Apple to use Google servers with Nvidia hardware for the new Siri

Early this year, Apple and Google issued a short joint statement saying that the next generation of Apple Foundation Models, upon which the new Siri and some other Apple Intelligence features will be based, will use Google’s Gemini technology.

Apple and Google have entered into a multi-year collaboration under which the next generation of Apple Foundation Models will be based on Google’s Gemini models and cloud technology. These models will help power future Apple Intelligence features, including a more personalized Siri that will reportedly be unveiled during Monday’s WWDC keynote.

Now, a report from The Information has a little more detail about the servers this will all run on. Apple’s going to try to run as much of the new Siri as possible on-device, but more advanced requests that require more compute power will run on Google’s cloud. That was part of the original announcement above, actually: “models and cloud technology” (emphasis ours). The compute resources Apple will use come in the form of Google’s servers that use Nvidia’s Blackwell B200 data center chips.

Nvidia has its own “confidential computing” feature that encrypts data as it’s being processed, which will be used with other privacy and security measures to protect user data. However, so does Apple, and it’s unclear exactly how Private Cloud Compute, announced at WWDC in 2024, fits into all this. That system was built to run on Apple’s own server hardware using Apple Silicon, but The Information claims that the new model being used for Siri ran too slowly on that system when Apple tested it.

Apple is expected to unveil the all-new Siri at WWDC on Monday, June 8.