Er, sorry to be blunt but that is plain bollocks. 
Developing a plug-in costs money, the economic viability is a function one of whose main variables is the potential number of licenses sold. 3dsmax, lets face it, is not used in high end visual effects much. The main package here is Maya.
This is the only reason there is no good RenderMan bridge for 3dsmax.
Because developing such a plug-in will take two man years.
Many hobbyist developers donât understand that when they start; they think merely exporting a RIB of every piece of geometry in your max scene is enough.
Once they understand their misconception, many stop because the tasks seems to daunting.
Commercial-scale development needs a good prospect in terms of license numbers (and people paying support afterwards) to be sustainable.
RMan-compliant renderers are used mainly in high end VFX. This is an absolute niche market (read: very few licenses sold) which is why the only companies offering good RMan plug-ins for even Maya are companies that also write such renderers themselves (DNAsoft & Pixar) or who use them in their own productions (Animal Logic) so the plugin becomes a sales multiplier for the main product (renderer) or is an existing asset of the company anyway (Animal Logic) whose making available, commercially, to third parties, can help ease the costs of developing it for the existing needs, in house, anyway.
BMRT has absolutely nothing to do with that. You donât need a renderer even to create such a plug-in. You donât even need a RIB binding, you can roll your own. Sometimes you even have to. I wrote my first RMan exporter for Real3D on Windows NT in 1996. In a FORTH dialect called RPL. There was no way you could use any exisiting RIB binding (there were 3, at the time, BMRT, RenderDotC and PRMan). So I rolled my own (as could an can anyone else).
Furthermore, the license agreement that was formerly necessary, with Pixar, to develop a RenderMan-compliant renderer, has been waived since years (by Pixar).
But what is more, it was never required to develop a plugin that spits out RIB or links with a RMan-compliant renderer directly!
These things Pixar had exempt from requiring a licenses from the beginning, since at the beginning (in 1989), the RI spec had been envisioned as an industry standard that should be encouraged to be adapted by as many people & companies as possible. That stance has certainly changed. 
If you wanted to develop a plug-in and not go through the hassle of writing your own binding, you can use Aqsis or Pixieâs bindings. Or even 3Delightâs.
I would thus be curious to learn about your âresearchâ that suggests the availability of MaxMan (or its lack, thereof) has anything to do with the disappearance of BMRT (or Entropy).
It is quite interesting that ONLY the plugins for Max dissappeared because as far as I know Animal Logic still develope and supprts Mayaman. But it is also quite interesting that the guy Pixar sued was hired by nVidia and developed Gelato, and Frantic has a great connction plugin for Max called Amerretto.
There never was a guy who developed âGelatoâ, Gelato was developoed by a team of software engineers. Larry Gritz, who you are probably referring to, was not hired by nVidia as a sole person, to do so.
Larry was the author of BMRT and chief engineer of Entropy, a renderer which was based on BMRTâs source. BMRT in term was written by Larry during his student days. Entropy was marketed by start-up called Exluna that had three principal founders (as far as I recall).
Pixar sued Exluna and their founders (among them Larry), personally, for trade secret infringement. Mind you, they worked on PRMan, at Pixar, before â almost all of them. The claim sounded kind of valid, greed aside.
We never know what the outcome of the lawsuit would have been, but the bottom line is that nVidia made an offer, at the time, to buy Exluna. At the same time Exluna reached a settlement with Pixar that would mean BMRT & Entropy would be pulled from the market and Pixar would withdraw the lawsuit.
After the aquisition through nVidia, Larry, Matt and all the others supposedly wrote a new renderer for their new employer, from scratch, named Gelato.
Which stupidly depended on an nVidia GPU to be present in the system it ran on, to do its deed. I say stupidly particularly because these are bright people, so I would bet almost any sum that this was not their idea but that of their new employer, nVidia.
I said âsupposedlyâ above because I donât believe that you can write a renderer like Gelato from scratch in such a short time frame it supposedly took them and also because it curiously suffered from some of the same bugs that Entropy had had (I was a beta tester of Entropy and had a close look at Gelato, for the reasons I touched above).
Gelato is almost just like Renderman, and I would say alot of advances beyond where Renderman has come thus far.
Like what, exactly?
Gelato is not a RenderMan compliant renderer. It is slow, was buggy, last time I tested it (I evaluated it for use in feature film pipeline in 2005, we ended up using 3Delight instead).
Gelato was total failure, commercially, because nVidia decided to make it depend on hardware. No one buys such a renderer. There are numerous threads were I commented on this on CGTalk in more detail.
You can find them by search for âGelatoâ and my user name.
Why Gelato totally failed, has many reasons. Hardware dependency was one. Another one was probably that is was still slower than most of its non-hardware accelerated competitors.
All this has absolutely nothing to do, however, with the availability of plug-ins to export to RenderMan compliant renderers, from 3dsmax.
Anyhow, I actually found a legal copy of Animal Logics plugin and it is unusable without BMRT, plus I dont have a license. Our friend here may have fallen victim to Pixar just like the developers before⌠who knows.
There never was a license for BMRT, it was free. Use Google and you find the latst version, for Widows, as a archive.
Beers,
Moritz