Four things that changed at CEDIA 2026

We spent three days on the floor at CEDIA Expo 2026 in Denver. Most of what gets written after a trade show is a list of products, which is useless a month later. So here is the shorter, more useful version: four things that actually changed, and what each one means for a house being designed right now.

One — Direct-view LED stopped being exotic

For years the answer to "should this be a projector or a big screen?" had an easy default. A dedicated dark room got a projector; everything else got the largest television that would fit. Direct-view LED — a wall built from LED modules rather than a single panel — existed, but at a price and a pixel pitch that kept it in boardrooms.

That has moved. There were more LED walls on the floor this year than projectors, at pitches that make sense in a house.

Pitch is the number that matters, and it is simpler than it sounds. It is the distance in millimetres between one LED and the next. Smaller pitch, finer image, higher cost. The working rule we use:

In practice — the pitch in millimetres is roughly the closest comfortable viewing distance in metres. A 1.2mm wall reads clean from about four feet. A 0.9mm wall holds up closer than that. If you are sitting twelve feet back, you are paying for resolution you cannot resolve.

The honest call has not changed as much as the marketing suggests. A dedicated room with controlled light still wants a projector — nothing matches it for scale per dollar, and black level in a dark room is a projector's home ground. But a room with real daylight that still has to perform at two in the afternoon is where LED now wins outright, and that used to be a compromise conversation. It isn't any more.

Two — The finish conversation moved upstream

The most useful thing on the floor was not a product. It was how much of the floor was finish samples.

Keypad and switch finishes. Downlight trim option boards. Speaker grilles printed to match the surface they sit in. Drapery hardware in the same metals as the cabinet pulls. Acoustic panel faces in woven, perforated timber and micro-perf metal.

All of that is a finish. It has a lead time, it has a colour, and someone has to specify it. Which means it belongs on the finish schedule, next to the paint and the millwork and the plumbing fixtures — not discovered at trim, when the answer is whatever the distributor has in stock.

In practice — if a keypad is being engraved, that is a six-to-eight week item with the room names on it. It cannot be decided the week the electrician is landing devices. We now ask for keypad locations and engraving text at the same time as the lighting plan, because the alternative is a beautiful room with a blank white rocker in it.

We carry Lutron, and their drapery hardware was on the floor in finishes that read as architecture rather than as hardware. That is the direction of the whole category.

Three — Voice control got less interesting, and that is the point

There was very little new to see in voice this year. No new wake words, no new party trick.

What there was instead was a lot of attention on failure modes. What happens when the internet drops. What happens when two people ask for different things in the same room. What happens when a command is ambiguous and the house has to pick. That is the unglamorous engineering that decides whether voice is something a family uses in year three or something they stopped using in month two.

Johann completed Josh.ai's Advanced Training at the show, and that was most of what it covered — not what the system can do, but how it behaves when something is wrong. A system that fails predictably is worth more than one with a longer feature list.

We carry Josh.ai and specify it in front of Lutron and Control4, so this is the layer we care about most.

Four — The oversold category was AI as a feature label

If a product had a screen, something on the stand said AI.

Some of it is real and unglamorous — room correction that actually measures, camera analytics that can tell a delivery from a raccoon, network gear that reroutes before anyone notices. Most of it was a label applied to logic that has been in the product for five years.

The test we used walking the floor, and the one worth using on any proposal: ask what the system does differently as a result, and ask what happens when it gets it wrong. A real answer is specific and slightly boring. A vague answer means the word is doing marketing work, not engineering work.

The pattern underneath

Reading these back, three of the four are not about technology at all. They are about when a decision gets made.

LED versus projector is a decision that has to happen while the room's daylight is still on a drawing. Finishes have to be specified while the finish schedule is open. Voice behaviour has to be designed before anyone is living with it. The technology in all three cases is ready — the failures are scheduling failures.

That is the same thing we say about prewire, and it is why we would rather be in a conversation early and cheap than late and expensive. The systems that age well are the ones where somebody asked the question in time.

If you are designing or building something now and want these decisions made in the right order, send us a note through the site, or call or text the office at 832-295-9098.

Houston · Austin · San Antonio · the Hill Country

StroTek LLC · TX Lic. B29770101

Photographed at CEDIA Expo 2026, Denver.