The Headset Was Never the Future. AR and AI Are.

I have believed in immersive technology since 2004. I still do. But I stopped believing the headset was the future a long time ago, and the last few years have shown why.

Look at where the industry is now. The company that bet the most on the metaverse has cut more than a thousand jobs from its VR division, closed three of its own VR studios, and moved its flagship virtual world off headsets and onto phones. VR headset shipments fell 18% year on year in the second quarter of 2026. Meanwhile, smart glasses with AI built in sold more than seven million pairs in 2025, three times the two years before combined, and sales nearly doubled again this year.

This is not a story about VR failing. It is a story about what people actually want from technology.

Why VR stalled

VR had two problems. The first everyone could see: the headset. It is heavy, it isolates you from the room and the people in it, and for many people it makes them feel ill. Wearing one is a commitment.

The second problem was quieter and more important: there was not enough to do once you put it on. Every world, every character and every scenario had to be built by hand, at enormous cost. So outside games and a few big-budget experiences, there was little reason to come back. Plenty of people bought headsets. Far fewer kept using them.

AI is now fixing the content problem, and in my view that will matter a great deal for VR in training, healthcare and simulation. But it does not fix the first problem. You still have to leave the world to use it.

Why AR and AI belong together

Augmented reality does the opposite. It adds to the world instead of replacing it. And for years it had the reverse problem to VR: it could put information in front of you, but it had very little worth showing. Without context, an overlay is just clutter.

AI is the missing piece. Glasses with a camera, a microphone and an assistant can see what you see and hear what you hear, and then decide what is worth telling you, at the moment you need it. That changes the question from "what can we display?" to "what does this person need right now?"

Some of it is already here:

Live translation, with captions in your line of sight when someone speaks another language.

Captioning glasses that let people with hearing loss follow a conversation without looking down at a phone.

Technicians guided step by step, with instructions and markers placed on the equipment in front of them.

An assistant you can simply ask: what is this, how do I fix it, where do I go next.

I have been designing for this for a while

In 2019 I designed a connected customer concept for airports and presented it to four of the UK's major airports. Alongside an AI assistant that stayed with the traveller from planning to boarding, it imagined AR glasses guiding people through the terminal: arrows on the floor to the gate, a gate change appearing wherever you happen to be standing, the lounge turning into a live travel dashboard. The airports liked it. The hardware was not ready.

It is getting ready now. The difference is not the glasses. It is the intelligence behind them.

What still has to happen

I am optimistic, but I do not think this is inevitable, and pretending otherwise does the idea no favours.

Privacy is the real test. Glasses that can see can also record, and facial recognition in everyday eyewear has rightly drawn strong objections from civil liberties groups and legislators. If the industry gets this wrong, it will repeat Google Glass. If it gets it right, with visible signals, clear limits and bystanders protected by default, people will trust it.

The technology also has some way to go. Today's display glasses show information in a small window rather than across your whole view. Batteries struggle to run cameras, AI and displays all day. True AR glasses, with graphics placed in the world around you, are only now reaching consumers and cost more than two thousand dollars.

Where this goes

The winners will not be the flashiest experiences. They will be the moments where hands-free, in-the-moment help is clearly better than taking out your phone: the traveller who is running late, the engineer up a ladder, the carer who should be looking at the patient rather than a screen, the visitor who does not speak the language.

That is a design problem as much as a technology one. The hard part is not deciding what glasses can show. It is deciding what they should leave out.

The headset asked us to leave the world. AR and AI meet us in it. That was always the more human idea.