At WWDC 26, NVIDIA announced its collaboration with Apple and Google to expand Apple’s Private Cloud Compute service to third-party data centers, utilizing NVIDIA Confidential Computing technology. This development marks the first time Apple’s service will extend beyond its own data centers, employing NVIDIA Blackwell GPUs to enhance Apple Intelligence workloads built with Google’s Gemini models. NVIDIA’s Confidential Computing is designed to protect user data during processing, ensuring that even system builders cannot access sensitive information, thereby addressing the growing demand for AI infrastructure that balances performance with strong privacy and security guarantees.

Apple: Apple designs consumer electronics, software, and AI services with a strong emphasis on privacy. It is expanding its Private Cloud Compute service beyond its own data centers for the first time, using NVIDIA Confidential Computing on Google Cloud to enable secure server-side inference for Apple Intelligence features. The move supports custom Foundation Models built with Google Gemini technologies.
Google: Google provides cloud computing platforms and develops the Gemini family of AI models. Its Google Cloud infrastructure now hosts Apple’s expanded Private Cloud Compute environment. NVIDIA Blackwell GPUs with Confidential Computing are running on this platform to support Apple Intelligence workloads that leverage Gemini technologies.
NVIDIA: NVIDIA develops graphics processing units, AI infrastructure, and security technologies for accelerated computing. Its Confidential Computing provides hardware-rooted trust, encrypted paths, and remote attestation to protect data during AI inference. In this development, NVIDIA is collaborating with Apple and Google to deploy Blackwell GPUs with this technology inside Apple’s Private Cloud Compute on Google Cloud.
Google Gemini: Google Gemini refers to the family of large language and multimodal AI models developed by Google. Technologies underlying Gemini are incorporated into custom Foundation Models built jointly by Apple and Google. These models power server-side components of Apple Intelligence running on NVIDIA Blackwell GPUs with Confidential Computing.
NVIDIA Blackwell: NVIDIA Blackwell is the company’s current-generation GPU architecture optimized for AI training and inference. It serves as the hardware foundation for confidential inference workloads in the expanded Apple Private Cloud Compute setup. The GPUs are integrated into Google Cloud while maintaining NVIDIA’s security architecture for Apple Intelligence processing.

`json
{
“Privacy”: “NVIDIA Confidential Computing ensures that no one, including system builders, can access user data, chats, or conversations during processing.”,
“Security”: “The technology includes hardware-rooted trust, encrypted communication, and remote attestation to verify platform integrity before sensitive data is released.”,
“AI Infrastructure”: “As AI experiences combine on-device and cloud processing, demand is growing for high-performance server-side inference that preserves strong privacy and security guarantees.”
}
`