You are mixing a couple of different stories I fear...
But a snap shot for this mailing list is this ... The PDP term was used by KO when he was funding Digital because the VC's the 60s did not believe that a computer company would succeed. But making things for the instrumentation market (Lincoln Labs, Livermore, etc.). This is how PDP-1 came to be. The PDP term was keep for the first 25 years (until the Vax and renaming of the PDP-10/PDP-20 to DECSystem 10/20). [The 1 begat the 6, 9, 15 an 10 families].
The concept of purchasing smaller system, was indeed true. This was the idea behind the >>mini-computer<< or minimal computer that Gordon Bell who lad left DEC temporarily to be a CMU Prof for a time began to explore. He took the idea and commercialized and the PDP-8 line was the first in that line. The 11 & 15 were full computer systems. DEC also made something called the PDP-16 'Register Transfer Modules' (RTMs) which was an attempt to make the small controller idea even more accessible, but the Intel microprocessor would eclipse them (I think I was the last group at CMU was talk a course using them. Another factoid, the predecessor to VHDL/Verilog and the like, ISPL and ISPS actually spit out 'code' for DEC RTM modules instead of gates).
But the key point is that in 1975 dollars, a PDP-11/40 system that was good enough to run something like Sixth Edition of UNIX cost somewhere between $50k-$150k. This would have been much, much cheaper than say a PDP-10, or a equivilent IBM 'mainframe' size system which would have started above $1M.
Clem