Quote:

Development (the heaviest part of total cost for small volumes): They had the competence in-house, Carda and Knäbel were doing the development "for themselves" (to build their own company). Had they had to pay some third party to do all the development and testing/bugsearching who would invoice every minute of their work (like AeonKit does), then I believe things would have been different.

Component cost: From what I understand (given the developers sponsoring programs, etc), they had some pretty tight relationship with key people over at the supplier of the Key Component, Freescale, who delivered the CPU's, the single most expensive component in the system. I seriously doubt they paid the standard list price.

Production: Same as development, they took care of it themselves. They had the know-how, and they had access to DCE's production facilities. AFAIK, they basically went there and operated the machines themselves, at the expense of some midnight oil and perhaps hosting a barbecue evening for Thomas Dellert. And some money probably, but you know what I mean.

They also did some actual market estimations and price calculations, and had the sense to *actually scrap* product ideas that they deemed wouldn't make sense from a customer price point of view. Take the Open Server Workstation for example, which would be directly comparable to the last Quad Core G5 PowerMac:





The plan was a *complete system* for $1,500 (this was in 2006 I believe), and when that wasn't possible (there were some slightly bitter messages at the time suggesting that IBM wasn't really willing to negotiate the price of the Key Components), they simply scrapped the idea and moved on. Genesi/bPlan did this several times, with several projects. To the frustration of many of us. But this however, made MorphOS reconsider the options, hence the support for existing PPC Mac HW, resulting in an impressive list of 80+ supported systems, ranging from tiny footprint Mac Mini to big-box Power Mac's, and two lines of Laptops(!). And when it became obvious that the PPC died, it caused MorphOS to reconsider the options once again, hence the x64 migration path. This has become the way forward for MorphOS.

MorphOS: One door closes, which happened several times, but each time another one (a better one) opened.


AeonKit/Hyperion however, does not think like this. Rather the opposite; the opposite to *everything* I have written above. They select an obscure, extremely expensive (dead-end) CPU as base for a new machine, they make the design overly complex with features they don't even know themselves what's it for (making design, testing and production more expensive than necessary), they pay a third party for all the development, they pay insane prices for components, they pay third party for production. The result? Around 2006 Genesi scrapped the idea of a Quad G5 system for drifting away north of $1,500. In 2012, about three "Moore's law generations" later, AeonKit pushes a €3,000 Euro system performing in 2005 *G4* level. And they claim (with no shame) to do it at loss(!). And this road was collectively kamikaze-applauded as "the way forward" (for OS4). And pretty much still is. Because of this, OS4 is bound to reach "the end of the road" soon, if it hasn't already. I believe it effectively did, some years ago actually, but only now is the effects starting to show.

OS4: Blind people yelled "hold that PPC door open", and AeonKit put down their hopeless X1000 door stop. And the door didn't close. Hence no better one opened.

Anyway, water under the bridges...