Google confirms partnership with Apple: Gemini takes over Siri functions, with Apple reportedly paid up to $1 billion per year

Google Cloud CEO Thomas Kurian publicly confirmed at the Cloud Next 2026 conference (Las Vegas, 4/22-23) that Google will assist Apple in developing the next-generation Apple Foundation Models, based on Gemini technology, to support a “more personalized Siri,” expected to be launched later in 2026.
(Background: Google Maps integrates Gemini, launching three major AI features focused on enterprise agents)
(Additional context: Google launches the eighth-generation TPU: two AI chips targeting training and inference, challenging Nvidia’s pain points)

Google has confirmed this in person. Thomas Kurian stated in his keynote speech that Google is Apple’s “preferred cloud provider,” and the collaboration aims to compress the large Gemini model into a small model that can run directly on the iPhone.

This partnership will be signed in January 2026, with preliminary results from knowledge distillation achieved by March, publicly showcased at Cloud Next this week, followed by the first demonstration at WWDC 2026 on June 8, and official release with iOS 27 in September.

The core technology is called “knowledge distillation” — compressing the massive Gemini model into a streamlined version so that computation can be done directly on the device (on-device). Data is fully controlled by Apple, with Google’s access limited; privacy architecture is a non-negotiable line at the negotiation table.

Here’s a noteworthy shift in direction: over the past decade, Google paid Apple about $20 billion annually for the “default search engine” position on iPhones; now, Google provides technology (reportedly costing $1 billion annually) in exchange for a foothold within Siri. The payment direction has reversed—Google has shifted from passively buying traffic to actively exporting models.