In a significant shift in its approach to artificial intelligence infrastructure, Apple has signed off on the use of Nvidia confidential computing chips hosted inside Google Cloud to handle demanding AI workloads for its overhauled Siri assistant. The move, reported by The Information, reveals how Apple is threading the needle between its strict privacy commitments and the raw computing power needed for next-generation AI capabilities.
Under the new architecture, simpler Siri queries will continue to be processed on-device using Apple’s own silicon, while more complex requests requiring substantially more compute will be routed to Google Cloud infrastructure equipped with Nvidia’s hardware. Apple is retaining the “Private Cloud Compute” branding it introduced last year, signaling that privacy assurances remain a central pillar of its AI strategy.
Interestingly, Apple is simultaneously working on distilling Google’s Gemini model into a smaller, more efficient version capable of running natively on iPhones and other Apple devices. This hybrid approach — on-device processing supplemented by encrypted cloud compute — reflects the industry-wide challenge of delivering capable AI features without compromising user data.
The timing coincides with Apple’s Worldwide Developers Conference, where the company is expected to unveil a substantially rearchitected Siri alongside new on-device AI capabilities. With competitors like Google and Microsoft pushing aggressively on AI features, Apple faces mounting pressure to demonstrate that its privacy-first approach can still deliver competitive AI experiences.
The partnership signals a maturing pragmatism in Apple’s infrastructure strategy — willingness to use best-in-class third-party chips and cloud capacity, provided encryption and privacy guarantees can be maintained throughout.
