To put it short: if Commodore had had decent management, it would now, with a little luck, be where Apple are now. Perhaps five percent of the worldwide computer market.
Now, there are several factors which could influence this, though:
Commodore did a lot of bad things towards the end of their life, as well as before. When does this theoretical competent management enter the picture? If we just propose that Commodore suddenly got rid of its old management in 1994 (straight before its death), we have to take into account two really odd Commodore path decisions from that time. One thing is the PA-RISC route. PA-RISC is now on HP's own death list, so right now, we would face right now a transition period either to the Itanium (HP's future choice), or to another processor, such as the PPC we all know and love. It should also be taken into account that the new Intel/HP processor still isn't aimed at the desktop market, both with regard to its environmental specs and its price. This is no problem for HP, which have always made PA-RISC workstations with very well thought-out cooling systems, but it would be for Commodore, whose systems are a lot more mainstream. Really, the only processors actually being developed with desktop use in mind today are Wintel and PPC, so any other processor choice could have proven fatal, since C= would have wound up with either an under-powered processor (ARM), one without a future path (Alpha, PA-RISC, probably MIPS, too), or one which is too expensive for consumer products (Alpha, PA, SPARC). These days would indeed have proven interesting times for a PA-bound Commodore.
There were also Commodore rumours about going for Windows NT, which would have killed the Amiga, without any doubt. You just can't coexist peacefully with M$.
Now, if we ignore those possible futures, what do we have then? Basically the Apple thing. At least Escom were clear about going to the PPC, so we have a processor for the future and no worries there, if we say Commodore would have done the same.
Back in 1994, Commodore's market is still basically intact. They release a new computer, either 68k or PPC based. Well, it would have been a work of wonder if they would manage a PPC transition in less than a year, but at least they would have a new computer out, or proceed to sell A1200s, A4000s and A4000Ts. They would lose a lot of funds on a PPC transition, and seeing as the lower stratum of the Amiga market was using A1200s, being wary of yet another upgrade in such a short time (both in Commodore's "one machine a decade" view and in the view of still teenaged Amiga users, who may have had an A500 for half their lives), they would have a lot of problems convincing the users, and as a consequence, the software houses, to switch.
Well, this might not be such a big problem, since we all remember the 68k Amiga market staying relatively healthy even for a few years after Commodore's demise, even with new production runs of the old machines to satisfy the market. But they would have to keep the old 68k line in parallel with the NG market for at least some years, I suspect. RTG was only to come in OS 3.2 or 3.3, and CGX/P96 only came out after C= died, so there was no culture of avoiding hardware banging back in 1994. Looking at the software offerings of today, all RTG friendly and as hardware independent as possible, Hyperion are in an enviable position compared to that of a 1994 Commodore OS team. Apple didn't have to deal with this, since it actually managed to enforce programming guidelines and didn't have an across-the-line unified custom chip architecture like Commodore. You can't aspire to run any major 80's app or game on OS4, but you have a fair chance doing that on a PowerMac.
Still, Commodore was heading in that direction, so the programmers had better listen up, so we may have that problem routed, albeit not as smoothly as in Apple's case. The games do remain, though. You can't do a game without accessing the hardware on a standard A1200, so all the old titles would have to be left behind, leaving the NG C= Amiga an uncompelling alternative to all the Amiga gamers, which would make Commodore lose a lot of income in the short term, unless they managed to make the A1200 still seem like a compelling games machine, perhaps even upgrading it in the process, which would disrupt the NG transition.
However, the big problem facing Commodore to a much larger extent than Apple in the nineties is the commoditisation of the computer, particularly the PC, which coincides very well with Commodore getting in a serious mess. In the eighties, common people didn't have a computer. But their teenage sons might. A few years later, we have a situation where the teenage son is begging his father, mother or sister, to let him use the family PC, which cost several times as much as the Amiga he would otherwise have had. The computer market grows a lot in the nineties, but the Amiga wouldn't get much of a share of it, unless Commodore were very focused for once and made a computer targetting the family and not their usual market of teenage boys and video graphicians. This is difficult, since so many of the new computer users are already familiar with the PC from work, and it may be built and sold by anyone, not just Commodore. Apple's Performa line was no rampant success, as I recall. Instead, it might even shrink Commodore's market, since the existence of a PC in every family may lead to fewer teenagers buying a computer of their own, and if they did, it would probably not be an Amiga, since they are already weaned on the PC by now, and there is an ample supply of what once made the Amiga big - pirate software.
This - the lack of down-market adoption of an incompatible NG Amiga, and the PC invasion of the home - could force Commodore upwards, into a smaller but richer market, that owned by Apple. Commodore would have similarly specced machines to Apple, and would focus on their own up-market slots, namely those of DTV and multimedia. Note that both of these are now important Apple markets, so if one non-PC vendor could do it, then so could Commodore, especially considering that they already had a foot in that market. Newtek would be kept on board, just as Apple have always tried to appease Adobe and Quark, and this would lead to Commodore having a competitive solution. Scala is another big name which would still have been with us, and generated sales for a reborn, high-end Amiga workstation. These uses alone would secure an industry need for all kinds of other graphics titles, such as ImageFX (though that one in particular is still with us), ADPro, Imagine, Cinema4D, Real 3D (note how the last three are all surviving on the Mac and PC, with C4D in particular being very popular). In the end, a maintained and fairly prosperous market would ensure a steady flow of other applications, too. Not too bad, though all of us might not be able to buy a brand new NG Amiga, just as some people say that Macs are too expensive.
In the end, Commodore, what with it still having both a games market and a pro market in the early nineties, could have gone the Apple route, but Commodore culture stood in the way. Since going into computers, they had always made their main income from the home computer market, one which they would run a big risk of losing either through outside factors (home PC market exploding) or internal factors (incompatible new developments).
Looking at what Dave Haynie has written about Hombre, though, it seems that at least some parts of Commodore were willing to let the Amiga move up a slot. After all, I don't think that AAA was supposed to have been ECS/AGA compatible, and they were planning to introduce integrated networking (albeit still favouring ARCnet) on the motherboards. With AAA, LAN, DSP and OS3.3, the Hombre Amiga would have been a mid-nineties powerhouse, but perhaps pleasing not so much most of us A1200 users as those who ran Toasters or Scala. But this would have secured the future of Commodore, I think.
As for custom chips and buses, I don't think that's a problem at all. Commodore were after all moving in that direction, and simple market realities would have pushed them in that direction sooner or later. Apple ditched Nubus for PCI, after all, and though Zorro is a notch above that, the new bus would according to Haynie have been so similar to PCI that they would have chosen to use that instead, sooner or later.
I still don't think we should say that Commodore was all bad, or deliberately pushing the Amiga as a games machine.
First of all, Commodore was one of the richest and biggest companies of the computer industry back in the eighties. Most other rich companies were IBM clone makers, and after looking at GW2000's handling of the Amiga or Compaq's handling of DEC, we know that that kind of company culture can't deal with alien architectures or innovate enough by themselves. Atari would have been just as bad, since they disappeared from the market at the same time as Commodore. Can you think of any other company? None in the computer business, at least. Commodore had funds, engineers and technology. How many computer manufacturers have their own semiconductor factories (Commodore had, with MOS)?
Commodore originally didn't try to push the Amiga as a games machine, either. Quite the opposite. The Amiga 1000 must have been the world's most expensive and difficult-to-operate games machine, in that case. I even remember reading about how Commodore held an early meeting with software houses to rally development for the yet-to-be-released Amiga. When two of the attendees said they were planning to write a game, they were thrown out. The Amiga was supposed to be a serious computer. Then, the story goes, the two rejectees went on to form Psygnosis. =)
Commodore released the A500 because Atari was eating the Amiga's market with its small and cheap STs. They still spent more effort and money on establishing the Amiga as a serious machine than I think Atari did. Atari never cared about making network cards, never wrote their own LAN software or TCP/IP stacks, they couldn't even decide on a working expansion bus for their high-end machines. Why, they never even produced a machine with more than one Megabus/VME slot, or an 040-based machine. Compared to an Atari takeover, Commodore was perhaps a real blessing, considering the development of the ST line.
