But despite every big company throwing a couple billions at the problem for the last few years (Facebook's $2 billion acquisition of oculus, Google's $500 million investment into Magic Leap, Apple's acquisition of Metaio/ARkit, Microsoft Hololens), all we have are literal toys!
So now you can't talk about AR as a serious subject anymore. People all dismiss it, saying we are still too early. I imagine it's like that for every "toy to big thing" transition.
Altair was a toy. Atari 2600 was a toy. TRS-80, Sinclair Spectrum, Apple II — all toys. Amiga was almost not a toy, and Aople Macintosh and IBM PC were mostly not toys anymore.
From there on, businesses started to use massive amounts of PCs, thanks to killer apps like word processors and spreadsheets.
Prices fell, home users took note, and a real explosion began.
It took mere 20 years, say, from 1980 to 2000.
On that note, who knows what would AR require in order to take off. It probably needs to be something you can get 80% of the killer features without wearing glasses. It also needs to support whatever frames glass wearers want to wear. It might require actual visual implants as I don’t think there’s a way to turn a contact lens into a display.
It of course had a large external induction coil to power it, and very low resolution.
Alternatively taking a long view, how many times do you think the dominant form factor of personal computers (in the general sense, including smart phones) will change? Another 3 times, another 5? Every 20 years forever?
The only exception I can think of is a direct neural link, but that’s generations away and even then I’m not entirely convinced. Maybe in the very long term.
What about contact lenses?
Voice is much too slow, as is listening - might as well use the current devices at those speeds.
Computers started with bookkeeping and mass production control. It took quite some time for them to penetrate office market or to (partially) replace the previously used communication tools.
But there are still more technical problems to solve before the killer application and corresponding headset will be cheap, useful, and lightweight enough for mass adoption.