Augmented reality has been five years away for about fifteen years. That is unusual: most technologies either arrive or are abandoned. This one keeps almost arriving.
The reason is not lack of investment or imagination. It is that the device everyone wants requires four things that are in direct physical conflict, and no amount of software solves an optics problem.
The four constraints
Imagine the target product: ordinary-looking glasses, all day on one charge, bright enough outdoors, with a wide field of view.
Optics. Getting a bright image into your eye from a source at the temple, through a lens you can also see through, requires a waveguide. Waveguides are inefficient — a large fraction of the light never reaches your eye — and their efficiency drops as field of view increases. A wide field of view therefore needs a much brighter light source.
Power. A brighter source draws more power. Sunlight is roughly 10,000 nits; a display competing with it needs to be very bright indeed. Batteries small enough to sit on your face hold very little energy, and battery density improves at a few per cent a year.
Heat. Everything that draws power produces heat, next to your skin, with no room for a fan. Comfortable skin temperature caps the total power budget at a couple of watts for the whole device — which is roughly a tenth of what a phone will happily dissipate.
Weight. Anything over about 70 grams becomes uncomfortable within an hour, and the weight distribution matters as much as the total. Weight on the front of the face is far worse than weight at the back.
Now the trap: improving any one of these worsens another. Wider field of view needs more brightness needs more power needs more battery needs more weight and more heat. This is why the category has stalled, and why the products that exist are all points on a curve rather than steps toward a single answer.
You can currently have a wide field of view, or all-day wear, or a socially normal appearance. You cannot have all three. Every shipping product picks two at most, and the interesting question about any new device is which one it gave up.
The three product categories that actually exist
Passthrough headsets
Fully enclosed, with outward cameras showing you the room on internal displays. Technically mixed reality rather than see-through augmented reality.
Advantages: excellent displays, wide field of view, genuinely convincing virtual objects, no optical compromise because you are looking at a screen.
Limits: heavy, hot, tethered to a battery, and impossible to wear in public without comment. Cameras introduce a few milliseconds of latency between the world and your perception of it, which most people tolerate and some find nauseating.
Where they work: seated, indoors, for a defined session. Design review, training simulation, immersive entertainment, and a genuinely compelling use as a large private display for working on a plane or in a small flat.
Display glasses
Glasses-shaped, with a small screen showing notifications, directions, subtitles or a virtual monitor. Limited field of view, no real understanding of the room.
Advantages: light, socially acceptable, all-day battery, and available now at reasonable prices.
Limits: the image floats rather than attaching to the world. It is a heads-up display, not augmented reality.
Where they work: live captions and translation — which is a genuinely significant accessibility application and probably the strongest use case in the entire category — plus navigation, teleprompting and a private screen.
True see-through augmented reality
Transparent optics with real spatial understanding, where virtual objects stay convincingly fixed in the room.
Advantages: the thing everyone actually wants.
Limits: narrow field of view, high cost, disappointing brightness outdoors, short battery life, and bulk. Mostly enterprise and development hardware.
What has genuinely improved
It would be unfair to imply nothing has moved.
Tracking is a solved problem. Determining where a headset is in a room, to sub-millimetre precision, at high frequency, from cameras alone, was extremely hard a decade ago and is now reliable and cheap. Nobody complains about tracking any more.
Hand and eye tracking work. Controller-free interaction is genuinely usable, and eye tracking enables foveated rendering — rendering at full detail only where you are looking, which cuts the graphics workload substantially.
Displays got much better. Higher resolution, better colour, higher brightness per watt. Real progress every year.
Microdisplay efficiency. New emitter technologies are improving the light-per-watt figure, which is exactly the constraint that matters most.
The pattern: everything downstream of the physical constraints has improved dramatically. The physical constraints have improved slowly.
Where the interfaces are still unsolved
Even granting perfect hardware, some design questions have no accepted answer.
Text entry. No good solution exists. Virtual keyboards are slow, voice is unusable in public and around others, and gesture typing has not worked. Any productivity claim runs into this immediately.
Managing attention. A display in your vision that you cannot look away from is a fundamentally different proposition from a phone in your pocket. The notification design that is merely annoying on a phone is intolerable on your face, and nobody has established the right restraint.
Other people. Cameras on a face change social situations. The first mainstream attempt at this failed for social rather than technical reasons, and the underlying objection has not gone away. A recording indicator is a partial answer, and trust in it is the actual issue.
What it is for, most of the day. This is the real one. Passthrough headsets have clear session-based uses. Glasses have clear brief uses. Neither has a compelling answer for the eight hours in between, and without that the category stays a peripheral rather than a platform.
A realistic timeline
Based on the rate at which the binding constraints are actually moving:
- Now to 2028. Display glasses get better and cheaper. Captioning, translation and navigation become genuinely mainstream uses. Passthrough headsets remain niche but improve, with the private-display use growing fastest.
- 2028 to 2032. Wider field of view in a glasses form factor, probably by moving compute and battery to the phone or a pocket unit. Real see-through augmented reality becomes a premium consumer product with meaningful limitations.
- Beyond 2032. The all-day socially normal device becomes plausible if microdisplay efficiency continues improving. It is not a certainty, and it depends on physics rather than on software or funding.
What to actually buy today
- For occasional immersive use, a large private screen, or spatial work: a passthrough headset is genuinely good now, provided you accept sessions rather than all-day wear.
- For captions, translation, navigation or a discreet second screen: display glasses are good value and the least over-promised part of the market.
- For augmented reality as it is usually depicted: wait. The products exist and they will disappoint you, not because they are badly made but because the optics are not there yet.
- For development: buy whatever your target platform is and build for a narrow field of view. Designs that assume a wide one will not run on the hardware people actually own.
The technology is not stalled through lack of effort or ideas. It is waiting on brightness per watt, and that is a materials science problem with a materials science timeline. Everything else — the tracking, the rendering, the interaction, the software — is largely ready and has been for a while.