Matter arrived with an unusually strong pitch: one standard, backed by Apple, Google, Amazon and Samsung simultaneously, that would end the compatibility mess in smart home hardware. Buy anything, it works with everything.
Several years on, it is worth being precise about what happened, because the gap between the promise and the reality is exactly where projects go wrong.
What it delivered
Basic device types genuinely do just work. A Matter light, plug, switch, sensor or lock will pair with any Matter controller and expose its core function. That is real, and it is not nothing — it removes a whole category of "will this work with that" from the specification process.
Thread is a good network. Low-power, mesh, IPv6-native, no vendor hub. For battery sensors and lightweight controls it is technically better than what came before, and the mesh genuinely self-heals when a node drops.
Multi-admin works. One device can be commissioned into several ecosystems at once. A lock that answers to both the family's iPhones and the house control system, without a bridge, is a meaningful improvement.
What it did not
The lowest common denominator problem. Matter standardises the common subset. A colour bulb exposes on, off, brightness, colour. The manufacturer's own app exposes the effects engine, the power-on behaviour, the fade curves, the firmware updates. You reach the standard's ceiling faster than you expect, and then you are back in the vendor's app anyway.
Border routers are the weak link. Thread needs one, and in most houses it arrives inside a speaker or a TV box that was bought for other reasons. That device now carries infrastructure duty it was never specified for. Whether the two border routers in a house form one Thread network or two competing ones is a question that has caused real, hard-to-diagnose faults.
Commissioning is still fiddly. The QR code flow works, until the device is already on a shelf at nine metres, or the code is on a label now buried behind plasterboard. Capture every code before first fix, or you will be pulling fittings down.
Adoption is uneven. Lighting and sensors are well covered. Cameras, high-end audio, HVAC and anything with a real interface are far patchier. The categories where interoperability would be most valuable are the ones least well served.
What this means when we specify
We treat Matter as a useful baseline, not an architecture.
For the wired backbone of a house — lighting control, blinds, heating zones — we specify a system chosen on its own merits, because those need to be right for twenty years and Matter does not yet reach deep enough to build on.
For the long tail — a sensor in an awkward corner, a plug behind a Christmas tree, a lock on an outbuilding — Matter and Thread are excellent. Cheap, reliable, no extra hub, no vendor lock-in.
And we put a proper Thread border router in the rack rather than relying on whatever consumer device happens to be in the room. It costs very little and removes an entire class of intermittent fault that is genuinely unpleasant to chase.
The honest summary
Matter made the cheap end of the market better and left the expensive end roughly where it was. That is a real achievement, and it is not the achievement that was announced.
If somebody tells you a house can be built entirely on Matter today, they are either working to a very simple brief or they have not tried.