Siri AI at WWDC 2026
Simon Willison · Simon Willison · 2026-06-08
Simon Willison analyzes Apple's WWDC 2026 Siri AI announcements with skepticism, highlighting that the Gemini-derived model now runs on Google Cloud infrastructure with NVIDIA GPUs under Apple's Private Cloud Compute privacy architecture.
Appears in
Extraction
Topics: apple-intelligencesiri-aiwwdc-2026private-cloud-computevision-llms
Claims
- Apple is licensing a custom Gemini-derived model for Siri AI that runs on its Private Cloud Compute infrastructure.
- For demanding agentic and reasoning tasks, Apple's PCC now extends to Google Cloud using NVIDIA GPUs.
- Apple uses vision LLMs to extract on-screen information, removing the need for app developers to write custom Apple Intelligence integration code.
- Apple's new Core AI library integrates with Meta's open-source PyTorch ecosystem, enabling developers to run their own models on Apple hardware.
- All PCC binaries will be published for public inspection, maintaining Apple's security transparency commitment.
Key quotes
Given how badly burned anyone who took Apple's 2024 WWDC Apple Intelligence announcements at face value was, I'm holding to a strict 'I'll believe it when I see it' policy for everything they announced today.
For the most demanding tasks, including agentic tool-use and complex reasoning, we worked with Google and NVIDIA to extend our PCC infrastructure to Google Cloud systems using NVIDIA GPUs, while maintaining Apple's powerful security and privacy protections.
It sounds like they'll be taking advantage of vision LLMs to extract information from the user's screen, which neatly sidesteps the need for every existing application to ship custom code in order to integrate with Apple Intelligence.