Apple's long-delayed overhaul of Siri is reportedly on track for a September 2026 launch alongside iOS 27, according to reports, relying on a hybrid architecture that pairs on-device processing with Google's Gemini models and Nvidia's Blackwell data center chips for more demanding tasks. The rebuild caps a project that was first teased at WWDC 2024 and delayed repeatedly since.
A Hybrid Approach to AI Processing
According to reports, simple functions like setting alarms, checking basic device information, and adjusting settings will continue to run directly on Apple hardware using on-device AI. More complex requests — such as summarizing lengthy documents, answering nuanced questions, or coordinating actions across multiple apps — will instead be routed to the cloud, where Google's Gemini models handle the heavier computational lift.
The Google and Nvidia Infrastructure Behind It
Apple is reportedly tapping into Google's fleet of Nvidia Blackwell B200 data center chips to power the cloud side of the new Siri, according to reports, with user data encrypted using Nvidia's hardware-based confidential computing features. That security layer appears designed to address privacy concerns that have historically made Apple cautious about routing user requests through third-party cloud infrastructure.
According to reports, Apple is paying roughly $1 billion per year for access to a customized version of Gemini, described as a very large model with approximately 1.2 trillion parameters. The scale of that investment underscores how central the Siri rebuild has become to Apple's broader AI strategy, even though the underlying model isn't built in-house.
A Rocky Road to This Point
The Siri overhaul has faced a difficult development path, according to reports, having been delayed multiple times throughout 2025 amid internal missteps and executive changes within Apple's AI organization. Critics have pointed to what they characterized as a lack of urgency in Apple's broader AI strategy, making the September target a significant milestone if it holds given the project's history of slipped timelines.
Why Apple Turned to an Outside Model
Apple's decision to lean on Google's Gemini rather than relying solely on its own in-house models reflects the scale of investment required to compete in frontier AI development, according to reports. Rather than attempting to match the enormous compute and model-training investments made by dedicated AI labs, Apple appears to be positioning itself as an integrator — combining its hardware and interface strengths with best-in-class external models.
What It Means for Users and Competitors
If the September launch holds, according to reports, it would represent Apple's most significant Siri upgrade in years, potentially closing a gap that has opened between Siri and rival assistants built by Google, Amazon, and OpenAI. The partnership also signals a notable moment of cooperation between Apple and Google on core AI infrastructure, even as the two companies continue to compete directly in other areas including search, mobile operating systems, and now, increasingly, AI-powered assistants.
Privacy Questions Apple Will Need to Answer
Apple has long built its brand around on-device privacy, according to reports, making the decision to route complex queries through Google's cloud infrastructure a notable departure that the company will likely need to explain carefully to users and regulators alike. The reliance on Nvidia's confidential computing hardware to encrypt data in transit and during processing appears to be Apple's answer to those concerns, though independent security researchers will likely scrutinize the implementation closely once the new Siri ships.
The September target places Apple's relaunch well behind competitors that have already shipped multiple generations of advanced AI assistants, according to reports, meaning the rebuilt Siri will need to offer a genuinely differentiated experience rather than simply catching up. Analysts note that Apple's advantage may lie less in raw model capability, which it is effectively licensing from Google, and more in deep integration with its hardware and app ecosystem — an area where rivals have historically struggled to match Apple's level of polish.
This article summarizes publicly reported technology developments and will be updated as more information becomes available.
