I'd bet somewhere between 200 and 350 total users of Aros...
Which is way up from the few dozen. There are now two commercially available systems that run AROS being sold and a third (ARM) in the pipeline and none of those are asking $2,000USD if a person opted to purchase a new system.
I will also point out to you that C= sold mostly A500/A600/A1200 and very few big box Amigas as a percentage of total sales. Their cash cow the low cost systems because it would sell in high volume. One could reasonably say that if C= had never made the A500/A1200, that would have been the end of the Amiga line because they couldn't generate that type of revenue (because the lack of popularity) for future sales and cover their R&D/T&E costs. It was the low end Amigas (A500/A1200) that generated momentum to create a market for the big box Amigas.
Or if you want to look at it another way, commercial coders want to have a reasonable size market to enter into to recoup their cost of making or porting software. Without the software, who is going to buy any system, especially if it's a high price one? If you sold a few hundred units, what commercial company is going to be interested vs selling tens or hundreds of thousand units @ year? A-EON going high end with high end price tag without having a sizeable low end base to attract users and devs and hope to achieve critical mass has missed something very obvious, IMO.