Alphabet’s (GOOG) (GOOGL) unveiled Private AI Compute in the cloud backed by its Gemini AI models.
Google said it is a new AI processing platform that combines its most capable Gemini models from the cloud with the same security and privacy assurances users expect from on-device processing.
Google noted that Private AI Compute runs on one seamless Google stack powered by its own custom Tensor Processing Units, or TPUs. Privacy and security is integrated into this architecture with Titanium Intelligence Enclaves, or TIE. This design enables Google AI features to use its most capable Gemini models in the cloud, according to the company.
Remote attestation and encryption are used to connect a device to the hardware-secured sealed cloud environment, allowing Gemini models to securely process user data within a specialized, protected space. Google added that this ensures sensitive data processed by Private AI Compute remains accessible only to the user and no one else, not even Google.
“Private AI Compute enables on-device features to perform with extended capabilities while retaining their privacy assurance. Using this technology, Magic Cue is getting even more helpful with more timely suggestions on the latest Pixel 10 phones. And with the help of Private AI Compute, the Recorder app on Pixel is able to summarize transcriptions across a wider range of languages,” said Jay Yagnik, vice president, AI Innovation and Research, at Google in a blog post.
Apple (AAPL) already has a Private Cloud Compute designed specifically for private AI processing.