Home AI server power consumption
Last updated
Home AI server power consumption is mostly idle time. Olares lists 30 W in standby for Olares One and GPU 175 W plus CPU 55 W at combined full load. Multiply watts by hours, then by your local electricity rate.
Olares One → Home AI server → Ollama · Open WebUI · Immich → Family AI that stays at home
Why does idle power matter most?
Because a home server waits far more than it works. Families send questions in short bursts, and between them the box sits idle, often for most of the day. So the standby figure drives the yearly bill. Peak power still matters for heat, noise and long jobs such as indexing a photo library or generating video.
How much power do common home AI machines draw?
Makers publish different kinds of figures, which makes direct comparison hard. Olares' FAQ gives a standby figure and component figures at full load. Apple gives a maximum continuous power rating. Most others give only a power supply rating, which is a ceiling. The table keeps each figure next to what it measures, as read on 12 and 13 November 2026.
| Machine | Figure | What it measures |
|---|---|---|
| Olares One | 30 W | Standby (Olares FAQ) |
| Olares One | GPU 175 W + CPU 55 W | Combined maximum load (Olares FAQ) |
| Olares One | 330 W | Power adapter rating |
| Mac mini M5 Pro | 155 W | Maximum continuous power (Apple) |
| NVIDIA DGX Spark | 240 W; GB10 TDP 140 W | Power supply rating (NVIDIA) |
| Beelink GTR9 Pro | 230 W; up to 180 W running an AI model on Ubuntu | Power supply; Beelink's own note |
| Mac Studio | 480 W | Maximum continuous power (Apple) |
| Used RTX 3090 tower | about 515 W | Sustained full load (SpecPicks estimate) |
| RTX 5090 desktop card | 575 W | Card alone, total graphics power (NVIDIA) |
Apart from Olares' standby figure, our sources give no idle figures, so measure your own.
How do you work out home AI server power consumption and cost?
Multiply watts by hours, divide by 1,000 to get kilowatt-hours, then multiply by the rate on your bill. Measure the watts with a plug-in meter for a week if you can, because maker figures are ceilings or lab values. Count idle hours and heavy-use hours separately, since they draw very different power.
Take Olares' standby figure of 30 W. Over 24 hours that is 0.72 kWh, or about 263 kWh over a year. Multiply by your own rate per kWh, then add the heavy-use hours at the higher figure.
How can you cut the bill?
Size the hardware to the models you actually use, and let the box return to idle quickly. A model that fits in VRAM finishes sooner than one spilling into system memory. A 575 W desktop card bought for a family chat assistant spends nearly all its life far below its rating, and you pay for the headroom in heat.
Olares One has a Silent Mode that lowers GPU power, but Olares' pages give different figures for it, so measure it yourself. Wake-on-LAN lets you wake a DIY server or Olares One over wired Ethernet when someone needs it. The quiet box guide covers noise, and the budget guide lists lower-power machines.
Questions & answers
How much electricity does a home AI server use?
Mostly its idle draw. Olares lists 30 W in standby for Olares One, which is 0.72 kWh a day if left on around the clock. Multiply by your local rate to get the cost.
Does a used RTX 3090 build use a lot of power?
Under load, yes. SpecPicks estimates its build at about 515 W at sustained full load. Idle draw is lower; measure it with a plug-in meter.
Sources for this entry
- Olares One FAQ (Olares docs) (Company-published)
- Olares One: maker's page (Maker-published)
- Apple Mac mini (M5 Pro, 64GB): maker's page (Maker-published)
- NVIDIA DGX Spark: maker's page (Maker-published)
- Beelink GTR9 Pro (Ryzen AI Max+ 395, 128GB): maker's page (Maker-published)
- Apple Mac Studio (M5 Ultra): maker's page (Maker-published)
- Used RTX 3090 tower build: maker's page (Maker-published)
- NVIDIA GeForce RTX 5090 (desktop card): maker's page (Maker-published)
Continue this reading path
How the related websites connect
Cite this entry
Best Home AI Server. “Home AI server power consumption.” Content reviewed 13 November 2026.
https://besthomeaiserver.com/home-ai-server-power-consumption/