Apple AI History: Siri, Apple Intelligence & Gemini Deal

New leadership, an old question

Tim Cook led Apple for fifteen years before handing the chief executive role to John Ternus on September 1, 2026, the transition announced in April. Ternus had been Apple's senior vice president of hardware engineering, running the iPhone, Mac and iPad hardware teams; Cook moved to executive chairman, staying on to work with policymakers. Apple's April announcement of the handoff made no mention of AI or Apple Intelligence, which is notable given that the company's most unsettled strategic question, what to do about a frontier model it has not managed to build in house, transferred to Ternus along with everything else. [2]

Apple's business is still overwhelmingly the iPhone, which accounted for more than half of company revenue at the end of 2025, alongside Mac, iPad and a wearables line that includes Apple Watch, AirPods and Vision Pro. The company reported $416 billion in revenue for fiscal 2025 and briefly crossed a $4 trillion market capitalization in October of that year. Apple remains the world's largest smartphone vendor by revenue, which is the installed base any on-device AI strategy has to work within. [3]

A decade of Siri, then someone else's model

Apple's AI history starts well before the current wave. In October 2011 it shipped Siri as the headline feature of the iPhone 4S, after acquiring the assistant the year before. For a decade afterward, "AI assistant" meant something that set timers and misheard names, which is a fair part of why ChatGPT's arrival landed the way it did: Apple had already spent ten years lowering expectations for the category. [4]

In June 2024, Apple announced Apple Intelligence with ChatGPT built in for questions its own on-device and Private Cloud Compute models could not handle. Apple's public description of the feature still describes it as proprietary, built around on-device processing plus Apple-silicon servers for harder requests, with no public mention of Gemini or any other outside model as of this writing. The ChatGPT integration announced in 2024 was the exception, not the architecture. [5][6]

That exception became the rule in January 2026. Apple and Google announced a multi-year deal under which Apple's next foundation models would be built on Gemini and Google cloud technology, with Apple maintaining that its intelligence features would still run on-device and through Private Cloud Compute. Reporting at the time centered on a rebuilt Siri arriving on Gemini rather than an Apple-trained model, and antitrust commentary flagged the arrangement as a bigger regulatory question than the App Store disputes that had occupied Apple's legal team for years. [7][8]

The lawsuit and the labels

Six months after choosing to license Gemini rather than finish a frontier model of its own, Apple sued OpenAI in July 2026, accusing former Apple employees of carrying trade secrets with them when they left for the rival lab. Both events trace back to the same shortage: AI research talent capable of building frontier systems is scarce enough that Apple sued over people leaving for it in the same year it decided not to compete for the underlying model itself. [9]

Apple's AI policy reach extends into products with no connection to Siri. In August 2026, Apple Music made AI-content disclosure mandatory for tracks where generative AI produced a material portion of the audio, upgrading a tag that had been optional since March, at a point when more than a third of monthly uploads were already AI made even though those tracks drew well under one percent of listening. It is a small policy move next to the Gemini deal, but it is the same company deciding, in a completely different product, what disclosure AI-made content owes a listener. [10]