NVIDIA GPUs with Confidential Computing are now used for confidential inference in Apple’s Private Cloud Compute (PCC), which expands beyond the company’s own data centers and into Google Cloud.
The information was revealed during Apple’s annual WWDC event for developers around the world. NVIDIA GPUs will support server-side inference for Apple Foundation Models, custom-built by Apple in collaboration with Google and leveraging technologies from the Gemini model family.
NVIDIA works with Apple and Google to enable some next-generation Apple Intelligence features. NVIDIA Blackwell GPUs powered by Confidential Computing have been integrated into the Private Cloud Compute hardware security architecture, which runs on Google Cloud.
Confidential Computing is essential in the era of AI experiences
NVIDIA Confidential Computing provides a hardware-based security layer for accelerated AI workloads. The technology protects data as it is processed, isolating tasks in trusted execution environments and allowing systems to cryptographically verify that the infrastructure has not been altered before sending any sensitive data to the server.
For end users, NVIDIA Confidential Computing means no one, not even the system builders, can access your data, chats, or conversations.
The adoption of NVIDIA Confidential Computing at this scale reflects a larger shift in AI infrastructure: as AI experiences combine on-device processing with cloud processing, the need for high-performance server inference arises while maintaining strong privacy and security guarantees.
How Confidential Computing ensures privacy and trust
NVIDIA Confidential Computing reflects NVIDIA’s commitment to trustworthy AI and includes these core capabilities:
- Hardware-based trust, which helps establish that systems are running on genuine, untampered NVIDIA GPUs.
- Encrypted communication paths, which help protect data as it moves between components.
- Remote attestation, which allows software to verify the security status of the platform before releasing sensitive data.
- Support for accelerated AI inference and training, which helps organizations run privacy-sensitive workloads without sacrificing GPU performance.
These capabilities become increasingly relevant for AI services that need to process sensitive information while maintaining strict user privacy controls.