A home network rack holds servers, storage drives, and a switch beside two security cameras.
Home lab owners monitor CPU temperatures, RAM pressure and disk I/O in obsessive detail, but few of us check what the whole rack pulls from the wall. XDA Developers writer Jasmine Mannan did exactly that. She put a TP-Link Tapo P110 energy-monitoring plug between the wall outlet and her lab's main UPS for a week. The result was a 24/7 baseline of about 72 W. That is roughly 12.1 kWh a week, or an estimated $114 a year at the current US national residential average.

That is not a scary number. It is a useful one, because the method is cheap and any always-on lab can repeat it.

A home network rack holds servers, storage drives, and a switch beside two security cameras. What was measured​

The plug sat between the wall receptacle and the lab's primary uninterruptible power supply. Everything downstream was therefore measured as one lump. The lab includes:

  • A GEEKOM IT13 mini PC running containers
  • A two-bay UGREEN DH2300 NAS with two HDDs
  • A UGREEN HomeAgent HA100 with two HDDs, acting as a 24/7 NVR
  • A managed PoE switch powering two PoE cameras

The reported draw ranged from about 62 W to 80 W and averaged 72 W. The lab was quiet for roughly 90% of the week. It never reached a deep-idle state, because the NVR kept writing to spinning disks and the cameras drew PoE power around the clock.

The arithmetic checks out. 72 W × 168 hours ÷ 1,000 is 12.096 kWh, which matches the reported 12.1 kWh.

Estimates, not meter readings​

The article gives a per-device breakdown, but the author says she did not meter the devices separately. Pulling live storage offline mid-week would have ruined the continuous baseline. The figures come from manufacturer specs and outside benchmarks, so treat them as informed estimates:

DeviceEstimated draw
GEEKOM IT13 mini PC (containers running)about 12–13 W
UGREEN DH2300, two HDDs, idleabout 12 W
UGREEN HomeAgent HA100, two HDDs, NVR dutyabout 20–26 W
Managed PoE switch (powering two cameras)about 25 W

UGREEN's own published DH2300 test is a loose cross-check. It lists about 13.7 W during drive access and about 3.3 W during hibernation. That was measured in UGREEN's lab with a single Seagate 2 TB drive. It does not validate the author's two-drive estimate, because drives, services and settings all change the result.

The author also noticed that total draw stepped up by about 4 W every evening at sunset and held until sunrise. She attributes this to the cameras' infrared night-vision LEDs. That is plausible, but a single upstream meter cannot prove which device caused the step. It is a good hypothesis, not a finding.

One more caveat from general electrical knowledge. A wall-side meter on a UPS also captures the UPS's own conversion losses and battery-charging overhead. That means the 72 W is the cost of the whole protected stack, not just the IT gear.

From watts to dollars​

The author annualised the week by multiplying by 52.14, giving about 630.7 kWh a year. That is a projection from one week. Seasonal load, maintenance, new containers, and outages could all shift the real figure.

For price, she used the US Energy Information Administration's October 2026 Electric Power Monthly, which covers August 2026. EIA's table shows a national residential average of 18.16 cents per kWh. That compares with 17.57 cents in August 2025, an increase of about 3.4%. EIA derives these numbers from retail revenue per kWh, so they approximate prices rather than reflecting any one customer's tariff.

At those prices, the same 630.7 kWh a year works out as follows:

Region (August 2026 residential average)Cents per kWhApprox. annual cost
Louisiana12.87about $81
North Dakota13.31about $84
US average18.16about $114.50 (roughly $9.50 a month)
California29.75about $188
Hawaii47.18about $298

Your own bill will differ. Fixed charges, time-of-use rates, tiers, taxes and riders all change the picture. The best number to use is your marginal per-kWh rate from your own statement.

Pick the right plug​

The article's hardware advice is the most transferable part. Smart plugs have voltage, frequency and current ratings, and these vary by region and model. TP-Link's US specification page for the P110 V1 lists:

  • 100–125 V, 50/60 Hz
  • A maximum load of 15 A and 1,800 W for general or resistive use
  • 1/6 HP for motors
  • Self-consumption of 1.64 W with the relay on and 0.64 W with it off

A 72 W lab is nowhere near those limits. A rack with a beefy server or a space heater under the desk is another matter. The author cites the US National Electrical Code's 80% continuous-load guideline, which caps a 15 A circuit at about 12 A of sustained draw. She also notes that a power strip does not raise that ceiling. Check the rating of your specific plug variant and your circuit, and follow the equipment instructions.

Collecting the data​

The author used the Tapo app's bar charts, which are fine for a weekly total. For finer detail, she points to local polling through Home Assistant. Home Assistant's official TP-Link integration lists the P110 as supported. It says energy sensors can feed the Home Assistant energy dashboard and that devices are polled locally every 5 seconds.

Some caveats from Home Assistant's documentation:

  • Tapo devices need your TP-Link cloud credentials to authenticate locally.
  • Some firmware versions require "Third-Party Compatibility" to be enabled in the Tapo app before Home Assistant can connect.
  • Automatic discovery will not work across subnets, so add the plug by IP address and give it a static address.
  • Newer Tapo devices do not report a lifetime total-consumption entity, so use the daily and power sensors.

So "local polling" is real, but it is not entirely plug-and-play. It depends on firmware and setup.

How to run your own audit​

  1. Confirm the plug's regional rating covers your expected peak load, including UPS charging and drive spin-up.
  2. Plug the meter in at the wall, upstream of the UPS or power strip, and let it run for at least a full week.
  3. Record average watts and total kWh. If you use Home Assistant, log the power sensor so you can see the baseline, the peaks and any nightly steps.
  4. Enter your actual utility rate into the Tapo app's bill estimate, or use average watts × hours ÷ 1,000 × your rate.
  5. To find the biggest consumers, take a spec-based estimate first. Then meter suspects individually during a planned maintenance window.

Why it matters​

The measured draw sits largely in the always-on infrastructure, not the compute. Spinning NVR disks, a PoE switch and cameras account for most of the 72 W. The mini PC is a minor line item. If you want to trim the bill, look at drive spin-down policies, PoE budgets and which services truly need 24/7 uptime. Do that before shopping for a "greener" CPU.

The $9.50 monthly figure is modest at average US rates, and it is easy to overstate. At Hawaii or California prices it gets more serious. At Louisiana rates it is a rounding error next to a streaming subscription. The value here is not the shock. It is the habit of measuring with an inexpensive plug before guessing.

 

References

  1. I plugged my whole home lab into a smart energy monitoring plug for a week, and the electric bill reality check shocked me XDA 2026-11-17T15:00:22+00:00
  2. Electric Power Monthly - U.S. Energy Information Administration (EIA) eia.gov
  3. Tapo P110 | Mini Smart Wi-Fi Plug, Energy Monitoring | TP-Link tp-link.com