Apple moved Siri’s brain to Google’s cloud and Nvidia’s chips.
At WWDC 2026 Apple unveiled a rebuilt Siri that no longer runs on‑device. The voice assistant now relies on a custom 1.2‑trillion‑parameter model built on Google’s Gemini architecture and is served from Google Cloud servers equipped with Nvidia Blackwell B200 GPUs. Apple presented the change as a performance upgrade, saying the shift does not alter its privacy commitments.
The move matters because it places Apple’s core AI service on infrastructure owned by two of its biggest competitors. Data still flows through Apple’s own layer, but the reliance on external inference hardware gives regulators and privacy advocates a new angle to scrutinise.
In short, Apple’s privacy story now includes a heavyweight third‑party supply chain.