The honest truth about smart home power usage
What a local-first Home Assistant setup actually draws, where the watts hide, and how to pick efficient gear without chasing meaningless savings.
On this page
Building a local-first smart home with Home Assistant keeps your control and your data at home. It also means you’re running hardware 24/7, so it’s fair to ask what that costs. The honest answer: a sensible local-first setup draws very little, the biggest variable is what you run Home Assistant on, and chasing the absolute lowest wattage usually costs you more in reliability than you save on the meter.
Know your power draw: the hub matters most
Your always-on hub is the thing that runs every hour of every day, so its idle draw dominates. A Raspberry Pi 4 running Home Assistant idles in roughly the 3–5 W range; purpose-built appliances like the Home Assistant Green and the Home Assistant Yellow (discontinued in October 2025, but fine if you already own one) are in the same low-single-digit-watt territory. Over a year, a few watts is a few kWh—pennies to low single-digit dollars depending on your rate.
Step up to a small x86 box (a mini PC or NUC-class machine) and idle draw is typically more like 8–20 W, climbing higher under load. That’s still modest, and the extra headroom is often worth it if you run add-ons like Frigate for camera object detection. The point isn’t that one is “right”—it’s that the hub, not your sensors, is where the steady watts live.
Radios sip power. A Zigbee coordinator like the SONOFF Zigbee 3.0 USB Dongle Plus or a Z-Wave controller such as the Z-Wave.Me RaZberry 7 Pro draws a fraction of a watt—negligible next to the host.
A note on ecosystems: if you also use Apple Home, the Apple HomePod mini or Apple HomePod 2nd Gen act as Thread border routers and speakers, but they’re full-time appliances with speaker-class draw, not a low-power “hub” replacement for Home Assistant. They’re not Z-Wave controllers, either—different job entirely.
Battery vs. mains sensors: the real efficiency question
For most sensors, the meaningful question isn’t grid watts—it’s battery life. Battery-powered low-power sensors like the Aqara Door/Window Sensor P2 and Aqara Motion Sensor P2 (both Matter-over-Thread) draw so little that they last months to years on a coin cell; they cost you nothing on the meter. Their “efficiency” is about how often they have to wake and report.
Mains-powered devices are where standby draw adds up across many units. Always-on Wi-Fi plugs, relays, and bulbs each pull a small amount continuously. None of it is large per device, but it scales with device count, so favor efficient gear where you’re deploying many.
Choose efficient devices without overthinking it
For controllable outlets, a metering plug—whether a Zigbee/Z-Wave plug or a Wi-Fi unit like the Shelly Plug S—lets you measure what appliances actually use, which is more valuable than guessing. For lighting, LED smart bulbs such as the Philips Hue White A19 or LIFX A19 Color are already low-wattage when on; the bigger savings come from automating them off, not from spec-sheet hunting.
For locks, be accurate about what you are buying. The Yale Assure Lock 2 is battery-powered and offered with swappable connectivity modules (Wi-Fi, Z-Wave, Matter), so it does not add to your grid draw at all. The Yale Assure Lock 2 Plus runs on batteries too, but it is the Apple Home Key model and does not take those modules—Home Assistant reaches it locally over Bluetooth instead. The Kwikset Halo Touch is a Wi-Fi lock (2.4 GHz, no hub required)—not a Z-Wave lock, and not something a HomePod controls. Pick the connectivity that matches your stack; the battery does the work either way.
For a Matter relay, the Sonoff MINIR4M is a low-draw Wi-Fi Matter device. As with any always-on Wi-Fi unit, its standby cost is small but real per device.
Don’t over-optimize for power saving
Here’s the tradeoff worth internalizing: the watts you’d save by buying the cheapest, lowest-power gear are usually trivial compared to the cost of an unreliable system. A flaky sensor that misses events or a hub that can’t keep up will cost you far more in frustration than the wattage difference ever saves in dollars.
Run Home Assistant on hardware sized for what you actually do. If you’re doing local camera AI, a mini PC’s extra idle watts buy you capability a Pi can’t match. If you’re running lights and sensors, a Green or a Pi is plenty. Either way, the radios and battery sensors are rounding error.
Bottom line: be realistic
A local-first Home Assistant setup is inherently efficient—the steady draw is your always-on host (a few watts on a Pi/Green, more on an x86 box), and almost everything else is negligible or battery-powered. Use a metering plug to find your actual energy hogs (usually appliances, not smart-home gear), automate things off when they’re not needed, and right-size your hub for your workload.
Quick verdict: A local-first smart home with Home Assistant sips power; the host hardware is the only meaningful 24/7 draw, and battery/mesh sensors barely register. Measure with a metering plug, automate aggressively, and prioritize reliability over chasing the last watt—the savings are real but small, and the value is in control and dependability.