Apple's new ChatGPT-like Siri may run on Nvidia chips

Apple is set to overhaul Siri using a combination of Nvidia’s AI chips and Google’s Gemini models, according to a report by *The Information*. The company plans to deploy Nvidia’s Blackwell B200 processors—its most advanced data center chips—to handle AI workloads for Siri, which will debut at WWDC 2026. This marks a departure from Apple’s usual practice of developing proprietary hardware and software, instead relying on third-party infrastructure to accelerate its AI ambitions. The upgrade will integrate Google’s Gemini AI models, hosted on its cloud, to process user requests while Nvidia’s hardware ensures efficient computing power. Apple is also adopting Nvidia’s Confidential Computing technology to encrypt sensitive data during processing, addressing privacy concerns central to its AI strategy. This aligns with Apple’s 2024 introduction of Private Cloud Compute, which processes AI tasks on Apple-controlled silicon while keeping data secure. The revamped Siri will feature long-awaited capabilities, including understanding personal context, analyzing on-screen content, and executing multi-step actions across Apple and third-party apps. A standalone chatbot experience is also in development, competing directly with services like ChatGPT and Claude. Users will interact via voice or text, with options to view conversation history, upload files, and submit multimodal requests combining text and visual inputs. Additionally, Siri’s interface will be updated with a Dynamic Island animation, offering a more interactive experience. The changes reflect Apple’s urgency to catch up in the generative AI race while balancing innovation with its core focus on user privacy and data protection.