Particle.news

Apple Asks Google to Host Gemini‑Powered Siri as Private Cloud Comes Up Short

Apple cites low Private Cloud Compute utilization alongside hardware limits to justify turning to external capacity.

Overview

  • Apple has asked Google to evaluate setting up servers in Google data centers to run a future Gemini‑based version of Siri that meets Apple’s privacy requirements, according to The Information.
  • Private Cloud Compute is reportedly used at roughly 10% of capacity on average, with some Apple‑built servers still sitting uninstalled in warehouses.
  • Current PCC hardware, believed to rely on modified M2 Ultra chips, is described as ill‑suited for frontier models like Gemini and slower to update within Apple’s fragmented cloud stack.
  • Expecting a surge in usage when the upgraded Siri launches, Apple is discussing leasing additional capacity from Google while maintaining its on‑device plus Private Cloud Compute privacy architecture.
  • Longer‑term plans such as a bespoke Baltra AI server chip are in development for later deployment, and reported financial terms for hosting—including figures around $1 billion annually—remain unconfirmed.