The Best Backup Power for Smart Home Is One UPS

The most useful smart-home outage story this year was not about a freezer, a solar generator, or a heroic whole-home battery. It was a short blackout. On Aug. 7, 2026, Creating Smart Home documented a roughly 10-minute outage where Z-Wave fire and leak detectors using direct associations still triggered a siren while the Home Assistant coordinator was offline. At the same time, hub-dependent devices exposed the ugly part of the failure: DIY blinds and Hue lights did not simply “stay battery-powered” or “come back fine”; they failed into bad default states and needed attention after power returned. [1]

That is the real answer behind “best backup power for smart home during power outage.” The first thing to back up is not every smart device. It is the control chain: ONT or modem, router, any small switch needed to connect the controller, the hub, and the Thread border router. If that chain stays alive, many battery devices still have someone to talk to, automations do not necessarily lose their coordinator, and the house avoids a chunk of the restore chaos that starts when every controller reboots at once.

Wirecutter’s 2026 UPS testing is useful here because it measured the kind of load a smart-home control shelf actually resembles. Across its UPS picks, a 20 W modem-and-router load ran for about 1.5 to 4.5 hours; the roughly $140 CyberPower LE1000DG ran about 3 hours at that load, and the roughly $196 CyberPower CP1350AVRLCD3 ran about 4.5 hours. Transfer times were measured around 4 to 10 milliseconds, fast enough in principle to keep routers and hubs from rebooting when utility power drops. [2]

“In principle” matters. A UPS spec sheet cannot promise that a particular ONT power brick, router, hub, or USB-C adapter will ride through the transfer cleanly. But the shape of the problem is clear: the protected load is small, and a normal sub-$200 UPS can keep it alive for the kind of short outage that otherwise turns into an 11 p.m. mesh-repair session.

The one-UPS rule

Put one UPS under the control shelf before buying backup power for lamps, speakers, cameras, or convenience devices. The UPS should have enough battery-backed outlets for the chain and enough runtime at a light networking load. The usual order is simple.

Do not make this harder than it needs to be. If the router, hub, and border router live in different rooms, the first fix is often layout, not battery capacity. Move the hub shelf. Wire one small switch. Put the pieces that decide whether the house has a control plane on one battery-backed strip.

If you are deciding between a UPS and a larger portable power station, use NestGrid’s UPS-or-power-station recipe for the broader tradeoff. If you already know you need multi-hour appliance support, that belongs in the home battery backup escalation stack. For the smart-home control problem, the first purchase is smaller.

What still works when the controller path disappears?

Battery power is only one layer. A lock can have fresh batteries and still be unreachable from the app. A leak sensor can wake up and still have no hub to receive the event. A Thread device can be awake while the only border router is dead. The useful question is not “does the device have power?” It is “what path still exists, and what state will the system be in when power returns?”

Z-Wave: direct associations are the outage feature people forget

The Creating Smart Home case is valuable because it separates two Z-Wave behaviors that often get blurred together. The Home Assistant coordinator was offline, so hub-driven logic was gone. But Z-Wave devices with direct associations could still act locally: the fire and leak detectors triggered the siren without waiting for Home Assistant to come back. [1]

That does not mean “Z-Wave survives outages.” It means a specific design survives a specific controller failure: battery-powered safety devices associated directly with another Z-Wave device that can still receive and act on the command. If the siren is mains-powered and unprotected, or if the association was never configured, the result changes.

For safety devices, this is worth more than another dashboard card. Pair the sensor-to-siren path directly where the devices support it, then test it with the hub offline. The test is not whether the hub logs the event. The test is whether the siren sounds when the coordinator is absent.

Zigbee: the coordinator and powered routers decide how much mesh remains

Zigbee battery sensors may remain powered through the outage, but the useful mesh can shrink quickly if the coordinator is down and the mains-powered repeaters around the house lose power. A door sensor sitting on a coin cell has not failed; the network around it may have disappeared.

This is where a UPS on the hub shelf helps more than it looks. If the coordinator stays up, battery devices do not have to rediscover a rebooted controller after every flicker. If a few key Zigbee routers are also on protected power because they happen to share the same shelf or rack, even better. But this is not a reason to put every smart plug in the house on backup power. It is a reason to keep the coordinator stable and know which powered repeaters your mesh leans on.

Thread and Matter: the border router is part of the control chain

Thread makes the one-UPS rule more important, not less. Battery Thread devices can remain awake, but they do not become magically self-sufficient when every Thread border router loses power. For Matter over Thread, the border router is the bridge between the low-power mesh and the IP network where controllers, apps, and automations live.

The post-outage pain point is recovery. Community tracker reports around Matter over Thread devices returning unavailable after power cycling, with routes taking time to rebuild, are useful context but should be treated as Investigating-grade rather than NestGrid Confirmed unless reproduced on the same controller, firmware, and border-router mix. The practical takeaway is narrower: if the border router reboots with everything else, you should expect some setups to need patience, rediscovery, or manual repair before the app view is trustworthy again.

This is also why “my Matter device has a battery” is not enough information. Ask which border router it uses, whether that border router is on the UPS, whether the controller is on the same protected network, and how the device appears in the app after an actual power cut. The outage may last minutes; the cleanup can last longer.

Wi-Fi and cloud devices: local power does not equal service

Wi-Fi devices split into two groups during an outage. Battery Wi-Fi devices may still have power, but they need the router alive to stay reachable on the LAN. Cloud-bound devices need more: the router, modem or ONT, provider infrastructure, and the vendor cloud path all have to work. A UPS under your router fixes only the part inside your house.

For mains-powered Wi-Fi devices, restore state becomes the main issue. A smart plug, bulb, relay, or shade controller that loses AC may come back in a default state you do not like. That is not a battery problem. It is a device behavior problem, and it belongs on the test list before storm season.

Hub-dependent devices: the failure may be delayed until power returns

The annoying failures are not always visible during the blackout. A blind that restores wrong, a light that returns in the wrong state, a bridge that needs time before automations are valid again, or a device that looks present but ignores commands can all appear after utility power is back. The Creating Smart Home outage showed that clearly: the controller dependency, not just the power source, determined whether a device behaved usefully. [1]

Hue restore behavior, Matter recovery, and bridge-specific restart quirks should be handled as setup-specific until tested. Record the normal restore setting for lights and relays. Cut power. Bring it back. Watch the app for unavailable devices, delayed route repair, and wrong defaults. If the device is safety-relevant, do not count an app status as proof; test the physical behavior.

The UPS spec that matters is boring

For this job, the winning UPS is not the most impressive battery in the house. It is the one with enough battery-backed outlets, clean transfer behavior, and enough runtime at a light networking load. Wirecutter’s measured 1.5-to-4.5-hour range at 20 W is the right scale for the decision; it maps to a modem/router/control-shelf load better than refrigerator math or power-station marketing does. [2]

A practical smart-home UPS check looks like this:

• Enough battery-backed outlets for ONT or modem, router, switch if needed, hub, and Thread border router.
• Runtime tested at a small network load, not judged from the largest VA number on the box.
• Transfer behavior that your actual power adapters tolerate without rebooting.
• USB or network monitoring support if you want Home Assistant to react to grid loss.
• A layout that keeps the whole control chain on the UPS, rather than protecting one impressive device while leaving a tiny switch unprotected.

Product rankings and sizing math change over time, so keep model picks separate from the control-chain rule. Use NestGrid’s tested smart-home outage gear picks for dated model choices and the backup power sizing guide if you are adding cameras, NAS hardware, access points, or longer runtime targets. The compatibility rule stays the same: protect the control path first.

Let Home Assistant know the grid is gone

Once the control chain is on a UPS, the next useful upgrade is observability. Home Assistant’s NUT integration connects to Network UPS Tools so the hub can monitor UPS state and use that information for shutdown logic or automations. [3]

That opens up cleaner outage behavior. The house can detect that utility power is gone, avoid nonessential automations, send a local alert while the network is still alive, or shut down a hub gracefully if the battery gets low. If you already have blackout load-shedding routines, pair the UPS signal with NestGrid’s heat-wave blackout automation patterns.

Do not overload this step. NUT does not make a cloud service reachable, and it does not fix a bad restore state on a smart bulb. It gives the controller a reliable signal that the power environment changed, which is exactly the kind of signal automations should be using before they make decisions during an outage.

The ISP caveat is not optional

Backing up the home control chain does not guarantee internet. Router and modem backup guides correctly focus on keeping home equipment powered, but the path also depends on provider infrastructure outside the house: fiber equipment, cable nodes, and cellular sites can have their own power limits during an outage. [4]

Fiber service adds one more local wrinkle. GFiber’s support page describes an Internet Battery Backup option that can provide up to about 2 hours of backup for eligible fiber equipment, which changes what a homeowner may need to power locally. [5] If your provider already backs up the ONT, your UPS may only need to cover the router, switch, hub, and border router. If it does not, the ONT belongs on your UPS list.

Either way, local control still matters. A Home Assistant box, Hubitat hub, SmartThings station, or Matter controller on a live LAN can keep useful local behavior available even when the internet path is degraded. Cloud-only devices are the exception you should identify ahead of time, not discover after the lights flicker.

Test the chain before you trust it

Do one controlled outage test with only the control chain on the UPS. Do not plug in lamps, chargers, speakers, or random convenience loads for the first run.

1. Plug the ONT or modem, router, required switch, hub, and Thread border router into battery-backed outlets.
2. Confirm the hub, router, and border router are online before the test.
3. Unplug the UPS from wall power and watch for any reboot, not just for internet loss.
4. Trigger one local automation, one battery sensor, and one safety path that should work without cloud service.
5. Restore wall power and watch which Zigbee, Z-Wave, Thread, Matter, Wi-Fi, and bridge devices return unavailable, delayed, or in the wrong state.
6. Fix the dependency you actually observed: a missing direct association, an unprotected switch, a border router off the UPS, or a restore setting that needs changing.

The best backup power for a smart home during a short outage is the battery that prevents the control chain from falling over in the first place. After that test, you can decide whether the house needs a larger battery plan. Start with the packet path.

References

1. Smart Home Power Outage: What a 10-Minute Blackout Taught Me About My Smart Home’s Weak Points — Creating Smart Home, Aug. 7, 2026
2. The 3 Best Uninterruptible Power Supplies (UPS) of 2026 — Wirecutter
3. Network UPS Tools (NUT) — Home Assistant
4. Battery Backup for Router and Modem — BLUETTI
5. Internet Battery Backup — GFiber