Particle.news
Download on the App Store

Apple Unveils Rebuilt Siri and Hybrid Apple Intelligence Platform

The system pairs on‑device foundation models with private cloud compute on Google Cloud using Nvidia hardware to keep personal data local through confidential compute.

Overview

  • Apple unveiled the new Siri AI at WWDC on June 8, released developer betas the same day, said a public beta will arrive next month, and tied a wider consumer launch to an autumn software update and new hardware.
  • The architecture uses a system orchestrator to route requests between on‑device models and a Private Cloud Compute layer so routine tasks run locally while heavy inference runs in the cloud.
  • Apple confirmed partnerships with Google Cloud and Nvidia to run its top-tier Apple Foundation Model Cloud Pro on Google’s infrastructure and Nvidia Blackwell GPUs, a deal that several outlets say may carry a large licensing cost though Apple has not confirmed a figure.
  • Apple stresses privacy by keeping personal data processing on users’ devices and using hardware-backed confidential compute when outsourcing inference, but it will limit some features by device capability, language and region at launch.
  • Analysts offered mixed readings: some see consumer upside and potential device upgrade demand, while others call the rollout cautious and note short-term investor concern after the WWDC announcements.