Apple History and Apple Psychology, How Apple Should Capitalize On Its AI Potential, Why It Probably Won't
Summary
- Apple Intelligence has destroyed shipping credibility that Apple accumulated over roughly two decades. Andrew Sharp calls the flagship product “essentially vaporware,” backed by a worldwide campaign for features that still do not exist; Ben Thompson argues the failure runs through Tim Cook because Apple advertised capabilities it apparently could not even demonstrate. His new default: “You would be a fool” to trust another comparable Apple promise before it reaches users.
- Scott Forstall’s removal helped stabilize Apple after Steve Jobs died but eliminated its strongest senior advocate for developers. Forstall led the software side of the iPhone, pushed Apple past Jobs’s web-app-only position, and helped create the App Store; Thompson even assigns him more credit than Jony Ive for the iPhone specifically. Yet Jobs’s death turned managed rivalry into “open war,” leaving Tim Cook to choose organizational cohesion over an abrasive executive whose employees were fiercely loyal.
- Apple’s hostility toward outside developers may originate in the near-death experience of the 1990s. Office, Photoshop, and other applications sold Macs, leaving Apple dependent on Microsoft and Adobe when its hardware and operating system fell behind; Jobs had to acknowledge in 1997 that “we need all the help we can get.” Thompson’s explicitly speculative psychoanalysis is that veteran executives became like “Depression-era babies who hoard,” determined never again to let a third party control Apple’s fate.
- Fifteen years of iPhone dominance let Apple neglect developer economics without paying an obvious price. Productivity software needed trials, subscriptions, or upgrade pricing, but Apple invested in easily ported games and a take-it-or-leave-it App Store because developers could not ignore its enormous market. The accumulated cultural lesson was worse than developer resentment: Apple learned “it’s okay to treat them like crap, and it doesn’t impact you.”
- Thompson locates Apple’s enduring software advantage in the operating system and its integration with hardware—not consumer apps, cloud services, or frontier models. His positive specimen is the iPhone X revealing notification text only after Face ID recognized its owner: hardware and software combining into “surprise and delight.” But extending integration into services creates brittleness and often shifts the benefit from users toward Apple’s revenue, lock-in, and infrastructure savings.
- On-device AI is strategically weak for an Apple-built assistant but potentially transformative infrastructure for independent developers. A smaller local model may underperform Google’s cloud-based system, while saving Apple server costs; opened broadly, however, the same hardware gives every developer effectively free inference on a predictable baseline such as “A18 or better.” That could let a small operator like Overcast’s Marco Arment experiment without funding a GPU service, then add paid Private Cloud Compute when a product merits scaling.
- The investable upside is Apple becoming an AI platform, but its psychology makes that outcome unlikely without leadership change. Thompson wants uncensored model access, device-aware APIs, sustainable subscription economics, and freedom for developers to discover products Apple cannot imagine; models themselves are becoming commoditized, so Apple need not match hyperscaler capital spending. His Microsoft analogy: Satya Nadella revived Microsoft’s developer-tools DNA, while Apple must move beyond being “the App Store iPhone company” and recover its older identity as a platform company.
Deep dive
1. Apple Intelligence turned a product miss into a credibility crisis
Sharp’s opening frame is deliberately broader than another Siri postmortem: the world’s most valuable technology company is “completely lost” trying to incorporate AI software into its business. Apple’s flagship AI product was “essentially vaporware,” while its global marketing campaign advertised features that “straight up don’t exist”—a reversal remarkable enough that “Apple suddenly looks like Microsoft.”
Thompson points readers to John Gruber’s “Something Is Rotten in the State of Cupertino,” calling it the definitive account. Gruber’s self-critique explains why observers suspended disbelief: Apple had built an unusually deep reserve of credibility around shipping what it announced, whereas 1990s Apple had been “the kings of vaporware” until Steve Jobs restored discipline.
The failure is not merely that development slipped. Thompson’s sharper question is how an undemonstrated capability entered the keynote and then a commercial: “How did it even make it into a keynote suggesting that they could?” After one charging-mat mulligan across roughly 20 years, he says comparable promises should no longer be believed “until it’s in people’s hands.”
Sharp’s hands-on verdict reinforces the gap between branding and utility: Image Playground was “an impressively useless app.” After trying the one product Apple did ship, he spent three minutes imagining a reason to reopen it and found none; Thompson agreed that it is “so bad, it’s amazing.”
2. NeXT supplied Apple’s technical DNA, while Jobs made the merger work
Thompson introduces Scott Forstall as a “mythic creature”: absent since roughly 2012 or 2013, he can absorb every wish of critics who imagine post-Jobs Apple would be perfect had he stayed. Thompson’s sole in-person impression during his 2010 Apple internship was overwhelming arrogance, yet he also encountered exceptional loyalty among people who had worked for Forstall.
His history starts with a corrective: the Apple II, not the Macintosh, paid Apple’s bills through much of the 1980s. The Mac eventually prospered in desktop publishing, but by the 1990s Classic Mac OS carried severe technical debt—one application could crash the entire machine—while Windows and Intel were “running circles around” Apple.
Near bankruptcy after losses of roughly $1 billion, Apple effectively let NeXT take over through its acquisition. NeXT brought the Unix-based, memory-safe foundation that still undergirds macOS, iOS, and tvOS; even “NS” prefixes preserve NeXTSTEP ancestry, while some underlying Unix code reaches back to the 1970s.
Jobs’s essential role was not personally designing everything but serving as “decider in chief, or the editor in chief.” He forced NeXT’s technically superior foundation and Apple’s design sensibility into one system, managed factions that disliked one another, and made a merger Thompson believes “does not happen without Steve Jobs as a force of nature.”
3. Forstall built iOS and fought to make the iPhone a platform
Apple put Forstall in charge of the iPhone’s software, making iOS “Scott Forstall’s baby”; Tony Fadell led key hardware underpinnings and Jony Ive led design. Thompson’s contested but categorical assessment is that Forstall deserves more credit than Ive for the iPhone specifically, while calling the iPod Ive’s crowning achievement.
After the iPhone launched, developers wanted native access to APIs Apple already used internally, but Jobs proposed web apps that could be saved to the home screen—a solution Thompson recalls being called a “shit sandwich” (he guesses Gruber coined the term). Users began jailbreaking phones and reverse-engineering function calls, making demand for a genuine application platform impossible to ignore.
Within Apple, Forstall was the leading advocate for native apps and developers. Senior developers believed they had an audience with him, while Thompson says the App Store’s existence, capabilities, and early developer support are “pretty universally” attributed substantially to Forstall.
Thompson does not romanticize the man: his emotional recollection is of “the most arrogant son of a bitch” he had encountered. The tension matters because Forstall could be simultaneously difficult for peers, intensely trusted by subordinates, and the executive willing to argue against Apple’s instinctive suspicion of developers.
4. Jobs’s death converted productive rivalry into an organizational threat
Jobs had cultivated “warring fiefdoms” whose leaders could despise each other yet still attend meetings and accept a final ruling. Thompson compares the arrangement to Red Bull Racing: after Red Bull’s founder died, contained factions became “open war”; without the singular authority, the system’s pointy setup could no longer hold.
Cook entered with operations credibility but little software credibility. To reassure Wall Street after Jobs’s death, Apple foregrounded Ive as the remaining creative genius—“Don’t worry, Steve’s gone, but Jony’s still here”—even though Thompson believes Ive needed an editor rather than unchecked control.
Thompson’s evidence for that concern is a decade of form outrunning function: MacBooks with unreliable keyboards and too few ports, the bending iPhone 6, and $17,000 gold Apple Watches. “Jony Ive’s brilliant, but he needs an editor. Steve Jobs was that editor.”
Apple Maps supplied the formal reason to remove Forstall, reportedly after Apple demanded that he personally take public responsibility, although Thompson notes Apple normally places such responsibility with its CEO. Firing him may have been necessary to keep a Jobs-tuned organization together, but the tradeoff was losing the senior leader who said, “No, actually developers matter.”
5. Apple’s 1990s dependence created a lasting fear of developers
The load-bearing line from Jobs’s August 1997 Macworld Boston speech was not merely that Apple could win without Microsoft losing. It was his admission of need: “If we want to move forward and see Apple healthy and prospering again, we have to let go of a few things,” accept responsibility for failure, and treat the maker of Office “with a little bit of gratitude.”
Thompson traces that dependence to applications that sold hardware: VisiCalc moved Apple IIs, while Photoshop and QuarkXPress made Macs essential for creative work. Once those developers could reach a larger market on technically superior Windows machines, Apple risked losing the very reasons customers still bought Macs.
With updated Office and Adobe software economically difficult to justify for the shrinking Mac market, Jobs effectively had to beg. Bill Gates appeared by satellite above the stage, promising Office support and an Apple investment; Thompson infers the investment may also have helped Microsoft’s antitrust posture, but the visual was nevertheless humiliating for Jobs.
During Thompson’s 2010 internship, younger employees spoke as if Apple were taking over the world; veterans spontaneously recalled Wired’s thorn-crowned “Pray” cover and the 1997 speech. His caveated supposition is that this cohort became like “Depression-era babies,” psychologically committed to ensuring Apple would never again be beholden to Microsoft, Adobe, or any developer.
6. iPhone dominance concealed a broken developer bargain
Forstall’s departure left no apparent senior advocate against Apple’s internal default: do what benefits Apple and let developers “take it or leave it.” That posture remained viable because the iPhone’s quality, first-mover position, and enormous user base made participation economically unavoidable even for developers Apple treated poorly.
Thompson rejects the old analogy that Android would defeat the iPhone as Windows defeated the Mac. DOS and its IBM-driven application base preceded the Mac, then Windows preserved that lock-in; the iPhone arrived before Android without an equivalent enterprise dependency, established its own developer base, and therefore could remain strong while imposing unfavorable terms.
His longstanding App Store criticism is that Apple promoted games, despite their being easy to port through engines and full-screen assets, while neglecting platform-specific productivity apps. Those products needed trials, subscription pricing, or paid upgrades to earn repeatedly from a narrower professional audience; Apple took years even to permit trials and other basic commercial mechanisms.
The iPad exposes the consequence: developers could imagine software uniquely suited to the device but struggled to finance companies by selling each application once and eventually exhausting customers. Today’s executives may carry less 1990s trauma, but many learned throughout their careers that developers would keep serving Apple regardless—arguably an even worse cultural inheritance.
7. Apple excels at operating systems but overextends integration
Listener Sergey asks what consumer-facing Apple software has felt magical since the first iOS. Sharp answers that the first iOS is his answer, noting it has been 18 years; Jens is harsher about cloud services—“They are all garbage”—and jokes that executives with real personal assistants might need a focus group to learn how ordinary people want voice assistants to work.
Sharp uses messaging as a clean incentives test: after adopting WhatsApp heavily, iMessage feels painful. Thompson’s explanation is structural—WhatsApp must persuade users to install and enjoy it, while iMessage “is just there”—showing why distribution strength can coexist with weak product pressure.
Thompson still calls Apple excellent at foundational software. macOS combines ease of use, extensibility, reusable frameworks, and third-party replacements that feel cohesive; his “Apple at its best” example is the iPhone X revealing notification contents only after Face ID recognizes the owner, producing delight through hardware-software integration.
Integration becomes dangerous when it reaches too far. Thompson’s analogy is a tall Lego structure, whose standardized connections remain sturdy, versus sticks and glue that achieve a bespoke result but become brittle at scale; adding services such as iCloud can therefore create complexity mainly to earn more money, deepen lock-in, or enforce Apple billing.
8. Apple should turn local inference from a constraint into an API
Google’s credible integration point is its cloud services: when users stay entirely inside Google, Android “all works.” Thompson fully expects Google within a few years to deliver the personalized assistant experience Apple only portrayed in videos—and to demonstrate it live—because infrastructure and cloud intelligence are precisely Google’s strengths.
Apple’s local-first approach has the opposite economics. A small model constrained by device memory may deliver worse results, while Apple captures much of the benefit by avoiding server expense; Private Cloud Compute adds capability and cost, but the first-party product still begins “with an arm tied behind their back.”
For an independent developer, however, local inference is “amazing” because it is effectively free. Marco Arment could add AI to Overcast without operating a GPU service, while Apple can credibly promise hardware such as “A18 or better,” vary capabilities across iPhone, iPad, and Mac, then offer Private Cloud Compute as a paid fallback, with subscription products helping cover that cost.
Thompson’s prescription is radical openness: expose the raw local models rather than a censored Image Playground component, provide reliable APIs, and make objectionable app behavior the developer’s responsibility. “Even if it’s crappy, see what people can do with it”—free experimentation is the feature Apple alone can distribute at massive scale.
9. Recovering Apple’s platform DNA may require new leadership
Thompson sees AI product discovery as an ecosystem problem because nobody yet knows what should be built. PCs enabled developers to invent desktop publishing after the platform existed; similarly, broadly available local inference could seed phone-native experiments that later graduate to much larger data-center models.
Hardware trends strengthen the opportunity. Apple sent two 512 GB Mac Studios to a developer Thompson tentatively identified as the founder of “Exo Labs or something like that.” The developer connected them and ran the full DeepSeek R1; performance was not amazing, but it worked, and models capable of operating within even 16GB should keep improving. “You skate to where the puck is going.”
Apple need not lead in models, which Thompson says are being commoditized, or replicate rivals’ capital expenditure. It could instead unleash “a tremendous well of innovation” around predictable high-end hardware—but he doubts current executives will do so because 15 years without consequences have entrenched the wrong developer psychology.
The hopeful analogy is Microsoft under Satya Nadella: Microsoft moved beyond Windows centricity by recovering its original identity as a developer-tools company. Apple similarly needs to leave behind “the App Store iPhone company” and return to being a platform company—an “old mindset” must go so an “ancient mindset” can return, perhaps through a leadership transition.