Which Smart Home Devices Survive a Polar Vortex Snowstorm?

Before a polar vortex drops snow on the driveway and ice on the lines, the useful question is not “Is my home smart?” It is: if grid power, internet service, or both disappear, which devices can still be controlled by a person standing in the hallway?

The January 2026 winter storms made that question less theoretical. At peak, more than 1 million customers lost power across the South, more than 800,000 were still out the next morning, and dangerous sub-zero lows stretched from the Northern Plains through the Northeast.[1] That is long enough for “temporarily offline” to stop being a dashboard message and start being a lock, thermostat, alarm, camera, or garage device with no usable path.

For polar-vortex smart-home prep, the rule is plain: a device survives only if its power source and its control path survive. A battery lock with a dead cloud bridge may still open from the keypad, but not from the app. A Zigbee sensor with fresh batteries may keep sensing, but only if the hub that receives its messages is alive. A Wi-Fi camera with no backup power is not “offline.” It is off.

Start with the failure path, not the product name

Every device in the house has two separate questions to answer. First: what powers it? Second: how does a command reach it? People usually notice the first one. The second one is where winter-storm plans get sloppy.

That table is the audit. Brand preference can wait. A beautiful app is only one possible route to the device, and in a storm it is often the least dependable route because it assumes power, Wi-Fi, internet, vendor servers, and account authentication are all still cooperating.

The network chain is the part to put on a UPS

If the house still has power but the internet is down, locally controlled devices may continue to work. If the internet still exists upstream but the router is dark, app control inside the house usually fails. If the router is alive but the hub is dead, battery-powered Zigbee, Z-Wave, or Thread gear may have no practical place to report.

The minimum protected chain is usually the modem or fiber terminal, the router, the primary Wi-Fi access point if it is separate, and the hub or bridge that actually controls devices. In some homes, that also includes a Thread border router, a Matter controller, a Home Assistant box, a security base station, or a bridge for lights and sensors.

This is where local protocols earn their keep, but only if the controller stays powered. Zigbee, Z-Wave, and Thread do not magically jump from a battery sensor to your phone. They need a route: hub, coordinator, border router, bridge, or controller. If that box sits in the same outlet strip as a lamp with no backup, the mesh may be local in theory and unavailable in the hallway.

A small UPS is usually more useful on the network shelf than on a random smart speaker. Keep the network path alive first. Then decide which devices are worth controlling through it. A router that stays up long enough to let you unlock a door, check a thermostat, silence a nuisance alert, or confirm a sensor state is doing real work.

Internet outage and power outage are different failures

Do not test for one and assume you tested the other. Pulling the WAN cable or disabling internet service checks whether local control works without the cloud. Cutting power to the router checks whether devices can be reached at all. Turning off a breaker checks whether mains-powered switches, lights, cameras, and bridges stay alive. Those are three different storm conditions.

The harsh case is both at once: utility power drops, the router and hubs are only alive if they are on backup, and the vendor cloud may be unreachable even if the local Wi-Fi name still appears on your phone. That is the condition worth preparing for before the snow arrives.

Locks: respect the keypad, key, and battery compartment

A smart lock is one of the few devices where “still works” has to mean more than “the app opens.” In a cold entryway, the important fallbacks are mechanical key, keypad code, interior thumb turn, external battery jump contacts if present, and known access to the battery compartment.

Battery-powered locks can keep operating through a grid outage because the lock motor and keypad are not depending on house power. Remote control is a separate matter. If the lock talks through a bridge, hub, Wi-Fi module, or cloud service, that route only survives if each piece in the route survives. A lock that opens from its own keypad has a cleaner storm behavior than one that expects a phone command to travel through a dead bridge.

• Replace weak lock batteries before the storm, not after the first low-battery warning in freezing weather.
• Confirm every household member who may need entry has a working code, key, or both.
• Test the lock with internet disabled, then test it with the hub or bridge unplugged. Note which functions remain.
• Keep the physical key outside the smart-home dependency chain: carried, lockboxed, or otherwise reachable by someone authorized.

Thermostats: separate heat safety from remote convenience

A thermostat’s storm value depends on the heating system behind it. The thermostat may have a display and batteries, but that does not mean the furnace, boiler, heat pump, air handler, zone valve, or ignition system has power. A thermostat can be alive and still unable to make heat if the equipment it controls is dead.

Before a polar vortex, set conservative heating schedules locally on the thermostat itself where possible. If the internet disappears, the thermostat should not need a cloud command to hold a safe temperature. If the thermostat depends on Wi-Fi for app control, assume remote changes are unavailable unless the router and upstream service remain active.

The useful test is boring: stand at the thermostat, change the setpoint, and confirm the heat responds. Then disable internet and try again from the wall control. If your setup uses a smart relay, room sensor, or remote temperature average, verify what the thermostat does when those devices cannot report.

Sensors: battery life is only half the story

Contact sensors, motion sensors, leak sensors, temperature sensors, and freeze sensors are often battery-powered, which makes them look storm-ready. They may keep detecting. The question is whether anything receives the detection, acts on it, or alerts a human.

A battery leak sensor under a sink is useful during an outage only if its alarm is local, its hub remains powered, or the homeowner can inspect it. A freeze sensor in a crawlspace may keep measuring temperature, but cloud alerts will not arrive if internet service is out. A door contact may keep reporting to a security base station on backup power, while a similar-looking sensor paired to an unpowered hub goes quiet.

Label sensor batteries with the month and year they were installed. This is not elegant, but it beats discovering during a cold snap that the “smart” part of the house was running on a coin cell nobody remembers replacing.

Lights, plugs, and switches: mains power decides most of it

Mains-powered smart bulbs, plugs, dimmers, and switches usually fail in the most literal way: no grid power, no device power. A local mesh does not help a bulb with no electricity. The useful preparation here is not pretending smart lighting will carry the house. It is deciding which lights matter and how they are powered.

Battery lanterns, flashlights, and charged task lights belong in the same plan as the smart-home gear. For the smart devices that do have power, local wall control matters. A smart switch that still behaves like a switch when the network is gone is preferable to a setup where the wall control only sends a wireless command to something else.

For critical plug-in loads, be careful. A smart plug on a heater, pump, router, or medical-adjacent device can become a single point of failure. If a device must stay on, make sure the plug’s default power-on behavior is known, and do not leave essential operation dependent on a cloud routine.

Cameras and doorbells: assume they are gone without backup power

Cameras are easy to overvalue in storm prep because their apps look confident on normal days. During a winter outage, they are among the first devices to lose practical value. A wired camera needs power. A Wi-Fi camera needs Wi-Fi. A cloud camera needs internet. A video doorbell may also need a powered chime circuit, a charged internal battery, or a bridge.

Battery cameras may continue recording locally if the product supports it, but many still need a hub, base station, Wi-Fi link, or cloud route for live view and alerts. If the camera is there to watch a driveway during a blizzard, check whether you can view it from inside the house with internet disabled. If you cannot, treat it as a normal-day convenience, not storm infrastructure.

Voice assistants and displays are poor emergency controls

Smart speakers and displays are pleasant when the network is healthy. They are weak as emergency interfaces. They need power, Wi-Fi, and usually an internet path to understand and execute most commands. Even when a limited local command works in a particular ecosystem, it is not the fallback to bet the front door or heat on.

Before a storm, make sure every important action has a non-voice path: physical switch, keypad, local app, wall thermostat, key, button, or automation that runs locally. The person trying to silence an alarm at 2 a.m. should not have to negotiate with a speaker that is waiting for the cloud.

Alarms and safety devices: know what is standalone

Smoke alarms, carbon monoxide alarms, security sirens, water alarms, and freeze alarms should be sorted into two piles: devices that make a local sound on their own, and devices that mainly notify an app or hub. The first pile can still help when the network is down. The second pile depends on the same chain as everything else.

If a security system has a base station with backup power and local siren behavior, it may continue some functions even while app features are degraded. If every alert depends on a push notification, then the useful question is whether the push notification can still be sent. In a combined power and internet outage, the answer may be no.

Do not silence a chirping alarm before a storm and promise to handle it later. Replace the battery, write the date on it, and test the local alarm sound. Smart integrations are secondary to the device doing its safety job in the room where the hazard happens.

Automations: only trust the ones that run where the power is

A routine can look local because all the devices are in the house, while the decision-making actually happens in a cloud service. During a storm, that distinction matters. If a freeze sensor is supposed to turn on a circulation fan, close a valve, or trigger an alert, find out where that automation runs and what happens if the internet drops.

The safest automations for an outage are simple, local, and already tested: sensor to hub to device, with the hub and device powered. Anything involving a voice assistant routine, cloud scene, remote notification, or third-party web service belongs in the “nice if available” category unless you have verified otherwise.

Also check startup behavior. After a power blink, some devices return to their previous state, some turn on, some stay off, and some wait for a network reconnect. That behavior can matter for lights, plugs, pumps, valves, and sirens. Test it before the breaker trips for real.

A status-labeled pre-storm checklist

Use three labels. “Confirmed” means tested under the relevant failure condition. “Workaround” means it can function with a UPS, manual control, or physical fallback. “Investigating” means it has not been tested and should not be trusted for storm-critical use yet.

A polar vortex is not the time to discover that the lock is battery-powered but the bridge is not, or that the freeze alert works only after a cloud round trip. Preserve the network path where it matters, test local control before the lights go out, replace or charge batteries, and know the physical override for every device that may lose power or cloud access.

References

1. January 2026 winter storms — NPR