Tech Illustration: The World Ledgers
TechApple Rebuilds Siri on Google's Gemini in a Landmark AI Partnership
At Tim Cook's final WWDC keynote, Apple confirmed its overhauled, more conversational Siri will lean on a custom Google model — a rare reliance on a rival's AI.
For years, Apple’s pitch was that it did the hard things itself. So it was a striking moment when, at its Worldwide Developers Conference keynote on June 8, 2026, the company confirmed that its long-promised, rebuilt Siri would run on a model from Google — its biggest rival in mobile software.
The admission, made at what was billed as Tim Cook’s final WWDC keynote as chief executive, marks one of the most consequential strategy shifts in Apple’s recent history: the company that prizes self-reliance is, for now, renting the brains of its flagship assistant.
What Apple announced
The new Siri is the centerpiece. Apple described an assistant that is markedly more conversational, able to handle follow-up questions and complex requests, with expanded “visual intelligence” for understanding what’s on your screen and around you. It will live both inside existing apps and as a standalone app.
It ships as part of iOS 27, and Apple seeded developer betas of iOS 27, iPadOS 27, macOS 27, watchOS 27, tvOS 27 and visionOS 27 immediately after the keynote.
The Gemini deal
Apple confirmed it collaborated with Google and its Gemini family of models to power the next generation of the Apple Foundation Models behind Apple Intelligence. Apple did not disclose financial terms or technical specifications on stage.
The details that have circulated come from earlier reporting. In November 2025, Bloomberg reported that Apple would pay Google roughly $1 billion a year for a custom Gemini model running at about 1.2 trillion parameters — a dramatic step up from the roughly 150 billion parameters in Apple’s then-current cloud model. Those figures trace to that report, not to anything Apple said publicly, and should be treated as well-sourced reporting rather than confirmed specification.
Why Apple needed help
The partnership is, at its core, an admission. Apple’s revamped Siri had been delayed repeatedly, and the first wave of Apple Intelligence features landed to mixed reviews. Building a frontier-scale model in-house, on Apple’s timeline, proved harder than the company hoped.
Leasing a rival’s model is a bridge. Apple has framed Gemini as an interim solution while it develops its own larger foundation models — a way to ship a competitive assistant now without conceding the long game.
Privacy, kept on message
Apple was careful to wrap the announcement in its usual privacy framing: on-device processing for many tasks, and Private Cloud Compute for heavier requests, with the company emphasizing that personal data is not used to train the underlying models. How a Google-built model fits inside that architecture is one of the more technically interesting questions the partnership raises.
The bigger stakes
The optics are complicated. Apple and Google already have a deeply scrutinized financial relationship through the multibillion-dollar deal that makes Google the default search engine in Safari — an arrangement at the center of a major U.S. antitrust case. Adding an AI dependency on top of it deepens the entanglement between the two companies just as regulators are paying closer attention.
There is also the leadership backdrop. The keynote was widely reported as Cook’s last WWDC as CEO, with hardware chief John Ternus expected to take the helm — a succession detail worth confirming against Apple’s official statements, but one that lent the day a valedictory air.
For now, the takeaway is simpler: the most valuable consumer-technology company in the world decided that the fastest path to a credible AI assistant ran through its fiercest competitor. Whether that is a clever bridge or a lasting dependency is the question Apple will spend the next few years answering.
Sources
#Apple #Google #AI #Siri #WWDC
