What matters is average AC power measured at the wall, including installed drives, network equipment and any UPS losses. A continuous 25 W load consumes 219 kWh a year; at an illustrative electricity rate of £0.25/kWh, that is £54.75 annually before any fixed standing charge. Each extra watt running all year adds 8.76 kWh. Measure a representative week, use the variable per-kWh price from your own tariff, and compare whole-system power rather than chassis specifications. For light Home Assistant and containers, a low-idle mini PCAliExpress price plus appropriately sized storage can be efficient; for file serving, a compact NASAliExpress price may avoid the power and maintenance overhead of a second computer. Do not sacrifice backups, cooling, reliable shutdown or required performance to save a few watts. The calculations below are reproducible examples, not lab measurements of particular products.
Calculate NAS Electricity Costs in Two Steps
Power is the rate at which a device uses electricity, measured in watts (W). Energy is power accumulated over time, measured in kilowatt-hours (kWh). Your electricity bill charges for energy, not the number printed on a power adaptor.
Energy (kWh) = average AC power (W) × running hours ÷ 1,000
Annual cost = energy (kWh) × variable electricity price per kWh
24/7 annual energy = average watts × 8.76
A 25 W system left on 24 hours a day for 365 days uses 25 × 8.76 = 219 kWh/year. At a sample rate of £0.25/kWh, that costs 219 × £0.25 = £54.75/year. This example deliberately excludes a household’s fixed standing charge: it normally applies whether the NASAliExpress price is present or not. If your tariff includes time-of-use pricing, apply the correct price to the energy used in each period rather than multiplying everything by one flat rate.
Quick rule: At £0.25/kWh, every continuous 10 W costs about £21.90 per year. At €0.20/kWh, the same 10 W costs €17.52 per year. Neither figure is a claim about current UK or Greek tariffs; substitute your actual import rate, including applicable taxes and variable charges.
24/7 NAS and Home-Server Running-Cost Table
The figures below are calculated power scenarios, not tested product benchmarks. They let you compare an existing machine with a potential replacement without pretending every N100 mini PCAliExpress price or four-bay NAS consumes the same amount.
| Average wall power | Energy per year | Cost/year at £0.25/kWh | Cost/year at €0.20/kWh |
|---|---|---|---|
| 10 W | 87.6 kWh | £21.90 | €17.52 |
| 15 W | 131.4 kWh | £32.85 | €26.28 |
| 25 W | 219.0 kWh | £54.75 | €43.80 |
| 40 W | 350.4 kWh | £87.60 | €70.08 |
| 65 W | 569.4 kWh | £142.35 | €113.88 |
| 100 W | 876.0 kWh | £219.00 | €175.20 |
| 150 W | 1,314.0 kWh | £328.50 | €262.80 |
| 250 W | 2,190.0 kWh | £547.50 | €438.00 |
These are annualised values assuming constant average draw. A server that briefly reaches 150 W while transcoding but idles at 20 W most of the day is not a 150 W continuous load. Conversely, a NAS that advertises 12 W during disk hibernation may spend little time in that state when cameras, containers or background tasks keep the disks active.
Measure the Complete System, Not the CPU TDP
There are at least four different numbers commonly confused in NAS buying guides:
| Number on a specification | What it actually describes | Suitable for annual-cost estimates? |
|---|---|---|
| CPU TDP or processor base power | A processor thermal/power design parameter under specified conditions | No — excludes the rest of the machine |
| Power adaptor rating, e.g. 65 W or 120 W | Maximum output capability of the adaptor | No — not typical input draw |
| UPS VA rating | Apparent-power capacity of the UPS | No — not energy consumed |
| Average AC watts or measured kWh | Electricity taken from the socket over representative operation | Yes — use this for the bill |
For example, Intel lists the Processor N100 with a 6 W TDP. That does not mean an N100 mini PC consumes 6 W at the wall: RAM, SSD, Ethernet controllers, motherboard regulators, USB devices, fan, firmware settings and adaptor losses all matter. Two N100 machines can have different idle power and very different power draw once storage and peripherals are attached. See the official Intel N100 specifications.
A NAS manufacturer’s power figures are more useful because they may include the enclosure and drives, but read the test conditions. Synology lists the DS925+ at 37.91 W in its access test and 12.33 W during HDD hibernation. These are two defined manufacturer test states, not a promise that every real installation will average one of those values. The number and model of installed drives, package activity and configuration can change results. Consult the DS925+ technical specifications and the testing notes before comparing it with another brand.
How to Measure Your NAS Power Properly
The most practical starting point is a plug-in energy meter that measures kWh, or a reputable energy-monitoring smart plug with an appropriate electrical rating and accessible readings. A real kWh counter is more useful than a screen showing only instantaneous watts.
- Decide your measurement boundary. Measure the NAS alone, or measure the complete group of NAS, mini PC, router, switch and UPS. Record which components are included so you do not compare unlike systems.
- Start with a safe, rated meter. Check local socket type, voltage, maximum current, earthing and manufacturer instructions. Do not use a low-quality adaptor or power strip simply to add monitoring. Avoid interrupting a running NAS to insert a meter without first performing a safe shutdown.
- Record baseline idle draw. Wait until boot, RAID checks, indexing and startup services settle. Note installed disks, active VMs, network links and whether HDDs are spinning.
- Measure normal activity. Include overnight backups, media scans, family file access, container updates and any camera recording. For most homes, a full seven days is a useful minimum; two to four weeks better captures irregular tasks.
- Read the accumulated kWh. Divide by the elapsed hours to find average watts:
average W = kWh × 1,000 ÷ hours. - Repeat after one change at a time. Keep the workload and measurement period comparable. A five-minute idle snapshot cannot validate a claim about annual savings.
For a NAS connected to a UPS, take two readings if possible: upstream of the UPS for actual grid cost, and at the device or backed-up outlet for the load. The difference includes UPS conversion and charging overhead. Never put a smart plug in a location that could remotely cut power to your storage by accident; disable switch controls where the product permits it, and use monitoring-only devices when possible.
Example: A One-Week Meter Reading
Suppose the energy meter records 4.20 kWh over exactly 168 hours. The average draw is 4.20 × 1,000 ÷ 168 = 25 W. At a flat £0.25/kWh, the estimated annual energy is 4.20 × 365 ÷ 7 = 219 kWh, costing £54.75/year.
That estimate is only representative if the week resembles the rest of the year. If your NAS also runs an eight-hour monthly full-backup job, a yearly media transcoding workload or a seasonal camera schedule, include those separately or monitor longer. A system used for occasional work is better modelled by its measured on-hours and off-hours than by assuming it runs flat out continuously.
A Worked Example: NAS, Mini PC, Network and UPS
A home lab is often more than one box. The following values are illustrative measured averages for a hypothetical setup, not specifications for real devices:
| Component | Example average wall-equivalent power |
|---|---|
| Four-bay NAS including installed disks | 40 W |
| Mini PC running Home Assistant and containers | 12 W |
| Small Ethernet switch | 8 W |
| Router | 7 W |
| Additional UPS overhead, measured at the wall | 6 W |
| Total | 73 W |
At 73 W continuously, this complete setup consumes 639.48 kWh/year. At £0.25/kWh the example bill is £159.87/year; at €0.20/kWh it is €127.90/year. If you measured the whole system upstream of the UPS and obtained 73 W, do not add the UPS overhead again. Likewise, a NAS’s internal drive power is already included in a whole-NAS wall measurement: adding separate HDD watts would double-count it.
This is why a purported 10 W mini PC may not lead to a 10 W home-lab installation. External disk docks, 10GbE adapters, USB peripherals and a separate switch can erode the apparent saving. A combined NAS-and-applications box might use more power than either component alone yet less than two separate machines together.
Choosing Efficient Hardware: Where the Real Differences Come From
A Two-Bay NAS for Backups and Shared Files
If the main job is file sharing, snapshots and scheduled backups, a compact two-bay NAS is usually easier to size and maintain than a large repurposed desktop. Look for documented wall-power measurements, supported disk choices, sensible fan control, sleep behaviour and a long software-support horizon. The drive configuration matters: two disks can have a different idle and active draw from four disks, even when the CPU is identical.
A small NAS is not automatically the lowest-power choice for Home Assistant or media transcoding. Check whether the model can run your required containers, whether it supports the relevant hardware video codecs and whether you need to leave a second mini PC running anyway. Compare options in our best two-bay NAS for home backups and DIY NAS versus prebuilt NAS guides.
A Low-Idle Mini PC for Home Assistant, Docker and Light VMs
A modern small computer can make sense when you need always-on automation, MQTT, ESPHome, media metadata services or several modest containers. Prioritise measured whole-machine idle consumption, enough RAM, dependable storage, supported virtualisation and the number of network ports actually required. A low-power CPU cannot compensate for an inefficient motherboard or permanently powered USB accessories.
Avoid buying an N305 or a larger desktop purely because it is faster on paper if your workload is almost always idle. Conversely, a system that runs at high utilisation and struggles with video decoding or several VMs can be a poor fit even when its idle figure is attractive. The relevant question is energy per completed task plus idle hours, not just peak watts. Our N100 vs N150 vs N305 comparison and mini PC plus NAS versus NAS-only home server explain the workload trade-offs.
A DIY Tower or Used Office PC for Expandability
A larger system offers PCIe slots, more drive bays, more RAM and often easier repair. Its power use depends on the actual platform, installed components, BIOS settings and disks. Do not assume that every old office PC is wasteful or that every purpose-built NAS is efficient. Measure the current machine before replacing it.
The hidden power costs of a DIY build can include a dedicated graphics card, several spinning disks, add-in network cards, high-speed fans and a PSU that is inefficient at very low load. A tower can still be the sensible choice if it replaces two or three other boxes, handles your required virtual machines or avoids buying a separate NAS enclosure.
Raspberry Pi and Other Single-Board Computers
A small board may draw little power, but total consumption includes its power supply, USB storage, SATA bridge, powered hub, Ethernet switch and any cooling. It may also introduce cabling and recovery complexity. For a small Home Assistant deployment or lightweight network service it can be appropriate; for a growing storage server with multiple disks, check the entire architecture and reliability requirements first.
Hard Drives, SSDs and the Hibernation Trap
Mechanical drives use power to spin platters and move heads; SSDs have different idle and active characteristics. However, SSD is not a synonym for lower whole-system energy in every workload. Capacity, cost, controller idle states, write endurance and network bottlenecks still matter.
A four-bay NAS with three or four active HDDs may use noticeably more power than the same chassis with fewer drives, but reducing disk count also changes redundancy, usable capacity and rebuild exposure. Buying much larger disks solely to save a small number of watts is rarely an obvious financial win. For disk-specific decisions, see HDD vs SATA SSD vs NVMe for NAS and CMR vs SMR for NAS.
HDD hibernation is conditional. It may not occur when surveillance cameras write continuously, Plex or Jellyfin refresh metadata, Docker containers write logs, a torrent client seeds files, or a monitoring tool polls storage too frequently. Even when disks sleep, the motherboard, RAM, network interface and PSU continue drawing power. Aggressive sleep intervals can also create repeated spin-up cycles and noticeable access delays; evaluate the actual drive behaviour and vendor guidance instead of assuming shorter is always better.
A safer approach is to inspect what wakes the disks, move high-frequency application writes to appropriate SSD storage if supported, and measure whether the change really lowers energy. Do not disable essential SMART checks, snapshots, RAID scrubbing or backup verification simply to keep a disk asleep.
Network Equipment and UPS Losses Are Part of the Bill
A NAS may be efficient while the infrastructure around it is not. A multi-gigabit switch, several PoE cameras, wireless access points, a 10GbE network adapter and a UPS all contribute to the electricity bill. If the NAS serves clients over 1GbE, adding a 10GbE switch may bring little practical benefit; if multiple editors regularly move large media files, higher network power may be justified.
When measuring a PoE switch, remember that the switch’s wall consumption includes the electricity delivered to attached PoE devices. Do not add the cameras’ rated PoE draw again when estimating the same circuit. If you need to compare switches, keep port count, link speeds, PoE loads and cooling requirements equivalent. Our 2.5GbE switch guide focuses on useful home-lab configurations.
A UPS protects data during outages but consumes some electricity itself. Its idle losses depend on design, loading, battery condition and operating mode. A UPS’s 600 VA or 1,000 VA label does not indicate how many watts it wastes. Use an upstream meter for real electricity cost and size the UPS for load, runtime and graceful shutdown. Read our UPS VA versus watts sizing guide and NUT shutdown guide.
Reduce Power Consumption Without Damaging Reliability
Work through the biggest opportunities first, and record the kWh change after each one.
- Consolidate duplicate always-on services. If two machines run trivial workloads, check whether one supported host can run both. Preserve isolation and backups where needed.
- Turn off unused equipment. Old switches, spare access points, idle USB disk docks and test servers can be surprisingly expensive if left on 24/7.
- Review BIOS/UEFI power settings. CPU idle states and platform power management can matter, but some devices have stability issues with particular settings. Record the original configuration and test carefully.
- Use appropriate OS power governors. Prefer supported balanced/idle policies; avoid disabling essential services to chase an artificially low number.
- Remove unnecessary peripherals. Unused USB radios, RGB accessories, add-in GPUs and extra NICs all deserve scrutiny.
- Schedule genuinely intermittent workloads. A backup-only destination may not need to run every minute, provided it powers up reliably for backup and remains reachable when required.
- Investigate constant disk writes. Look for verbose logs, databases, containers, indexing and camera recording that defeat HDD hibernation.
- Keep cooling functional. Lower fan speed is not worthwhile if it causes hot drives, thermal throttling or shorter hardware life.
- Avoid blind BIOS tweaks. ASPM, deep C-states, wake-on-LAN and disk standby settings can affect device compatibility and recovery from outages. Change one variable and test.
- Recheck UPS behaviour. A lower idle draw is useless if your storage no longer receives a shutdown signal or fails to recover after mains power returns.
A device that spends much of the day sleeping may save energy, but only if it still meets availability requirements. Home Assistant automations, family file shares and remote-access services generally need an always-on host; a cold backup server may not.
Track NAS Electricity in Home Assistant
Home Assistant can display individual-device consumption if your meter or smart plug provides suitable energy entities. Its official individual-device energy documentation describes compatible monitoring approaches, and the Energy dashboard documentation explains the data model.
For a simple setup, use a supported energy-monitoring plug or meter on the NAS’s supply, then add its kWh energy sensor as an individual device in the Energy dashboard. A sensor that only reports watts is instantaneous power, not energy. Home Assistant can derive kWh with the Riemann-sum integration, but check the sensor’s reporting interval and whether it already exposes a native accumulated-energy reading.
A valid energy sensor generally needs device_class: energy, a suitable state_class such as total_increasing, and units such as kWh. A power sensor normally uses device_class: power, state_class: measurement, and W or kW; see the official energy FAQ. When a parent meter measures an entire power strip and child meters measure devices within it, configure the upstream relationship or exclude duplicates rather than adding every number together.
Recommended dashboard: show today’s kWh, the last seven days, the previous month’s kWh, estimated cost at your tariff, and the highest-power recurring workload. Alert on abnormal increases only after you understand the baseline. A 30 W rise during a RAID rebuild is not automatically a fault; a 30 W increase that remains for days after a software update may deserve investigation.
A Simple Spreadsheet or Notebook Log
| Date/period | Devices included | Meter start/end (kWh) | Hours | Average W | Cost at your tariff | Change made |
|---|---|---|---|---|---|---|
| Week 1 | NAS + disks | 110.0 → 114.2 | 168 | 25.0 | £1.05 at £0.25/kWh | Baseline |
| Week 2 | Same NAS + disks | 114.2 → 118.0 | 168 | 22.6 | £0.95 at £0.25/kWh | Removed unused service |
The second row represents a hypothetical 0.4 kWh/week reduction. That is about 20.9 kWh/year, or £5.21/year at the example tariff. Such a small saving is useful if free, but it would not justify expensive replacement hardware by itself.
Should You Replace an Older Server to Save Electricity?
Use incremental cost, not only annual power draw. Suppose your current server averages 45 W and a replacement would average 30 W doing the same job. The 15 W difference saves 15 × 8.76 = 131.4 kWh/year. At £0.25/kWh that is £32.85/year. If the replacement costs £150 more than keeping the current system, the simple electricity-only payback is about 4.6 years (£150 ÷ £32.85).
That calculation ignores hardware failure risk, resale value, maintenance, software support, performance, storage expansion, repairs and environmental costs. It also assumes the new system really averages 30 W with the same drives and workload. If you would have to buy the replacement anyway because the old machine cannot run your services reliably, power savings become one useful factor rather than the whole justification.
| Proposed change | Energy impact | Other costs and constraints |
|---|---|---|
| Replace a working 45 W server with a 30 W unit | Saves 131.4 kWh/year if always on | Purchase cost, migration, backup, RAM and storage |
| Shut down an unused 20 W test box 20 hours/day | Saves 146 kWh/year | Wake/restart process; must not host essential services |
| Add a 12 W mini PC to a 40 W NAS | Adds 105.1 kWh/year if both run continuously | May improve app performance and separation |
| Consolidate a 12 W mini PC into an existing NAS | Could save up to 105.1 kWh/year | NAS CPU/RAM capacity, container support, availability |
The savings in the table are arithmetic scenarios, not promises about any brand. For the shut-down example, 20 W × 20 h/day × 365 ÷ 1,000 = 146 kWh/year; the device is still assumed to run four hours each day. For consolidation, the NAS may draw more power once it takes over the mini PC’s workload, so the actual net saving could be smaller.
Which Architecture Is Most Efficient for Your Workload?
| Main workload | Sensible starting point | What to verify before buying |
|---|---|---|
| Family files and automated backups | Compact two- or four-bay NAS | Installed-drive draw, backup targets, noise, vendor support |
| Home Assistant, MQTT and a few containers | Low-idle mini PC or supported NAS apps | Full wall power, RAM, storage endurance, USB passthrough |
| Plex/Jellyfin and family storage | NAS with suitable media support, or NAS + efficient mini PC | Actual codec support, transcoding demand, idle power |
| Several VMs and development environments | Efficient mini PC or DIY server sized for RAM/PCIe | Workload performance, NICs, idle power, upgrade headroom |
| Surveillance recording | NAS/NVR built for continuous writes | Drive activity, PoE cameras, retention, backup strategy |
| Weekly off-site or cold backup | Scheduled backup NAS or rotating external drive | Automatic wake, verification, offline separation, recovery |
There is no universal winner. A low-power device that cannot meet your needs is false economy; a powerful server that idles at a high draw to host two trivial services may be unnecessary. The best setup is the lowest measured energy cost that still satisfies reliability, capacity, speed and recovery requirements.
Frequently Asked Questions
How much does a 30 W NAS cost to run 24/7?
It consumes 30 × 8.76 = 262.8 kWh/year. Multiply by your actual electricity rate: £65.70/year at £0.25/kWh or €52.56/year at €0.20/kWh. Add only incremental charges attributable to the device, not the household standing charge that would exist anyway.
Is a 6 W Intel N100 mini PC really a 6 W computer?
No. Intel’s 6 W figure is the processor TDP, not the complete system’s AC consumption. The motherboard, memory, SSD, NICs, power supply and workload determine the wall reading. Measure the exact configured machine.
Does a NAS consume less electricity when its disks hibernate?
Usually it can, but the actual saving depends on how many drives enter standby, the enclosure’s remaining draw and whether services repeatedly wake the disks. Manufacturer hibernation figures are measured under defined conditions. For always-on recording or busy containers, assume hibernation may be rare until proven otherwise.
Does a UPS increase electricity use?
A UPS normally introduces some conversion, control and battery-maintenance overhead. The amount depends on the model, loading and operating mode. Measure upstream of the UPS to include it in the bill, and never use VA capacity as a substitute for measured watts.
Is SSD storage always more power-efficient than HDD storage?
Not necessarily on a per-terabyte or whole-system basis. Compare the actual drive models, number of drives, idle states, workload, cost and required usable capacity. SSDs can be excellent for application data and frequent small writes; large HDD arrays can remain economical for bulk storage.
Should I power off my NAS overnight?
Only if no backups, remote access, automations, surveillance, family services or maintenance tasks need it. Scheduled shutdown can save energy, but verify clean filesystem shutdown, startup behaviour, access windows and backup completion. A storage server that is supposed to be available 24/7 should be sized efficiently rather than repeatedly forced off.
Can Home Assistant calculate my NAS’s electricity bill?
Yes, if you provide suitable energy data and an appropriate tariff model. A measured kWh sensor is preferable to guessing from processor TDP. Check time-of-use rates, taxes and whether your dashboard already includes the same consumption via a parent meter.
Is a more efficient new NAS worth buying just for electricity savings?
Calculate the measured watt difference, annual kWh saved and purchase premium. A 15 W reduction is £32.85/year at £0.25/kWh; replacing an otherwise adequate server purely to achieve that may take years to pay back. Reliability, software updates and capacity may be stronger reasons to upgrade.
Related NAS and Home-Server Guides
- DIY NAS vs Prebuilt NAS: Complete Build and Running Costs
- Intel N100 vs N150 vs N305 for Home Assistant
- Mini PC Plus NAS vs a NAS-Only Home Server
- Best 2-Bay NAS for Home Backups
- UPS VA vs Watts: Sizing Backup Power for NAS, Router and Server
- A 3-2-1 NAS Backup Plan: USB Drive, Second NAS and Cloud
Manufacturer Documentation and Further Reading
- Intel — Processor N100 specifications: processor TDP, not system wall power.
- Synology — DS925+ technical specifications: manufacturer access and HDD-hibernation power figures and test notes.
- Home Assistant — Home energy management: energy and device monitoring concepts.
- Home Assistant — Individual device energy: smart-plug measurement and parent/child device accounting.
- Home Assistant — Energy FAQ: sensor metadata and troubleshooting.
- Seagate — IronWolf product manual: definitions of drive idle, operating, standby and spin-up measurements; use the manual for your exact model.
Bottom line: Buy and configure around the workload, then measure kWh at the wall. A week of actual readings and your own tariff are more informative than a CPU’s TDP, a PSU’s maximum wattage or an optimistic hibernation specification. Preserve your backup and shutdown safeguards while making changes.