Apple's Deliberate Approach to Privacy with Private Cloud Compute Servers

Apple's software SVP Craig Federighi has revealed more details about the company's Private Cloud Compute (PCC) servers used for Apple Intelligence features, stating that they are "really basic" - and deliberately so.

The PCC servers, which handle AI processing for features like Siri when on-device processing isn't sufficient or ChatGPT can't help, are designed from the ground up to maximize user privacy. They have no ability to store data long-term, lacking even SSDs for permanent storage; all processed data is wiped upon reboot.

"What was really unique about the problem of doing large language model inference in the cloud was that the data had to at some level be readable by the server so it could perform the inference. And yet, we needed to make sure that that processing was hermetically sealed inside of a privacy bubble with your phone," says Federighi. "So we had to do something new there." The technique used was to ensure all data processed on PCC servers is immediately deleted after being sent back to the user's device, and can never be recovered.

Apple has also created an unprecedented transparency mechanism for PCC servers. Every production server build is recorded in a cryptographic attestation log, with each entry including a URL to download that individual server image. The company says this level of transparency creates a complete trust model around the servers; iPhones will refuse to send Apple Intelligence data or queries to any server whose signature has not been logged.