It was just like every other high end Xeon Workstation, and Apple took hits because it was compared to much less expensive machines (without similar capabilities).
When it was released, it had the same basic architecture as every other Mac, meaning that software companies did not need to care how many were sold, just what was the overall size of the Mac market. Today that would not be true. Anyone porting would have to do two ports to the Mac ecosystem with very different optimizations. Companies that have not ported up until now will not be even less likely to port to such a split ecosystem (especially given the tiny number of those machines they would sell).
Second, since any investment in this system would could only be amortized over its sales (nothing could be shared with other Apple Silicon machines), and given that Mac Pro sales have always been low, it would mean that these systems would be very expensive but would not out perform very cheap Windows or Linux systems for most benchmarks.
It would mean that Apple would have to keep a team working on performance tuning macOS for these machine and supporting this different architecture, and it would detract from Apple’s message that Apple Silicon was its future.
This machine would the worst of all possible worlds: an expensive machine with a tiny market share that can never be meaningfully faster than commodity systems for most applications.
