Apple's Siri Overhaul Taps Google Gemini and Nvidia Blackwell

Apple is finally bringing Siri into the generative AI era. The delayed overhaul, expected to launch with iOS 27 in September 2026, will rely on a hybrid approach: simple requests processed on-device, while complex queries are routed to Google's Gemini models running on Nvidia Blackwell B200 data center GPUs. This marks a rare departure from Apple's vertical integration, as it partners with Google Cloud and Nvidia for the cloud layer. The Blackwell chips offer higher memory bandwidth and faster inference than previous Hopper GPUs, and Nvidia's confidential computing technology encrypts data during processing—a key selling point for Apple's privacy narrative. Apple reportedly explored running similar models on its own Private Cloud Compute infrastructure but found performance limitations, leading to the external partnership.

At WWDC 2026 (June 8), Apple is expected to preview its broader AI roadmap, including how Siri will handle multi-step reasoning, app integration, and context-aware tasks. The new system represents Apple's bid to catch up with conversational AI leaders like ChatGPT and Google Assistant, which have long outpaced Siri. The overhaul has faced multiple delays since its initial tease in 2024, but now aims for a full rollout across iOS 27, iPadOS 27, and macOS 27. Separately, macOS 27 will drop support for Intel Macs, signaling the final phase of Apple's silicon transition. For professionals, this could finally make Siri a viable productivity tool for scheduling, research, and cross-app workflows—if the execution matches the ambition.