The Atari ST vs the Amiga: The 16-bit rivalry that split bedrooms

The strangest fact about the Atari ST and the Commodore Amiga isn’t how fiercely their owners argued — that’s true of every 8-bit and 16-bit rivalry this desk has covered — it’s that both machines nearly ended up being the same company’s product. Jack Tramiel, having just been forced out of Commodore in a boardroom fight, bought Atari’s home computer division in 1984 and had his engineers rush the ST to market specifically to beat his former company to the 16-bit shelf. Commodore, meanwhile, had already acquired Amiga Corporation — the small California outfit actually building the chipset that would become the Amiga — in a deal partly motivated by keeping that same chipset out of Atari’s hands, since Atari had been circling Amiga as a potential partner before Commodore’s cheque cleared first. The two machines that ended up as each other’s defining 16-bit rival both trace their existence back to the same tangle of boardroom betrayal, which makes the rivalry that followed almost personal by inheritance before either machine had shipped a single unit.

The ST’s bet: get 16-bit power out fast and cheap

Tramiel’s Atari moved with the urgency of a man settling a score. The 520ST, launched in 1985, hit the market before Commodore’s Amiga 1000 and undercut it substantially on price, built around a Motorola 68000 processor — the same chip family powering the Macintosh and, not coincidentally, the Amiga too — paired with a GEM-based graphical operating system that gave it a proper windowed desktop years before that was standard on cheaper machines. The ST’s most consequential design decision, though, had nothing to do with graphics: it shipped with a built-in MIDI port, a decision that turned the machine into the de facto standard for music sequencing software through the rest of the decade and well into the 1990s, adopted by studios and touring musicians who had no interest in its gaming library at all. That’s an odd kind of legacy for what was primarily sold as a home computer, but it’s a genuinely durable one — Steinberg’s Cubase began life as ST software, and professional musicians kept STs running in studio racks long after gamers had moved on to other machines entirely.

Atari’s speed to market came at a cost the ST never fully shook. Its graphics hardware, while perfectly capable for business software and the GEM desktop, lacked dedicated custom chips for sprite handling and hardware scrolling, leaning on the 68000’s raw processing power to do in software what the Amiga’s custom chipset did in dedicated silicon. For business applications and MIDI sequencing this barely mattered. For games, it meant ST ports routinely ran with less smooth scrolling and thinner soundtracks than their Amiga counterparts, a gap that became one of the rivalry’s most frequently rehearsed talking points and, unlike some retro-computing arguments, one that was almost entirely true.

The Amiga’s bet: build the custom silicon first

Commodore’s Amiga 1000, arriving in 1985 shortly after the ST, walked in with a genuinely radical set of custom chips built by the original Amiga Corporation engineers — Agnus, Denise and Paula, handling graphics, video and sound respectively, each offloading work the 68000 processor would otherwise have had to do itself. Hardware blitter operations let the Amiga move large blocks of graphics data around the screen without burning processor cycles on it, which is the specific piece of engineering behind the smooth parallax scrolling that became the platform’s signature party trick. Shadow of the Beast’s multi-layer parallax effect is the single most cited proof of what that chipset could do, and it worked precisely because the Amiga’s parallax scrolling wasn’t a software trick being faked — the custom hardware was doing real, independently scrolling background layers the ST’s architecture had no equivalent way to produce without grinding the 68000 to a halt trying.

That custom silicon came at a real, compounding cost. The Amiga’s chipset, revolutionary in 1985, aged awkwardly as PC graphics hardware caught up and then overtook it through the following decade, because dedicated custom chips are expensive and slow to redesign compared to a general-purpose architecture that can simply be paired with faster off-the-shelf parts. Commodore’s later chipset revisions — the Enhanced Chip Set on the A500 Plus and A600, the Advanced Graphics Architecture on the A1200 — improved things incrementally, but the company never fully solved the problem of custom hardware falling behind a PC market moving on commodity silicon and Moore’s Law rather than bespoke chip design cycles.

Music software and games software as the actual dividing line

The rivalry’s popular memory frames it as a straightforward graphics argument, and on games specifically that’s largely fair — the Amiga’s custom chipset routinely produced smoother, more colourful, better-scored conversions of the same titles the ST also received, and the Amiga canon of games that justified owning the machine is considerably longer and more celebrated today than any equivalent ST list. But the two machines actually split their user bases along a line the graphics argument mostly misses: the ST’s MIDI port and GEM desktop made it the sensible choice for anyone doing music production, desktop publishing or general business work, while the Amiga’s chipset made it the obvious choice for anyone whose primary interest was games, video work, or the demoscene culture that grew up specifically to show off what Agnus, Denise and Paula could do when pushed past their documented limits.

This split shows up clearly in the demoscene’s long shadow over game design — the overwhelming majority of demoscene groups that grew directly out of the software-cracking scene chose Amiga hardware specifically, because the custom chipset gave them more spectacular tricks to discover and show off than the ST’s more conventional architecture offered. The cracking scene’s intro screens, which became competitive art in their own right, are almost entirely an Amiga and C64 phenomenon rather than an ST one, for exactly this reason. The ST wasn’t a worse machine; it was a machine built for a different job, and the demoscene simply preferred the platform that rewarded showing off.

Team17 and the platform loyalty that outlived the rivalry

Some of the era’s most enduring studios picked a side and stayed there long enough to define what “Amiga game” meant as a genre unto itself. Team17 built its entire early identity on Amiga shooters before Worms eventually carried the studio onto other platforms, and Turrican’s sprawling, systems-dense action design became one of the clearest arguments for what the Amiga’s chipset could do when a developer pushed every custom chip simultaneously rather than settling for a straightforward conversion. The ST rarely produced studios with that kind of platform-defining loyalty, partly because its strongest commercial identity had already been claimed by music software rather than games, and partly because a machine whose games library leaned heavily on ports rarely inspires the kind of studio devotion that produces genre-defining originals.

The price war and the machines that lost the middle

Both companies spent the mid-to-late 1980s cutting prices aggressively enough that the rivalry played out as much in bundling and discounting as in raw hardware argument. Commodore’s Amiga 500, launched in 1987 as a cheaper, keyboard-integrated redesign of the original 1000, brought the machine’s price down into direct competition with the ST’s established value proposition, and it’s the 500 rather than the pricier original 1000 that most people actually mean when they talk about owning an Amiga. Atari answered with its own STFM revision, bundling more RAM and a modulator for driving a television directly, chasing the same budget-conscious buyer Commodore was now courting. Both companies discovered the same lesson Commodore had already learned fighting the Spectrum a few years earlier: whichever machine reached the widest audience at the lowest workable price built the larger software library, and the larger software library then reinforced the price advantage by making the cheaper machine look like the obviously sensible buy regardless of which chipset actually did more.

Neither machine ultimately won that price war outright in the way the Spectrum’s reach outpaced the early 64 in Britain specifically. Commodore’s Amiga line sold better overall and built the larger, more fondly remembered games library, but Atari’s ST held onto entire professional niches — studio music production chief among them — where price competition from the Amiga barely registered, because a working musician wasn’t choosing a machine on games-library depth in the first place. That’s part of why the rivalry never resolved into a clean verdict the way some of the era’s other platform arguments eventually did: the two machines were competing hardest in the middle ground of general home use, while each retained a flank of loyal, specialised users the other machine was never seriously contesting.

What outlived both machines

Commodore’s slide into bankruptcy in 1994 and Atari’s own retreat from computer hardware not long after both machines’ commercial peaks meant neither company survived to referee its own rivalry’s aftermath, which left the argument to be inherited, much like the Spectrum-versus-64 one, by emulation communities and retro hardware enthusiasts rather than settled by sales figures anyone still cared about. The Amiga’s demoscene legacy proved the more durable of the two by simple volume — new demos for original Amiga chipsets still appear from active scene groups decades on, extending tricks the custom silicon’s original engineers never anticipated. The ST’s legacy runs quieter but no less real: MIDI sequencing software descended from Cubase’s ST-era lineage still underpins large parts of modern music production, an inheritance most musicians using it today have no reason to trace back to a 1985 rush job built to beat Jack Tramiel’s former employer to market. Both machines, in their own register, are still working.

Why the rivalry still reads as unresolved

Unlike the Spectrum-versus-64 argument, which at least settled into “cheap reach against raw capability,” the ST-versus-Amiga rivalry never quite agreed on what it was even arguing about. Musicians will tell you the ST won outright and mean it sincerely, pointing at studio racks that ran Cubase on ST hardware into the 1990s. Gamers will tell you the Amiga won just as sincerely, pointing at a games library and a demoscene legacy the ST simply couldn’t match chip for chip. Both are right, because Tramiel’s speed-to-market gamble and Commodore’s custom-silicon gamble solved different problems well enough that each machine built a devoted audience around the problem it actually solved best. Two computers born out of the same boardroom betrayal ended up proving that a 16-bit rivalry, unlike its 8-bit predecessor, didn’t need a single winner to leave a lasting mark — it just needed each side to be right about something the other one wasn’t built for.