I wonder if they boxed themselves into a corner by putting so many features into ASICs. Custom chips take forever to design and iterate even today, I can only imagine how painful that was in the 80s. Couple that with a frugal management and I can see how they would be reluctant to sink resources into a fantastically expensive development process.

The custom chipsets in the Amiga went from being its secret weapon to its biggest liability. Amiga simply could not afford to keep them competitive with commodity PC hardware, especially not while Irving Gould was busy looting Commodore of every penny it was once worth.

But was it Amiga's chipset the noncompetitive part? My recollection from that era was that graphics/sound capabilities of the era were far ahead of the PC, but PC had better CPUs, which was a problem of Intel vs Motorola.

Anyway, in the early 90s my father wisely chose PC over Amiga to replace our aging ZX Spectrum. The reason - PC was open and easier to upgrade. Maybe that's what actually killed Amiga and Atari ST.

Upvoted. These are indeed the real reasons.

Commodore didn't give the hardware development team the resources and go-aheads that they needed in order to keep the Amiga competitive, mainly as a gaming machine and incidentally as an extremely popular demo platform. And it wouldn't have had to be "insane" upgrades to at least buy Commodore more time in the race to figure out the platform's future shape:

* A few more (and ideally also a bit wider) sprites.

* 8 channels of sound, with actual panning. The low sample rates were "OKish".

* Chunky mode, or at least hardware chunky-to-planar like the CD32 has.

* Bumping e.g. the 14 MHz EC020 in the A1200 to the 25 or 33 MHz variant.

* More chipmem and an equal amount of real on-board fastmem.

Other companies have made the same mistake as Amiga, before and after.

They spent a ton of R&D getting ahead on things that would be commoditized in a few years. Later, their sound card had to be better than the best of a half-dozen PC variants, and so did the video card, CPU, monitor, keyboard, disk drive, etc, etc.

Sure, they could have bought commodity versions of some of those (and probably did), but then they’d have to be the best at choosing commodity components, and there were 100’s of strong competitors doing just that for PCs.

Now that moore’s law is slowing down, and everything has long-moved into system on chips, vertical integration seems to make sense again (look at Apple).

Android was trying to create an ecosystem of interchangeable hardware vendors (like Microsoft did with DOS, and, later, Windows), but it basically failed because driver compatibility is so poor in that space, and key components, like cell modems are locked behind anticompetitive OEM licensing agreements.

Android succeeded at other things, but it’s definitely not the case that you can assemble components into something with better specs than a flagship android for half the price, even if you had access to a pick + place / IC soldering machine. It was/is that way for desktop PCs.

That Commodore didn't understand that software sells hardware was obvious back in the early 80's before they bought Commodore. The VIC-20 was incompatible with the PET, The CBM-II line was incompatible with the PET, the C64 was incompatible with the VIC-20, the Plus/4 was incompatible with the C64, the C128 was incompatible with the C64, although it had a compatibility mode. All the C64 compatibility mode did was ensure the C128 mode was rarely used.

The only reasons Commodore was successful is that their hardware engineering was great and their products were inexpensive. Apple had weak hardware and high prices, Atari had great hardware and high prices.

Jack Tramiel never used a computer until he was given an iPad. He never understood computers. He would have been just as successful selling bicycles.

All by Irving Gould.