Private Cloud Compute

Private Cloud Compute (PCC) is Apple's answer to a specific problem in AI products: cloud inference requires sending user data to a server, and users have no reliable way to verify what happens to that data. PCC is designed to make verifiable privacy possible at the infrastructure level. When a request is sent to PCC, it is processed, the response is returned, and the data is not retained. Apple publishes cryptographic evidence of this that outside security researchers can independently audit.

For developers, PCC is significant in two ways. First, it is the cloud layer behind Siri AI and many Apple Intelligence features, so understanding it helps builders reason about what data their iOS apps are sending and what protections apply. Second, through the Apple Foundation Models framework, qualifying small developers can now run Apple's more capable cloud models on PCC at no API cost, which lowers the barrier to building AI features that require more intelligence than the on-device model can provide.

PCC runs on Apple silicon servers in data centers. At WWDC 2026, Apple announced that its most capable cloud model (AFM Cloud Pro) runs on NVIDIA GPUs in Google Cloud, all still under the PCC privacy architecture. This is a notable infrastructure model: cloud compute from a third-party provider, privacy architecture enforced by Apple's attestation system. For builders thinking about enterprise AI deployments, PCC is a useful reference point for what verifiable privacy guarantees in AI infrastructure could look like.