Wi-Fi smart devices and the costs nobody prints on the box

The one failure that accounts for most Wi-Fi setup problems#

If a smart plug, bulb, or camera will not complete setup, start here before anything else.

Almost all smart home devices are 2.4 GHz only. They use a single-band radio because 2.4 GHz travels further, passes through walls better, and costs less to build. That has not changed in 2026 and there is no sign of it changing, because none of the three reasons has gone away.

Most modern routers, meanwhile, present 2.4 GHz and 5 GHz under one SSID and use band steering to push capable clients to 5 GHz. Now consider what onboarding actually does: your phone joins the device's temporary network or talks to it over Bluetooth, hands over your SSID and password, and the device tries to join. Several things can now go wrong at once:

• Your phone is sitting on 5 GHz, so the app hands the device credentials for a network the device cannot see on that band, or the provisioning flow refuses to proceed because it thinks you are on the wrong network.
• The device joins, then the AP tries to steer it, and a simple IoT stack handles the 802.11v, 802.11k and 802.11r roaming exchange badly and drops.
• The app never finds the device on the LAN because the phone and the device end up on different bands with client isolation on.

The fix is structural, not per-device: separate the SSIDs. Run your main network on 5 and 6 GHz and a dedicated 2.4 GHz IoT SSID, optionally on its own VLAN. It also makes every future device setup boring, which is the goal. The full ordered diagnosis, including DFS, WPA3, hidden SSIDs and captive portals, is in the device will not join 2.4 GHz Wi-Fi, and the network design is in the network setup a smart home wants.

Airtime: the cost that does not show up as a number#

A Wi-Fi channel is a shared medium. Only one device transmits at a time, and every transmission occupies the channel for a duration that depends on the data rate it negotiated. A slow, distant, 2.4 GHz-only smart plug transmitting a small status update takes far longer to send those bytes than a modern client would, and while it does, nothing else on that channel transmits.

That is why the damage from Wi-Fi IoT devices is not proportional to their traffic volume. Twenty cheap plugs sending a few bytes each are not "almost no traffic". They are twenty slow clients, plus their beacons, probes, DHCP renewals, cloud keepalives and NTP checks, all competing on the busiest band in the house with your neighbors doing the same thing.

Two practical consequences:

1. Moving high-throughput clients off 2.4 GHz helps everything. Get laptops, phones and TVs onto 5 or 6 GHz and the IoT band gets quieter for free.
2. Every Zigbee, Thread or Z-Wave device you buy instead is airtime you did not spend. This is the strongest technical argument in which radio for which job.

How many Wi-Fi devices can a home network take?#

There is no credible published figure, and anyone quoting one is guessing. Consumer router vendors do not document association limits in any consistent way. What actually constrains you, roughly in the order you will hit it:

The honest guidance: past roughly 50 Wi-Fi IoT devices, the answer is not a bigger router. It is either dedicated access points for the IoT SSID, or moving the light, plug and sensor population onto Zigbee, Thread or Z-Wave. Buying a mesh Wi-Fi system to solve a device count problem usually moves the problem rather than removing it.

Where Wi-Fi is the right answer#

It genuinely is, in three cases:

• Anything carrying video. Cameras and doorbells need megabits. IEEE 802.15.4 carries 250 kbit/s for the whole channel, so Zigbee and Thread are not candidates. Use Ethernet or PoE if you can, and 5 GHz Wi-Fi if you cannot. See cameras and doorbells.
• Devices with a screen or a speaker. Displays, TVs, speakers and hubs are mains powered, high bandwidth, and often are the border router or hub for everything else.
• A small number of devices in a house with no hub. If you own four smart devices and no intention of building a system, Wi-Fi is fine and a hub is overhead. Do you need a smart home hub treats this seriously rather than pushing hardware.

Where Wi-Fi is the wrong answer#

Battery devices. A Wi-Fi radio must maintain association, renew leases and respond to the access point. A sleepy Thread, Zigbee or Z-Wave end device turns its receiver off and lets a parent hold its messages. The difference is measured in years of battery life, and it is the reason sensors should never be Wi-Fi.

Anything you need in an outage. A cloud-dependent Wi-Fi device is offline the moment your ISP is, even though the device and the phone are three meters apart on the same LAN. What still works when the internet goes down has the device-by-device picture, and the outage audit checks your own list.

Anything at volume. See airtime, above.

Matter over Wi-Fi, specifically#

Matter runs over Wi-Fi, Ethernet and Thread. A Matter over Wi-Fi device is a normal Wi-Fi client with all the properties above, plus a dependency on mDNS working correctly across your network, because that is how controllers discover it after commissioning. It also uses Bluetooth LE during commissioning and then stops using it.

So Matter over Wi-Fi does not fix the 2.4 GHz-only problem, the band steering problem, or the airtime problem. What it fixes is the vendor lock-in problem: the device works with multiple ecosystems and keeps working locally. That is worth a great deal, and what Matter still cannot do is the honest boundary.

Wi-Fi 6, 6E and 7 for IoT#

Mostly irrelevant, and it is worth being blunt about it. The headline features of Wi-Fi 6E and Wi-Fi 7 are about 6 GHz and very wide channels, and no smart plug is going to use either. What helps indirectly:

• OFDMA, which lets an access point serve several small clients within one transmission opportunity rather than one at a time.
• BSS coloring, which reduces unnecessary deferrals between overlapping networks.
• Moving your fast clients up a band, which is the biggest single win and needs no new standard at all.

There were reports in January 2026 that the Wi-Fi Alliance introduced Wi-Fi 7 certification covering 20 MHz-only station devices, which would be aimed squarely at IoT. Treat that as reported rather than settled, and do not buy on it.

Wi-Fi HaLow: not yet#

Wi-Fi HaLow (802.11ah) is the version of Wi-Fi that should own the long-range low-power niche. It has been certified since November 2021, runs sub-1 GHz, and the Wi-Fi Alliance describes range of approximately one kilometer. Silicon exists from Morse Micro and Newracom, Morse Micro's HaLowLink 1 gateway shipped in February 2025 at $99 as a developer product, its development platform received Matter certification in June 2025, and a Quectel FGH200M module cleared FCC, IC, CE and RCM approvals in July 2026 for industrial use.

What does not exist, as of September 2026, is any mainstream ecosystem support or any mass-market HaLow smart home device. The realistic niche is long driveways, gates and wells, and for those Z-Wave Long Range is the answer you can actually buy today.