It wasn't so much MS that dropped support, but rather Compaq that said it was stopping marketing Alpha boxes for NT. I think it also withdrew the funding Digital had been paying to MS to create the port. I believe there were also Digital staff in MSwho were "let go".
I personally believe that NT on Alpha didn't make sense. There were no mainstream applications on NT that needed the extra CPU horse power. What we needed at the time was faster i/o and the NT Alphas didn't deliver this. Instead it devalued the Alpha brand and made it much harder to sell the much more expensive Alpha's for running VMS or one of the Unix variants.
I believe this rendition of Compaq's actions being the reason for the end of MS support is correct. I recall being shocked the first time it came from a Compaq representative at an industry show. When I probed the thinking behind the decision, they really couldn't make sense of their actions. My interpretation was that the decision was taken at a high level. Subsequently I would learn that Compaq sold off Alpha's architectural elements to Intel, who would later use them as their own designs in the Pentium series.
As for NT/Alpha not making sense - I must disagree. At the time, there was no high end PC platform anywhere near the Alpha's performance. In an industry not sure it could really make a transition to this IBM PC "clone" platform as a "real, dependable" alternative, the idea that if needed - the horsepower could be bought was a critical decision point. Manufacturers knew there was a "big wrench" they could bring to bear, should their projections of CPU need be badly wrong for a given product.
The PC / Alpha series was responsible for legitimizing the raw platform before available software support warranted it. The fact that it was a DEC product gave many VARs a warm feeling about attempting a PC product direction.
As for the commercial success of the combination - we should not lose track of who the intended customers were. High end Commercial and Government users who frequently wrote their own applications, not general consumers who weren't sure who "Computer" was, or what he would be doing in their home.
NT was the most advanced GUI based OS, MS could offer. The fact that the primary author of the NT kernel was himself a DEC alum would have added considerable comfort to both sides. The success of Apple in marketing to the Scientific / Government community took DEC by surprise. Mandating a GUI based OS for Alpha would have been a big bonus in this effort.
Even though it not commonly known, many Alphas were sold. The failure of DEC itself, is what ultimately doomed the Alpha, which by any other measure was a highly successful processor.
