The slow-burn bet
Apple has never been first. It didn't invent the personal computer, the MP3 player, the smartphone, or the tablet. It perfected them, then sold hundreds of millions of them.
The pattern is so consistent it barely registers as strategy anymore. But AI is proving harder to sit out. The mirage and real market forces pushing every tech company toward building AI pulled Apple into a race it has never historically been set up to win.
Increasingly, it seems Apple is remembering its roots and finding its own way. At last year's Worldwide Developers Conference, buried beneath the underwhelming Siri news and a glassy new design language, Apple opened its on-device AI models to third-party developers for the first time.
The Foundation Models framework lets developers build AI features that run entirely on the device, with no cloud
connection required. The models themselves are modest — 3 billion parameters compared to the trillion-scale rumored for frontier models.
What Apple is doing is handing its AI infrastructure to the same developer community that turned the App Store into an economy. Over a billion active iPhone users, served by millions of developers.
Free, private, offline AI baked into every app that wants it. It’s a bet on its legions of supporters and partners to build what Apple hasn't figured out how to build itself.
Thinking different,
carefully
Apple has stumbled before. The "Think Different" campaign that launched in 1997 was itself a kind of resurrection story. Apple was near bankruptcy. Steve Jobs had just returned. The ads celebrated Einstein and Picasso and Muhammad Ali, people who broke rules, upended expectations, refused to fit. It was aspirational branding for a company that needed people to believe in it
again.
Fifty years in, Apple is a different kind of institution. It is the establishment. It sets the prices, controls the platform, takes the cut. A federal judge found the company in contempt last year for willfully violating court orders in the Epic Games case, a ruling that could reshape how the App Store operates. The company that once celebrated misfits has had to be dragged, legally, toward a more open
ecosystem.
What's interesting is that openness may be exactly what Apple needs right now. The Foundation Models framework is a concession that Apple can't win the AI race alone. Tim Cook, at Apple's anniversary celebrations this month, quoted the original Think Different copy: "The people crazy enough to think they can change the world are the ones who do." He was talking about Apple users. He
might also have been talking about the developers Apple is now counting on.
The privacy-first, on-device approach remains genuinely differentiated. In an era when AI companies are hoovering up personal data and users are increasingly unsettled by what gets stored where, Apple's architecture is a real alternative. Whether that matters more than raw capability is the question the next few years
will answer.
Those same years may tell us whether the AI arms race was worth running at all. Every model gets beaten by the next one. Every breakthrough becomes a commodity.
The companies best positioned to win that race are pure software
companies, and Apple has never been one. Remembering that before it was too late might have been the smartest move it has made in years.
It’s not trying to out-build OpenAI or Google, but doing what it has always done: taking someone else's breakthrough and making it useful for a billion people. That's not thinking different. That's thinking like Apple.
—Jackie Snow, Contributing Editor | 
.png)
