Assuming that color management is seriously addressed.
But depth is a multiplier that the hardware requirements discussion has possibly overlooked:
I see most calculations above use 8 bits per channel, but 12 bits per channel is the range of professional video production, it 4:4:4 signal.
Panasonic's little GH5-s just announced records in 4:2:2 for 64* the color information of consumer 8 bit cams.
(Note the -s designation, the difference is all important. This video dedicated variant seems to have been launched without any leaks whatsoever, but garnered across the board coverage immediately. The importance of the camera has been list unfortunately to fresher news from Fujifilm and Sony, but the impact on the video formats (sensor size, lens mount,codec, workflow bandwidth) game may be significant.
As a result the price of capturing a signal you can apply the same quality of 3D blue screen graphics to, has plummeted. I know that the Sony A7ii has comparable capability at only four times the price but the recording of the full signal is bored, shifting the cheapest option to be ten times the price of this sweet little thing. No skin in this game except it's made a project long in development, finally viable)
We're much more susceptible to color information errors in terms of accepting new realities, once the other hurdles are overcome. This is because we're a part of the color information equation that leads to difficulty such as perceptual models, and equivalence elipses...
64 times the color information, anyhow is the fact that you deal with when matching your talent plus a blue screen background and CGI animated effects or infographics, which is the standard set-up in many studios. I'm no longer able to ask clarification personally from a friend but I believe that Sky News's anchor desk is virtual for some time now.
The color 'depth" is vital to create convincing mattes for consumption and the difference is instantly recognisable by most people over normal broadcast television. We're that good at adjusting for color variations that we jank bad on inconsistencies.
I can immediately think of the kind of application for which custom rendering hardware can get orders at the kind of tickets that might drive developments.
High end fabric weaving is one.
This can involve looms four stories high being fed by hundreds of bobbins. Threads are woven into the (vertical) by air suspension carried shuttles to create the complex wefts the industry is thriving on to meet with insatiable consumer demand for high value patterns in short quantity.
Set-up and teardown costs are considerable.
Color matching and selection, particularly in combination and in combination with varying reflectance and absorption spectra which interact including by means of elasticity which also governs shuttle speed and hence overall weaving rates is a scientific application of significant commercial value.
I can imagine thread suppliers wanting to deliver hardware modeller boards for the color properties, and high end FPGAs would be entirely justifiable to protect decades of research investment.
Being able to make interactive vr selections for new designs could make or break the most successful cloth manufacturers, all of whom need to capture the fast moving and ephemeral fashion orders.
In printing for magazines as well, I know that press proofing is often a vital role in contract with new and existing print supply alike. The sheer cost of errors, which if as late as being found in the make ready for a run are highly hazardous to anyone who has thin margins, let alone the drama of stopping a scheduled run with a bigger customer running right behind you... this is the single least appreciated cost involved in magazine production and technology has been too long almost about to resolve the risks. Too long to save smaller print shops. Too long to keep many independant publishers afloat. And it's a primary tech barrier to trading advertising positions independently.
I used a Symbolics xl1200 when new, and I disagree with the online inflation calculators, believing they are far too conservative. I would suggest that$120K in 1986 is more than $239,000 today. Twice that at least and maybe ten times. My hundred thousand dollars apartment went up sixteen fold in the same time.
You can get a lot of hardware for the current market value of that lispM, today.
If you accept my high end estimate of$1MM then I think people will build aggressive designs for the abilities this kind of display technology will encourage.