Apple’s AI Plans for iOS 27 Reportedly Depend on Google Cloud and Nvidia Chips

Apple is preparing to unveil iOS 27, its upgraded Siri experience, and several new “AI advancements” at WWDC next week, and a fresh report now reveals how the company plans to power many of those tools behind the scenes while still keeping privacy as a major focus.

The Information reports that Apple continues prioritizing on-device AI processing for future Apple Intelligence features. However, the company reportedly still faces major hardware limitations when trying to run larger AI systems entirely on its own infrastructure. Because of this, Apple has turned to Google and Nvidia for additional cloud computing support tied to Siri and other AI requests.

Apple reportedly uses Gemini to train smaller AI models

According to The Information, Apple uses a version of Google’s Gemini AI model to help train smaller models that can run locally on iPhones, iPads, and Macs. This process, known as distillation, allows Apple to offer AI tools directly on devices without depending fully on cloud servers for every request.

The report also says Apple has explored acquiring smaller AI companies that specialize in local AI processing. One company reportedly under consideration is Liquid AI, a startup focused on building efficient AI systems designed to run directly on devices.

Google Cloud and Nvidia reportedly handle larger AI tasks

Some Siri requests in iOS 27 will still rely on Google Cloud because larger Gemini models require huge amounts of computing power that Apple’s current Private Cloud Compute system struggles to handle efficiently.

To improve security, Apple reportedly approved Nvidia’s confidential compute technology in recent weeks. This system encrypts data while AI models process requests inside Nvidia GPUs, helping Apple maintain stronger privacy protections even when tasks run in the cloud.

The report adds that Apple will continue using its “Private Cloud Compute” branding for upcoming Apple Intelligence features, even though some processing reportedly takes place on external cloud infrastructure instead of Apple-owned servers alone.