Copland's Long Shadow
How the Siri LLM struggle echoes Apple's most notorious unfinished project
I recently turned 40, and while the downsides of aging are well known—sore joints, gray hair, and hangovers measured in days instead of hours—there is one undeniable upside: perspective. When you've been around long enough, you start to notice patterns. I think I may be noticing one that could be missed by Apple observers too young to have had an Apple product with a six-color logo stamped on it.
Apple, as you may have heard, is struggling to make a version of Siri that leverages Large Language Models (LLMs) - you know, the magic that changed everything and made OpenAI very famous (and very, very rich). The state of the art is advancing so quickly that last month's gold standard model is old news and two breakthroughs behind the new hotness gold standard, which has unlocked even more superhuman AI powers for its users.
But against this backdrop of breakneck innovation, Apple seems… stuck, and now a proper "AI Siri" has been further delayed to iOS 20, which is 2 years from now (in consumer tech years this rounds up to 1 eternity).
And so, watching Apple struggle with its much-hyped AI-powered revamp of Siri, I can't shake the feeling that I've seen this movie before. The repeated delays, the fragmented development, the leadership struggles—for a really important project where the rest of the industry is running ahead—this is all eerily reminiscent of Apple's ill-fated Copland project in the 1990s. Copland (for those of you whose earliest associations with Apple were not "beleaguered computer maker" but "cool iPod maker") was an attempt to build a next-generation Mac OS to replace the functional (but aging) Mac OS Classic that had kept the Macintosh party going since 1984.
Of course, history doesn't repeat itself, but it does rhyme. The problems that doomed Copland—overambition, shifting strategies, internal turmoil—are now threatening Apple's AI efforts. Siri's long-overdue reinvention has been pushed back to iOS 20, likely not arriving until 2027, with an interim, LLM-powered Siri update in iOS 18.5 serving as a half-measure in the meantime. The result is a fragmented, two-track development process that mirrors Apple's old struggle to maintain System 7 while attempting to build its successor.
The analogy, of course, is not perfect. Apple today is worth nearly $3 trillion—whereas 1996 Apple was worth around $3 billion (and falling). Today's Apple has enough money to have essentially achieved corporate immortality. They have virtually limitless resources to build whatever they want. Getting an AI Siri is not make-or-break for Apple in 2026 in the way that saving the Mac was in 1996.
But necessity is so often the mother of invention. Copland's mounting failures forced a dying Apple to make a desperate gambit: abandoning its homegrown OS efforts and acquiring NeXT, which brought in not just their new next-gen OS, but some guy named Steve who, as it turned out, had a pretty good knack for running the place. They shipped Mac OS X, and some other things too, and things worked out pretty okay for them.
But would today's Apple be capable of taking such a bold leap? Apple's sheer scale and dominance make it hard to imagine the kind of existential crisis that drove the NeXT acquisition. Yet that same security raises an uncomfortable question: does Apple's corporate immortality breed a level of complacency that even the best management and brightest engineers can't shake? If Siri's reinvention continues to falter, would Apple have the courage to scrap its internal efforts and make a similarly audacious move?
Who, then, would be Apple's NeXT?
And more importantly—who would be Apple's next Steve?