NVIDIA Blackwell GPUs with Confidential Computing now power Apple's Private Cloud Compute, announced at WWDC 2026

At Apple's WWDC 2026 developer conference, NVIDIA announced that its Blackwell GPUs with Confidential Computing are now integrated into Apple's Private Cloud Compute (PCC) infrastructure. The deployment enables privacy-preserving server-side inference for Apple Intelligence features, with hardware-level guarantees designed to prevent any party — including Apple and NVIDIA itself — from accessing user data during processing.

What's new

NVIDIA confirmed the integration in a June 9, 2026 blog post: "NVIDIA GPUs with Confidential Computing are now used for confidential inference in Apple's Private Cloud Compute (PCC)."

The technical stack includes:

• NVIDIA Blackwell GPUs handling server-side inference for Apple Intelligence features
• Confidential Computing encrypting data in-use via hardware-level memory isolation
• Remote attestation and GPU verification enabling cryptographic proof of system integrity for both Apple and end users
• Infrastructure running on Google Cloud as the underlying platform for Apple's PCC

The GPUs power inference for Apple Foundation Models built in collaboration with Google: "NVIDIA is collaborating with Apple and Google to support some of the next-generation Apple Intelligence features, using NVIDIA Blackwell GPUs."

For end users, NVIDIA summarized the privacy guarantee directly: "for end users, NVIDIA Confidential Computing means that no one, not even the system's builders, can look at their data."

Context

Apple introduced Private Cloud Compute in 2024 as a mechanism for offloading compute-intensive AI tasks from iPhones and Macs to secure cloud servers without compromising its privacy commitments. Unlike conventional cloud AI inference, PCC was designed so that even Apple's own engineers cannot access the data being processed — a property users can verify through remote attestation.

The WWDC 2026 expansion scales PCC significantly to handle the compute requirements of next-generation Apple Intelligence features, which exceed what on-device Apple silicon can efficiently handle. The choice of NVIDIA Blackwell — NVIDIA's latest GPU architecture — reflects the performance demands of frontier-tier inference at consumer scale.

Why it matters

NVIDIA's Blackwell GPU entering Apple's tightly controlled AI supply chain is notable. Apple designs its own silicon and has historically run its AI features on Apple chips. PCC represents one of the few production surfaces where NVIDIA GPUs are embedded directly in Apple's infrastructure at scale.

The Confidential Computing approach also sets a precedent for consumer-facing cloud AI. If hardware-verifiable privacy at inference time proves reliable at this scale, it could pressure other cloud AI providers to offer similar attestable guarantees — particularly for AI assistants processing sensitive personal conversations.