I think the main problem was that systems to take advantage of it tended to be either fully bespoke or were produced in runs of dozens or hundreds at most. Customisation and installation (including wiring the terminals, in the days before networking was common) were protracted processes as well.

As a result, they were priced against the low-end of mid-range systems - so rather more than you might expect from looking at the raw bill of materials.

Their niche was rapidly eroded by simply running multiple single user PCs at the low end, and networked small Unix systems at the high end - both of which benefitted from higher economies of scale and needed less systems integration work.

Definitely interesting in a "what might have been?" way, though. I suspect that if DR had done a deal with IBM, then we might have ended up going down that sort of path for most of the 80s.