Apple taps Nvidia Blackwell GPUs to power massive iOS 27 Siri overhaul at WWDC 2026: Report

Apple is planning to use chips from Nvidia for running its next-generation Siri AI upgrade that it plans to introduce at WWDC 2026 in California next week. While Apple has already signed a deal with Google for Gemini to act as the brains behind its voice assistant, a new report by The Information (via 9to5Mac) gives us a peek at how Nvidia could play a key role in fulfilling some of the computing needs.

Reportedly, Apple will need to process some of the requests made to Siri using the Gemini AI models hosted on Google Cloud infrastructure. In order to support those requests, the company is said to be planning to use Nvidia's AI chips for the computing needs.

Specifically, the report states that Apple is planning to use Nvidia's Blackwell B200 data centre chips. The Cupertino-based tech giant is also said to enable Nvidia's confidential computing feature, which will encrypt data as it is processed on the chips.

Reportedly, the Nvidia Blackwell B200 is the company's flagship data centre GPU designed for training and running large AI models. The chip is built on the Blackwell architecture, which offers faster AI training performance, lower cost, and lower energy consumption for running LLMs compared to the previous-generation Hopper technology.

“NVIDIA Blackwell includes NVIDIA Confidential Computing, which protects sensitive data and AI models from unauthorized access with strong hardware-based security,” the company explains in a blog post.

The report also notes that Apple's new move departs from its usual strategy of controlling all the critical ingredients of its products. It is also yet unclear how the company's Private Cloud Compute will fit into the new Siri AI features launch.

Notably, the company introduced Private Cloud Compute with the initial batch of Apple Intelligence at WWDC 2024 to assure iPhone users that their data would remain private even when AI queries needed to be processed in the cloud by running them on Apple's own silicon.

In a report in January, Bloomberg's Mark Gurman also noted that Apple was planning to run the Siri chatbot directly on Google servers via tensor processing units (TPUs). However, it seems that the plan has now changed to using Nvidia chips.

Apple's Siri plans for WWDC 2026:

Apple will unveil the next generation of Siri upgrades at its WWDC 2026 event next week. The company will not only finally bring to the fore the revamped Siri, with the ability to understand personal context, see what's on users' screens, and take multi-step actions across native and third-party apps on behalf of users.

However, with WWDC 2026, the company is planning to go a step further with the introduction of a new Siri app that is expected to act like a full-fledged chatbot such as ChatGPT or Claude. The app is said to come with an option to turn voice mode on or off. There will also reportedly be the ability to see past messages and upload files and images into the chatbot to make multimodal requests.

Siri will also reportedly find a place inside the Dynamic Island with a new animation that Apple will unveil at WWDC 2026. Triggering Siri will open the voice assistant in the Dynamic Island with a "Search or Ask" omni-bar present there.