Google on Wednesday introduced Private AI Compute, a new secure cloud infrastructure designed to power advanced artificial intelligence (AI) features while maintaining user privacy.
The new platform “unlock(s) the full speed and power of Gemini cloud models for AI experiences, while ensuring your personal data stays private to you and is not accessible to anyone else, not even Google,” Jay Yagnik, the company’s vice president of AI innovation and research, said in a blog post.
The announcement comes as Google pursues an aggressive strategy to integrate generative AI across its product ecosystem — a strategy mirroring Apple Inc.’s Private Cloud Compute system.
Private AI Compute operates on what Google calls “one seamless Google stack,” utilizing the company’s proprietary Tensor Processing Units (TPUs) equipped with integrated secure elements called Titanium Intelligence Enclaves (TIE). The system lets devices establish direct, encrypted connections to a protected cloud environment, theoretically isolating user data from unauthorized access, including from Google.
The platform’s security architecture relies on an AMD Inc.-based Trusted Execution Environment (TEE) that encrypts and isolates memory from the host system. According to Google, independent security analysis conducted by NCC Group has verified that Private AI Compute meets the company’s stringent privacy standards.
Google claims the new service delivers security comparable to local, on-device processing while harnessing the superior computational power of cloud infrastructure. This combination enables access to Google’s most sophisticated Gemini AI models — capabilities that would be impossible to achieve on consumer devices like smartphones and laptops alone.
“Private AI Compute allows you to get faster, more helpful responses, making it easier to find what you need, get smart suggestions and take action,” Google said, emphasizing the platform ensures “your personal data stays private to you and is not accessible to anyone else, not even Google.”
The system processes sensitive information within what Google describes as a “trusted boundary,” adding an additional security layer beyond the company’s existing AI safeguards.
Remote attestation and encryption technologies create what the company calls a “hardware-secured sealed cloud environment” where Gemini models can process user data without exposing it to external parties.
Google has already implemented Private AI Compute in several features on its latest Pixel 10 smartphones. The Magic Cue feature now provides more timely suggestions, while the Recorder app can summarize transcriptions across an expanded range of languages, capabilities enabled by the platform’s secure access to more powerful cloud-based AI models.
