Apple WWDC 2026: Siri AI Launch with Gemini-Derived Model and Google Cloud Backend

Apple announced 'Siri AI' at WWDC 2026 on June 8, 2026, powered by a custom Gemini-derived model integrated into Apple's Foundation Models for on-device tasks [1][12]. The top tier, Apple Foundation Model Cloud Pro (AFM Cloud Pro), runs on Google Cloud infrastructure with NVIDIA GPUs for demanding agentic and reasoning work [4][2]. Apple simultaneously announced that iOS 27 will let users set Claude, ChatGPT, Gemini, or Grok as their default AI assistant [7]. Apple's stock fell after the keynote despite the company entering the event near all-time highs [8].

After a widely criticized 2025 Apple Intelligence rollout [?], Apple has made a structural bet: rather than building competitive foundation model capability in-house, it is licensing Google's model and running cloud inference on Google's infrastructure while claiming to retain software-layer privacy controls. This choice — combined with opening iOS 27 to competing AI defaults — represents a significant departure from Apple's typical model of end-to-end ownership of its platform stack.

Apple used WWDC 2026 to relaunch its AI strategy under the 'Siri AI' name, describing the new assistant as capable of multi-turn conversations, on-screen awareness, and app-based task completion in a single session [1]. The keynote was pre-filmed, and demos featured multi-second pauses between prompts and responses — details Ars Technica noted without editorializing [1]. Apple executives framed the company's approach as 'human-centered,' contrasting it with competitors they said were 'racing forward pursuing AI for the sake of AI' [1].

The technical backbone is a two-layer architecture. On-device and lighter cloud tasks run on an updated Apple Foundation Model that Apple says incorporates a Gemini-derived model distilled for local processing [2][3]. Heavier tasks — described as 'demanding agentic tool-use and complex reasoning' — run on Apple Foundation Model Cloud Pro (AFM Cloud Pro), which Apple confirmed at WWDC runs on Google Cloud with NVIDIA GPU infrastructure [4][2]. Apple stated it retains complete control over the Private Cloud Compute software stack regardless of where infrastructure is hosted [5], and committed to publishing all PCC binaries for public inspection [2]. A separate Apple-built World Knowledge Service also runs within the PCC environment [6].

Two additional announcements expanded the ecosystem picture. Apple announced that iOS 27 will allow users to set Claude, ChatGPT, Gemini, or Grok as their default AI across Apple Intelligence [7] — a notable platform opening. Apple also announced a Core AI developer library integrating with Meta's PyTorch ecosystem, enabling developers to run their own models on Apple hardware [2]. Despite the scope of announcements, Apple's stock declined after the keynote, an outcome one financial commentator described as paradoxical given the company entered the event near all-time highs [8].

Reception is split along predictable lines. Technically oriented observers like Simon Willison acknowledge the features look more feasible than Apple's 2025 effort but withhold judgment until delivery, citing how badly Apple's 2024 WWDC AI announcements disappointed [2]. Enthusiastic commentators called it the biggest Siri overhaul ever and emphasized the Google-NVIDIA-Apple partnership's market implications [9][10]. Privacy-focused observers are watching whether PCC protections genuinely extend to the Google Cloud environment, where Apple says full protections are still ramping [11].