It's only a matter of scaling and adapting further technologies which will further decrease costs and improve quality of digital.
Here's a crazy thought experiment (numbers from Epson and Heidelberg websites). You can caveat this to pieces but if that's your line of thought you're missing the point.
An Epson 9900 proofer retails for $5,995 and is at least 4 year old technology.
It can produce 40" x 60" prints from 15:26 to 40:05 (Normal is 24:20). Let's use 24 minutes as a nice round number.
That's 1.64 sq. inches per second
A new Heidelberg XL106 retails for ???
At 18,000 it can produce 29 x 41 sheets per hour
That's 5,945 sq. inches per second
That's roughly 3,625 Epson 9900 printers to match the production of one XL106.
That's $21,731,875 MSRP for the Epson printers.
Take 20% off the top for volume purchase and your down to $17,385,500
Add your number for unique plant engineering to build a (crazy) array that can feed and deal with the output of these thousands of proofers, plumb them all to one set of giant ink reservoirs, climate controls, building, etc. I'm going to throw out $10,000,000.
Your at $27,385,500.
Say there's a leap in technology with the Epson without any consideration to the external engineering and support costs:
2x todays speed $18,692,750
4x todays speed $14,346,375
6x todays speed $12,897,584
8x todays speed $12,173,188
10x todays speed $1,738,550
Ask yourself how many Heidelberg XL106 offset machines can you buy at that cost including building, support equipment, engineering, delivery, etc.? How long will it take for that Epson cost to come down? How long until they double the speed of the Epson? How long until machines are available without all that crazy custom engineering?