{"id":68814,"date":"2026-06-04T03:00:20","date_gmt":"2026-06-04T01:00:20","guid":{"rendered":"https:\/\/www.apfelpatient.de\/?p=68814"},"modified":"2026-06-04T03:01:19","modified_gmt":"2026-06-04T01:01:19","slug":"gemini-siri-apple-nvidia-chips-google-cloud","status":"publish","type":"post","link":"https:\/\/www.apfelpatient.de\/en\/rumors\/gemini-siri-apple-nvidia-chips-google-cloud","title":{"rendered":"Gemini Siri: Apple is expected to use Nvidia chips in Google's cloud"},"content":{"rendered":"

A new report describes how the revamped, Gemini-powered Siri is expected to work technically. According to the report, Apple would rely on Nvidia's most powerful data center chips \u2013 running in Google's cloud.<\/strong><\/p>\n\n\n\n

The next generation of Siri is drawing near, and with it, the technical details of how it will work behind the scenes are becoming more concrete. The fact that the Gemini-powered Siri is slated for release in 2026<\/a> is considered a given \u2013 the latest report from the industry publication The Information<\/a> now adds details about the hardware Apple plans to use to process some of the requests. Two names are at the center of this discussion, names one wouldn't necessarily expect to see in an Apple product: Google and Nvidia.<\/p>\n\n\n\n

Which chips should be used?<\/h2>\n\n\n\n

As early as the end of May, The Information reported<\/a> that some of the requests to the new Siri in Google's cloud would run on a licensed version of the Gemini model. Apple had reportedly approved the use of a privacy technology from Nvidia in this context \u2013 an indication that Nvidia chips would be used for at least some of the computing power in the Google cloud.<\/p>\n\n\n\n

The new report is now more precise: Apple is reportedly planning to access Google's inventory of Nvidia Blackwell B200 chips. The company intends to activate Nvidia's "Confidential Compute" feature, which encrypts data while it is being processed on the chips.<\/p>\n\n\n\n

Encryption during processing<\/h2>\n\n\n\n

\"Confidential Compute\" is a hardware-based security system that protects data while it is being actively processed by Nvidia's GPUs. According to Nvidia, the feature preserves the confidentiality and integrity of the deployed AI models and makes it possible to run sensitive AI workloads securely and with near-native performance even in shared or cloud environments. Nvidia has published<\/a> a dedicated whitepaper on this.<\/p>\n\n\n\n

For Apple, this step would be crucial: The company would have to allow customer data to be processed on external cloud servers without compromising its own data protection and security standards. Encryption during processing would be precisely the mechanism that would technically secure this balancing act.<\/p>\n\n\n\n

What the Blackwell B200 hardware can do<\/h2>\n\n\n\n

The Nvidia Blackwell B200 is among the manufacturer's most powerful data center GPUs and is designed for training and running very large AI models. It is based on the Blackwell architecture, the successor to the Hopper generation. Nvidia positions Blackwell as a platform for running and training extremely large models \u2013 up to models with a trillion parameters \u2013 and promises significant improvements over Hopper in inference, memory bandwidth, and scaling across multiple GPUs.<\/p>\n\n\n\n

A remarkable break with Apple's principle<\/h2>\n\n\n\n

The Information classifies this move as a departure from Apple's usual strategy of controlling the critical components of its products. This is precisely where the report's true significance lies. Around two years ago, Apple introduced Private Cloud Compute, its own server system based on Mac chips, thus presenting a particularly privacy-friendly approach to cloud computing. However, Gemini reportedly ran too slowly on this in-house system for the next Siri generation \u2013 which is why Apple is now switching to Google's infrastructure and Nvidia's chips.<\/p>\n\n\n\n

According to the report, it remains unclear what role Private Cloud Compute will play in the upcoming Siri launch. The reliance on external partners is part of a strategic realignment surrounding the voice assistant<\/a>, in which Apple is relying more heavily on acquisitions and partnerships than before, instead of manufacturing every component in-house.<\/p>\n\n\n\n

September will be a test of resilience<\/h2>\n\n\n\n

If the reports are accurate, the new Siri is expected to launch in September \u2013 as a significantly revamped assistant with its own app<\/a>. How well Apple manages the balancing act between external processing power and its own data privacy promises will only become clear in practical use. Apple has not officially confirmed either the partnership in this form or the hardware details \u2013 the information so far comes exclusively from reports. (Image: Shutterstock \/ otelzucianas)<\/p>\n\n\n\n