Apple WWDC 2026 Keynote: Gemini-Powered Siri, iOS 27, and Tim Cook's Final Act

On Monday morning at 10 a.m. PT, Tim Cook walked onto the stage at Apple Park’s Steve Jobs Theater for the last time as CEO. The theme was “All Systems Glow” — a not-so-subtle signal that this year’s Worldwide Developers Conference was going to be about artificial intelligence, finally and fully. It delivered.

After two years of stalled promises, botched demos, and a generative AI rollout that left users underwhelmed, Apple used WWDC 2026 to reset its entire AI strategy. The results are sweeping: a Siri rebuilt from scratch on a licensed Google Gemini model, an iOS 27 redesign that bakes intelligence into the OS at a system level, and a new Extensions framework that turns Apple devices into a platform for multiple competing AI providers. The company that once insisted it could build everything in-house has made a $1 billion-per-year bet on Google — and it showed every move on stage.

The New Siri: Powered by Gemini, Designed by Apple

The centerpiece of WWDC 2026 is a Siri that bears almost no resemblance to the assistant Apple shipped a decade ago. The rebuilt system runs on a custom 1.2-trillion-parameter variant of Google’s Gemini model, licensed at a reported cost of approximately $1 billion annually — a commercial arrangement that represents Apple’s largest AI infrastructure deal to date.

The new Siri operates as a standalone app for the first time, shipping across iOS 27, iPadOS 27, and macOS 27. Users interact through a persistent iMessage-style chat interface with full conversation history synced via iCloud. Siri now lives within the Dynamic Island on iPhone, providing a continuous visual presence rather than the ephemeral overlay of previous versions.

More substantively, Siri gains deep personal context — access to emails, photos, files, calendar entries, and contacts — enabling it to answer questions that previously required users to open multiple apps. It can see what’s on-screen, act across apps without manual switching, and complete compound tasks (“find that email from my accountant last month and summarize it into a to-do list”) that no prior Apple assistant could handle.

Apple calls the cloud component of this system “Siri Intelligence” and is explicit that some queries route to Google’s servers via Private Cloud Compute, its privacy-preserving inference infrastructure. On-device capabilities, powered by Apple Silicon and its Neural Engine, remain for sensitive requests.

iOS 27: Liquid Glass at Scale

iOS 27 extends the Liquid Glass design language Apple introduced last year across the entire system, with refinements that address earlier criticism of the aesthetic being visually noisy. The status bar, notifications, and control panels all receive the treatment, giving iPhones a cohesive translucency that adapts dynamically to wallpapers and ambient light.

AI touches permeate iOS 27’s features. Image Playground gains the ability to generate photorealistic outputs — a significant upgrade from its previous cartoon-only constraint. Writing Tools, Apple’s AI text-editing system, adds a “Coach” mode that provides structured feedback on tone, clarity, and structure rather than simply rewriting prose. The Photos app gains a generative search mode that can surface images based on emotional descriptions (“pictures from that beach trip where everyone was laughing”).

Visual Intelligence, which uses the iPhone camera as a real-time query interface, now integrates with the new Siri to enable compound lookups: point at a restaurant, and Siri can simultaneously pull reviews, check your calendar, and make a reservation through a third-party app.

iOS 27 also introduces accessibility expansions including natural-language Voice Control, more granular subtitle customization for video, and enhanced VoiceOver integration with AI-generated scene descriptions.

The Extensions System: Apple Opens the AI Platform

Perhaps the most strategically significant announcement at WWDC 2026 is the AI Extensions framework. For the first time, Apple is allowing third-party AI providers — including OpenAI’s ChatGPT, Google Gemini directly, and Anthropic’s Claude — to serve as the intelligence layer powering Apple Intelligence features.

Users can designate a preferred AI provider in Settings, and that provider will handle requests routed through Writing Tools, Image Playground, Siri’s chatbot interface, and other Apple Intelligence surfaces. Apple’s own Gemini-powered Siri remains the default, but the Extensions system transforms the company from a walled-garden AI provider into an AI platform — a move that likely reflects both competitive pressure and the practical impossibility of keeping pace with the frontier labs on all dimensions.

For developers, the Extensions API opens up an entirely new class of application: bespoke AI experiences that integrate natively into Apple’s operating system surfaces rather than living inside a standalone app. The implications for enterprise software, productivity tools, and education applications are substantial.

Foldable iPhone: Software First

Apple confirmed that iOS 27 includes native support for the form factors of its upcoming foldable device — referred to internally as iPhone Fold. The software introduces a split-view mode for the first time on iPhone, supporting two apps side-by-side when the device is unfolded, and a standard single-app layout when closed. The multitasking system borrows from iPadOS’s Stage Manager but has been redesigned specifically for the foldable’s transition between display states.

The hardware announcement itself is not expected until September, when Cook’s successor John Ternus will likely preside over an iPhone event. But the iOS 27 scaffolding signals that the product is real, shipping-ready, and a central part of Apple’s hardware roadmap for the second half of 2026.

Tim Cook’s Farewell

Before the product demos began, Tim Cook delivered brief remarks that carried the weight of a transition. He acknowledged that Apple Intelligence had “not yet delivered on everything we promised” — a rare public admission of stumbling from a company famous for tightly controlling its public narrative. He framed WWDC 2026 as the genuine beginning of Apple’s AI era, the pivot point the company had been building toward since it acquired the technology and talent to attempt it.

Cook, 65, announced in April that he would step down as CEO on September 1, with hardware chief John Ternus taking the helm. His tenure oversaw Apple’s growth from a $350 billion company to one worth more than $4 trillion. The WWDC stage has been a signature venue for Cook’s presentations since Jobs died in 2011. His farewell was characteristically understated: a short acknowledgment, a wave, and a handoff to the next demo.

What Comes Next

The developer betas for iOS 27, iPadOS 27, macOS 27, watchOS 27, tvOS 27, and visionOS 27 are now available. Public betas are expected in July, with general availability targeting September alongside new iPhone hardware.

The Google Gemini licensing deal changes Apple’s competitive posture in ways that will take months to fully understand. On one hand, it gives Apple’s billion-plus device users immediate access to a genuinely capable AI assistant at a level the company could not have shipped on its own timeline. On the other, it creates a deep dependency on a competitor — Google — and raises questions about what Apple’s AI differentiation looks like five years from now when Gemini is just the default, not the differentiator.

For now, Apple has made its bet. The platform is open, the model is licensed, and the keynote is done. What WWDC 2026 most definitively proved is that Apple, after two years of hedging, has decided to compete.